Re: [Emc-users] who has used thin client pcs
I'm impressed with the Pi4 for embedded computing. I have a Pi4 right here on my desk that is running a robotic motion control application. It does a full kinematic solution for a 12 degree of freedom waking "dog" robot. This involves many 4x4 matrix multiplications and dozens of forward and inverse trig functions per cycle. All this is written in Python. The Pi4 can do about 2100 cycles per second. I was very impressed. At the very most my use case needs 300 per second and I'll likely run the loop at 100 or less per second. This is my "servo control cycle". I can also run the same code in my Xeon-based workstation. One Xeon core can run the loop at about 4,000 cycles per second. Only twice as fast as the Pi4. (But the Xeon has 16 cores) In other tests the Pi does poorly. It is not good at all at moving data around between the network and disks or disks and memory and it is even worse when it comes to graphics performance. As a machine controller I am very impressed with the Pi4. I am running on battery power so I do care about power use. I suspect that the only thing I will find that outperforms the Pi4 in terms of computing power for its size and weight is a cell phone. All Android phones run Linux under the hood. But interfacing an Android phone to a milling machine is nearly impossible unless you just use it as a kind of remote controller pendant. It looks like it is getting hard to run software step generation on a new PC class computer.It's just as well because now there are very good low-cost microcontrollers. On Fri, Feb 11, 2022 at 3:40 AM gene heskett wrote: > On Thursday, February 10, 2022 3:05:57 PM EST Chris Albertson wrote: > > Yes, "Mini ITX" is just a shape and a standard for where mounting holes > > go but it is a shape that is sold to a cost-sensitive market where low > > power ususage and low cost maters. The ITX market is large enough > > that mass production drives prices down. So it is a good fit to > > hobby-level machine control. I would think a good place to look as > > Aliexpress as most of this stuff ships from China at very low cost. > > > > If you are never going to use the computer as an interactive desktop > > you could use a Raspberry Pi4.Then use any standard PC notebook to > > remote log-in and run the display from the notebook over WiFi. A Pi4 > > is 1/2 the cost of the cheapest ITX board. > > I did this just to see if I could, but the interfacing from the pi to the > lathe IS the cost of the ITX board over and above the 2 gig rpi4. When I > did it, the 7i90HD was suggested, but while it has numerous ways to > configure 72 i/o's, its fpga has no buffering and the 3 50 in i/o sockets > are a PITA to wire uo to, and leave the fpga wide open to being blown by > noise fron ground loops etc. Enter the 7i42TA, the protection magic for > the 7i90HD, but it takes 3 of them. Now I think it would survive the EMP > of a tactical nuke. And because the i/o is there, and now on green > terminal strips that are much easier to wire, I have all sorts of extra > stuff hooked up and running, like a pair of $20 100ppr dials to replace > the missing cranks, speed selectable on a 1,2,5 etc scale from .0001" to > 20 thou per click, one per powered axis. All the modular outputs were > used from gpio-0 to the end of the config installed, and I used the gpio > from gpio-71 down for such things as controlling ALL the lathe power with > a couple 40 amp AC SSR's which are switched by the state of F2. I could > go on, but if this lathe ever grows a controllable tool changer like the > EMCO-5 or similar I have buckets of i/o currently unused. > > And that rpi4 is running the full desktop raspian buster install. With a > preempt-rt kernel I built on that rpi4. Obviously I shouldn't run FF at > the same time as LCNC as FF plays hell with the latency, but I have > carved air while running FF just to hear it stutter. And it does, but > nothing else seems to bother it that much. > > I'm useing the SPI interface version of the mesa file for the 7i90HD and > the 3 wire interface has been bulletproof. No comm errors. > > At about 25 watts total for the rpi4, interfacing and 11 watt monitor, I > don't even have a power switch to shut it off. With a small $35 ups and a > 20kw generac in the back yard, uptimes are most of a year, till I need to > add something else, I have the i/o left to do it, whatever it might be. > > SPI to/from the pi is faster than any other comm method we have, its in > 32 bit packets at 41megabaud from pi to mesa card, and 25megabaud coming > back from the mesa card to the pi. > > So it wasn't as cheap as I was expecting but the plethora of i/o has had > me considering converting the rest of my machines and saving
Re: [Emc-users] who has used thin client pcs
Yes, "Mini ITX" is just a shape and a standard for where mounting holes go but it is a shape that is sold to a cost-sensitive market where low power ususage and low cost maters. The ITX market is large enough that mass production drives prices down. So it is a good fit to hobby-level machine control. I would think a good place to look as Aliexpress as most of this stuff ships from China at very low cost. If you are never going to use the computer as an interactive desktop you could use a Raspberry Pi4.Then use any standard PC notebook to remote log-in and run the display from the notebook over WiFi. A Pi4 is 1/2 the cost of the cheapest ITX board. On Thu, Feb 10, 2022 at 5:57 AM andy pugh wrote: > On Thu, 10 Feb 2022 at 03:04, andrew beck > wrote: > > > > Its just kinda hard to ship to New Zealand from UK. > > Anyone got a oceania source or asia > > Mini-itx is just a motherboard standard. > > mini-itx.com just happens to be a UK supplier. I often suggest using > their board finder, but then you can hopefully find a local supplier. > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] who has used thin client pcs
A CNC controller is an "off label" use case for a thin client machine. What you should be looking for is a "mini ITX" motherboard designed for industrial embedded processing. these sell new for just about $200 with a CPU soldered (not socketed) to the board. The Intel Celeron processor usually does not use a fan, except for the case fan. That said, these Mini-ITX board are many times what you find inside a "thin client" PC. The Celeron is a powerful CPU if you don't need graphics On Wed, Feb 9, 2022 at 6:13 PM andrew beck wrote: > thanks chris > > specifically i am thinking of running my cnc machines on them. they are > cheap fanless and seem to be pretty solid. > > but just seeing who else has gone down that way > > On Wed, Feb 9, 2022 at 5:58 PM Chris Albertson > wrote: > > > They work well for their intended use case. We used them at one place I > > worked. All the data and the user sessions were kept on a server and > then > > any employees and sit down at any computer and sign-in and get back to > the > > desktop he left. Also if one failed, there was no data loss as there > was > > not data on the PC. > > > > I would not use this for 3D CAD or video editing or gamming but for > normal > > office work and web rowing they are good. > > > > The trick is to match the PC to these case. I've typing this on a > > "workstation" class PC, 16-cores and a decent Nvidia GPU but I only have > a > > 256GB SDD drive installed. All the data lives on a network storage > > device. I use this for 3D CAD, robotic simulations and other big > compute > > jobs > > > > But of late the "thin Client" PC is become obsolete. What we use now at > > Apple iPad. For most people these can replace a notebook computer > > especially if you have a little iPad sized keyboard for when you need to > > type more than a few lines of text.The iPad connects to the same > > network storage (NAS) and if needed I can run an instance if Microsoft > > Windows 10 in the NAS with the Windows desktop exported to the iPad > screen > > > > A $200 Chromebook can work just like the iPad but is slightly less > > portable.The "Thin Client" is a desktop, non-portible version of > this. > > > > A very w-cost chin client PC can be made by attaching a Raspberry Pi4 to > > the back of any flat screen monitor with some double sided mounting tape. > > > > But, bottom like is these are CLIENT side devices. The compete and > storage > > is some place else. Thee mostly provide screen and keyboard and very > > limited local computing. > > > > > > Al that said. it is very easy to getcome spoild with "my screen and data > > follows me" style computing > > > > On Wed, Feb 9, 2022 at 4:05 PM andrew beck > > wrote: > > > > > hey everyone > > > > > > just wondering who has used the thin client computers > > > i am looking for a fanless pc that i can buy in oceania > > > > > > https://www.youtube.com/watch?v=tVXn_fEbDTs > > > > > > ___ > > > Emc-users mailing list > > > Emc-users@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > > > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] who has used thin client pcs
They work well for their intended use case. We used them at one place I worked. All the data and the user sessions were kept on a server and then any employees and sit down at any computer and sign-in and get back to the desktop he left. Also if one failed, there was no data loss as there was not data on the PC. I would not use this for 3D CAD or video editing or gamming but for normal office work and web rowing they are good. The trick is to match the PC to these case. I've typing this on a "workstation" class PC, 16-cores and a decent Nvidia GPU but I only have a 256GB SDD drive installed. All the data lives on a network storage device. I use this for 3D CAD, robotic simulations and other big compute jobs But of late the "thin Client" PC is become obsolete. What we use now at Apple iPad. For most people these can replace a notebook computer especially if you have a little iPad sized keyboard for when you need to type more than a few lines of text.The iPad connects to the same network storage (NAS) and if needed I can run an instance if Microsoft Windows 10 in the NAS with the Windows desktop exported to the iPad screen A $200 Chromebook can work just like the iPad but is slightly less portable.The "Thin Client" is a desktop, non-portible version of this. A very w-cost chin client PC can be made by attaching a Raspberry Pi4 to the back of any flat screen monitor with some double sided mounting tape. But, bottom like is these are CLIENT side devices. The compete and storage is some place else. Thee mostly provide screen and keyboard and very limited local computing. Al that said. it is very easy to getcome spoild with "my screen and data follows me" style computing On Wed, Feb 9, 2022 at 4:05 PM andrew beck wrote: > hey everyone > > just wondering who has used the thin client computers > i am looking for a fanless pc that i can buy in oceania > > https://www.youtube.com/watch?v=tVXn_fEbDTs > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] What Would You Suggest?
nto the end > to > > > get ft-lbs or ft-in until it turns. That's the torque required to > overcome > > > static friction. Double that to choose your motor. > > > > > > Say that is 1 ft-lb or 192 oz-in.If you choose 3:1 for your > reduction > > > ratio you get 600 oz-in. Look at the motor torque curve (they are all > > > different and if the supplier can't give you that buy one somewhere > else) > > > and see where the torque drops below 400 oz-in. Say that's 180 RPM. > > > That's 3 RPS which multiplied by 2000 steps per rev for micro-stepping > is > > > 6000 steps/second which achieves your 1 RPS on the handle. > > > > > > Or if you find it's 2 ft-lb or 400 oz-in choose a much larger motor > like > > > 1200 oz-in > > > http://www.automationtechnologiesinc.com/download/9259/ > > > Notice the curve at 3000 half steps per second is about 3.2NM. That's > > > 12,000 steps per second (7.5RPS) with 8 micro-steps per step well > within > > > the reach of even a parallel port controller and 450 oz-in. That's > well > > > above the 1 RPS you need and even just 3:1 still gives you 1600 oz-in. > > > > > > My two cents... > > > John Dammeyer > > > > > > > > An alternative would be to provide more gearing, but I don't think > it's > > > > practical to get more than about a six to one ratio in a single belt > > > > reduction and I'd like to avoid mechanical complexity if I can. > > > > > > > > Thoughts? > > > > > > > > Ken > > > > > > > > Kenneth Lerman > > > > 55 Main Street > > > > Newtown, CT 06470 > > > > > > > > > > > > > > > > On Fri, Feb 4, 2022 at 7:13 AM Chris Albertson < > > > albertson.ch...@gmail.com> > > > > wrote: > > > > > > > > > If looking for lowest cost solution you can us the old "Atom" > computer > > > to > > > > > control the grinder as long as you do not need to run the mill and > > > > > grider at the same time. Get an Eiternet interface Mesa card for > the > > > new > > > > > machine, You need two config files, just load the one for the > mill or > > > the > > > > > one for the grinder. > > > > > > > > > > Then someday you buy a second computer you only have to move the > > > Ethernet > > > > > cable over. The best option is a newer version of the Atom. They > > > seem to > > > > > sell for just under $200. Finally Newegg.com always has many > used oe > > > > > refurb PCs Used PCs sourced locally can be a cheap as "free" > > > > > > > > > > But 9ld PCs tend to burn up a lot of power. I am trying to get > mone > > > to do > > > > > "wake on LAN" so it can not use power until I need to log onto it > > > > > > > > > > On Thu, Feb 3, 2022 at 6:52 PM Kenneth Lerman > > > wrote: > > > > > > > > > > > I'm considering converting a surface grinder to CNC. To start, > I'll > > > > > > probably just convert the longitudinal and transverse axes. > > > > > > > > > > > > I'll go with steppers for this -- I'm thinking NEMA-42 motors. > > > > > > > > > > > > My current Bridgeport clone uses servos and Jon Elson's hardware > on a > > > > > > little Intel Atom Box. I'm thinking of using a Rpi for this. It > will > > > > > need a > > > > > > minimal display/control panel when completed, but initially will > > > need a > > > > > > display with touchscreen or mouse and possibly a keyboard. In the > > > long > > > > > run, > > > > > > some buttons. and perhaps an mpg might be useful. > > > > > > > > > > > > I'd like to use a raw Rpi without adding special hardware > directly. > > > That > > > > > > probably means using a USB or ethernet interface to control the > > > steppers. > > > > > > I'm thinking of using Mesa hardware. > > > > > > > > > > > > Can someone suggest the most cost effective way to do this? > > > (Although I > > > > > > have to admit, that after buying the timing belts and pulleys, > the > > > > > > s
Re: [Emc-users] Mini-itx 64bit mother board with parallel port
On Thu, Feb 3, 2022 at 4:25 PM Andy Howell wrote: > > Starting from scratch, I would likely stay away from parallel ports. Our > current CNC router's controller uses the parallel port, so that is what > I have to go with. Buy an Ethernet Interfaced Mesa board for $89 and it has a DB25 connector on the back that works just like a parallel port. Then you can connect that DB25 to your $10 Chinese breakout board. Basically, it is an Ethernet to Parallel port bridge, but what you end up with for the $89 is two parallel parts that have *much* better performance than a real parallel port. The Mesa card would be a significant performance upgrade while still allowing the CNC machine to keep its existing 25-pin parallel cable. Also it opens up the option to use any computer, even if it does not have a built-in parallel port. A $35 Raspberry Pi could even work (but you'd be better off with any of those cheap Intel boards that have two Ethernet ports. Or I'm sure you can find an older PC in someone's junk pile -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] What Would You Suggest?
If looking for lowest cost solution you can us the old "Atom" computer to control the grinder as long as you do not need to run the mill and grider at the same time. Get an Eiternet interface Mesa card for the new machine, You need two config files, just load the one for the mill or the one for the grinder. Then someday you buy a second computer you only have to move the Ethernet cable over. The best option is a newer version of the Atom. They seem to sell for just under $200. Finally Newegg.com always has many used oe refurb PCs Used PCs sourced locally can be a cheap as "free" But 9ld PCs tend to burn up a lot of power. I am trying to get mone to do "wake on LAN" so it can not use power until I need to log onto it On Thu, Feb 3, 2022 at 6:52 PM Kenneth Lerman wrote: > I'm considering converting a surface grinder to CNC. To start, I'll > probably just convert the longitudinal and transverse axes. > > I'll go with steppers for this -- I'm thinking NEMA-42 motors. > > My current Bridgeport clone uses servos and Jon Elson's hardware on a > little Intel Atom Box. I'm thinking of using a Rpi for this. It will need a > minimal display/control panel when completed, but initially will need a > display with touchscreen or mouse and possibly a keyboard. In the long run, > some buttons. and perhaps an mpg might be useful. > > I'd like to use a raw Rpi without adding special hardware directly. That > probably means using a USB or ethernet interface to control the steppers. > I'm thinking of using Mesa hardware. > > Can someone suggest the most cost effective way to do this? (Although I > have to admit, that after buying the timing belts and pulleys, the > steppers, power supply, stepper drivers, ..., it's too late to be really > cost effective.). And the surface grinder only cost me $300. > > Thanks, > Ken > > > > Kenneth Lerman > 55 Main Street > Newtown, CT 06470 > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Mini-itx 64bit mother board with parallel port
I just looked around. There seem to be many at this price point of just under $200 that all have 6 Watt CPUs in them and several have parallel ports. Anyways, sub-$20 Intell machines for industrial use seem to by plentiful. Much better then a Raspberry Pi for not much more However if I were setting up a new system I'd not want a parallel port. Ethernet seems to be better and many of these little mini-ITX boards have two Ethernet ports on them, as I assume they are designed to be file servers. In actually I just bought one, kind of. It is a Synology NAS with one of these low-power quad core CPUs inside. It does RAID and will hold all my files with a 10 Watt CPU. (I know it is nothing to to with LCNC, except file storage. I'm looking to consolidate storage and backup On Wed, Feb 2, 2022 at 6:44 PM gene heskett wrote: > On Wednesday, February 2, 2022 9:08:01 PM EST Chris Albertson wrote: > > I followed that link.Wow, that is a good deal. Especially when you > > look at the power supply. It uses a 12 volt barrel jack and a large > > size wall-wort. The CPU burns all of 6 Watts.It could run on > > battery power. > > > > It is good to look for low-power PCs if they are going to run all day, > > every day. The cost of power really adds up. The bix Xeon powered HP > > I use for development work costs maybe 12 cents an hour. That is > > about $400 at the end of the year. > > > > The machine I use to power LCNC and my 3D printer is an Intel i5 and I > > could justify downgrading it based on power-saving along. I had not > > realized there was such things at 6 Watt quad-cord Intel CPUs. > > > > On Wed, Feb 2, 2022 at 3:35 PM Andy Howell wrote: > > > On 2/2/22 16:31, gene heskett wrote: > > > > On Wednesday, February 2, 2022 4:42:22 PM EST Andy Howell wrote: > > > >> I was hoping to update to a recent Debian and LinuxCNC version. > > > >> However, I have a 32bit motherboard. This is for our school, so > > > >> I'm > > > >> trying to contain cost by just replacing the motherboard. > > > >> > > > >> Any suggestions for a 64bit mini-itx motherboard? > > > > > > > > Most any NEW mobo today will need a new cpu too since the sockets > > > > are > > > > changed and probably won't take your old memory for the same > > > > reason. > > > > > > > > ATM I'm running a normal sized Asus Z370-AII with the cheapest 6 > > > > core i5 on it and I'm as happy as I can be. Asus makes decent > > > > stuff. Draws about 140 watts less than the phenom it replaced. > > > > > > > > And stay away from OLOy memory, I had a failure and they needed > > > > more data than I had to replace it, so I had to buy a different > > > > brand to replace it. That's BS, so be sure, get it in writing, > > > > that you can get in warranty replacements by simply shipping the > > > > bad one back with a photocopy of the bill of sale. If they won't > > > > do that, go down the list to the next vendor. > > > > > > Gene, > > > > > > It looks like most of the Mini ITX boards have the cpu soldered in. I > > > don't imagine I'd need a lot of memory. 2GB maybe? > > > > > > Looks like this will do the trick. $200 with 2GB of memory. > > > > > > https://www.onlogic.com/pd14ri/ > > > > There is however, one detail that would discourage me, its already EOL, > came out in q4-15, lifespan 4 years, so its approaching 2 years since > last shipped. > > Where is the support, I never got that page to load. > > > Thanks, > > > > > > Andy > > > > > > > > > > > > > > > > > > > > > ___ > > > Emc-users mailing list > > > Emc-users@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > Cheers, Gene Heskett. > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author, 1940) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Mini-itx 64bit mother board with parallel port
About disk drives... We've all seen how a Linux Demo CD can boot and run off the CD using RAM as a "fake" disk drive. Tis is how all the Linux installs are done or if you want to just try Linux and not write anything to you hard drive. It is also possible to boot from a network server, like you bot off a CDROM and then run on the RAM disk.Then once running the PC can mount a disk over the network and have access to any number of terabytes from a server. read/write speed is limit by the network cable to about 100 megabytes per second buy that is very usable. I think LCNC could be run this way, with no disk at all on the PC. I used to work at a place that had many of the corporate desktop PCs configured to run like that, with no internal storage at all. You think it might be slow but RAM-disks are even faster than SSDs by at least 10X. They are small but they only hold scratch files On Wed, Feb 2, 2022 at 8:25 PM Chris Albertson wrote: > They use a really tiny ATX power supply that convert the 12 volt input to > whatever the ATX pins are. These things cost about $20 The power supply > is some small it is built into the cable > https://www.amazon.com/Power-Supply-Htpc-Mini-box-Mini-itx/ > <https://www.amazon.com/Power-Supply-Htpc-Mini-box-Mini-itx/dp/B00QVK3ZPU/ref=sr_1_6?crid=1AAP9OO84LECA=mini+atx+pico+power+supply=1643862229=electronics=mini+atx+pico+power+supply%2Celectronics%2C124=1-6> > > On Wed, Feb 2, 2022 at 7:32 PM Andy Howell wrote: > >> I >> >> On 2/2/22 20:08, Chris Albertson wrote: >> > I followed that link.Wow, that is a good deal. Especially when you >> > look at the power supply. It uses a 12 volt barrel jack and a large >> size >> > wall-wort. The CPU burns all of 6 Watts.It could run on battery >> power. >> > >> > It is good to look for low-power PCs if they are going to run all day, >> > every day. The cost of power really adds up. The bix Xeon powered HP I >> > use for development work costs maybe 12 cents an hour. That is about >> $400 >> > at the end of the year. >> > >> > The machine I use to power LCNC and my 3D printer is an Intel i5 and I >> > could justify downgrading it based on power-saving along. I had not >> > realized there was such things at 6 Watt quad-cord Intel CPUs. >> >> Chris, >> >> Its says it uses a ATX 24pin power supply. Hope that is right, I'd >> rather not have rig up an external supply for it. Though I do have a >> couple that would work. >> >> Yeah, this will run 24/7. Like lights, the kids don't turn them off. >> >> I had a bunch of development servers in room in my garage. Noise and >> heat drove me to build the room. >> >> Thanks, >> >> Andy >> >> > >> > On Wed, Feb 2, 2022 at 3:35 PM Andy Howell wrote: >> > >> >> On 2/2/22 16:31, gene heskett wrote: >> >>> On Wednesday, February 2, 2022 4:42:22 PM EST Andy Howell wrote: >> >>>> I was hoping to update to a recent Debian and LinuxCNC version. >> >>>> However, I have a 32bit motherboard. This is for our school, so I'm >> >>>> trying to contain cost by just replacing the motherboard. >> >>>> >> >>>> Any suggestions for a 64bit mini-itx motherboard? >> >>>> >> >>> Most any NEW mobo today will need a new cpu too since the sockets are >> >>> changed and probably won't take your old memory for the same reason. >> >>> >> >>> ATM I'm running a normal sized Asus Z370-AII with the cheapest 6 core >> i5 >> >>> on it and I'm as happy as I can be. Asus makes decent stuff. Draws >> about >> >>> 140 watts less than the phenom it replaced. >> >>> >> >>> And stay away from OLOy memory, I had a failure and they needed more >> data >> >>> than I had to replace it, so I had to buy a different brand to replace >> >>> it. That's BS, so be sure, get it in writing, that you can get in >> >>> warranty replacements by simply shipping the bad one back with a >> >>> photocopy of the bill of sale. If they won't do that, go down the >> list to >> >>> the next vendor. >> >>> >> >> Gene, >> >> >> >> It looks like most of the Mini ITX boards have the cpu soldered in. I >> >> don't imagine I'd need a lot of memory. 2GB maybe? >> >> >> >> Looks like this will do the trick. $200 with 2GB of memory. >> >> >> >> https://www.onlogic.com/pd14ri/ >> >> >> >> Thanks, >> >> >> >> Andy >> >> >> >> >> >> >> >> >> >> >> >> >> >> ___ >> >> Emc-users mailing list >> >> Emc-users@lists.sourceforge.net >> >> https://lists.sourceforge.net/lists/listinfo/emc-users >> >> >> > >> >> >> ___ >> Emc-users mailing list >> Emc-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/emc-users >> > > > -- > > Chris Albertson > Redondo Beach, California > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Mini-itx 64bit mother board with parallel port
They use a really tiny ATX power supply that convert the 12 volt input to whatever the ATX pins are. These things cost about $20 The power supply is some small it is built into the cable https://www.amazon.com/Power-Supply-Htpc-Mini-box-Mini-itx/ <https://www.amazon.com/Power-Supply-Htpc-Mini-box-Mini-itx/dp/B00QVK3ZPU/ref=sr_1_6?crid=1AAP9OO84LECA=mini+atx+pico+power+supply=1643862229=electronics=mini+atx+pico+power+supply%2Celectronics%2C124=1-6> On Wed, Feb 2, 2022 at 7:32 PM Andy Howell wrote: > I > > On 2/2/22 20:08, Chris Albertson wrote: > > I followed that link.Wow, that is a good deal. Especially when you > > look at the power supply. It uses a 12 volt barrel jack and a large > size > > wall-wort. The CPU burns all of 6 Watts.It could run on battery > power. > > > > It is good to look for low-power PCs if they are going to run all day, > > every day. The cost of power really adds up. The bix Xeon powered HP I > > use for development work costs maybe 12 cents an hour. That is about > $400 > > at the end of the year. > > > > The machine I use to power LCNC and my 3D printer is an Intel i5 and I > > could justify downgrading it based on power-saving along. I had not > > realized there was such things at 6 Watt quad-cord Intel CPUs. > > Chris, > > Its says it uses a ATX 24pin power supply. Hope that is right, I'd > rather not have rig up an external supply for it. Though I do have a > couple that would work. > > Yeah, this will run 24/7. Like lights, the kids don't turn them off. > > I had a bunch of development servers in room in my garage. Noise and > heat drove me to build the room. > > Thanks, > > Andy > > > > > On Wed, Feb 2, 2022 at 3:35 PM Andy Howell wrote: > > > >> On 2/2/22 16:31, gene heskett wrote: > >>> On Wednesday, February 2, 2022 4:42:22 PM EST Andy Howell wrote: > >>>> I was hoping to update to a recent Debian and LinuxCNC version. > >>>> However, I have a 32bit motherboard. This is for our school, so I'm > >>>> trying to contain cost by just replacing the motherboard. > >>>> > >>>> Any suggestions for a 64bit mini-itx motherboard? > >>>> > >>> Most any NEW mobo today will need a new cpu too since the sockets are > >>> changed and probably won't take your old memory for the same reason. > >>> > >>> ATM I'm running a normal sized Asus Z370-AII with the cheapest 6 core > i5 > >>> on it and I'm as happy as I can be. Asus makes decent stuff. Draws > about > >>> 140 watts less than the phenom it replaced. > >>> > >>> And stay away from OLOy memory, I had a failure and they needed more > data > >>> than I had to replace it, so I had to buy a different brand to replace > >>> it. That's BS, so be sure, get it in writing, that you can get in > >>> warranty replacements by simply shipping the bad one back with a > >>> photocopy of the bill of sale. If they won't do that, go down the list > to > >>> the next vendor. > >>> > >> Gene, > >> > >> It looks like most of the Mini ITX boards have the cpu soldered in. I > >> don't imagine I'd need a lot of memory. 2GB maybe? > >> > >> Looks like this will do the trick. $200 with 2GB of memory. > >> > >> https://www.onlogic.com/pd14ri/ > >> > >> Thanks, > >> > >> Andy > >> > >> > >> > >> > >> > >> > >> ___ > >> Emc-users mailing list > >> Emc-users@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-users > >> > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Mini-itx 64bit mother board with parallel port
I followed that link.Wow, that is a good deal. Especially when you look at the power supply. It uses a 12 volt barrel jack and a large size wall-wort. The CPU burns all of 6 Watts.It could run on battery power. It is good to look for low-power PCs if they are going to run all day, every day. The cost of power really adds up. The bix Xeon powered HP I use for development work costs maybe 12 cents an hour. That is about $400 at the end of the year. The machine I use to power LCNC and my 3D printer is an Intel i5 and I could justify downgrading it based on power-saving along. I had not realized there was such things at 6 Watt quad-cord Intel CPUs. On Wed, Feb 2, 2022 at 3:35 PM Andy Howell wrote: > > On 2/2/22 16:31, gene heskett wrote: > > On Wednesday, February 2, 2022 4:42:22 PM EST Andy Howell wrote: > >> I was hoping to update to a recent Debian and LinuxCNC version. > >> However, I have a 32bit motherboard. This is for our school, so I'm > >> trying to contain cost by just replacing the motherboard. > >> > >> Any suggestions for a 64bit mini-itx motherboard? > >> > > Most any NEW mobo today will need a new cpu too since the sockets are > > changed and probably won't take your old memory for the same reason. > > > > ATM I'm running a normal sized Asus Z370-AII with the cheapest 6 core i5 > > on it and I'm as happy as I can be. Asus makes decent stuff. Draws about > > 140 watts less than the phenom it replaced. > > > > And stay away from OLOy memory, I had a failure and they needed more data > > than I had to replace it, so I had to buy a different brand to replace > > it. That's BS, so be sure, get it in writing, that you can get in > > warranty replacements by simply shipping the bad one back with a > > photocopy of the bill of sale. If they won't do that, go down the list to > > the next vendor. > > > Gene, > > It looks like most of the Mini ITX boards have the cpu soldered in. I > don't imagine I'd need a lot of memory. 2GB maybe? > > Looks like this will do the trick. $200 with 2GB of memory. > > https://www.onlogic.com/pd14ri/ > > Thanks, > > Andy > > > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Stepper PID
Many times with steppers, if the sound seems to go away, it didn't. You only move the frequency to a range your ears can't detect. Remember that the sound is likely not a primary frequency but maybe a harmonic or "beat" frequency and those would move unpredictable, or at least unintuitively. I'm trying to think how a PID P gain can affect noise. Typically, the way to get the P gain is to make it better until the system oscillates, then back down some. Very unscientific but this works, With the high P-value I wonder if noise was from the oscillation. Sound from motors is hard to predict. Some times it is obvious like a 1KHz step making a 1Khz tone. But it might be the pulse edge interacting with the motor housing. Some years ago, believe it or not, in the days when we would build motor drivers ourselves, I had a motor that sounded like an AM radio talk show. I could hear most of the words. The motor ran correctly at different speeds but my controller placed an audio signal on top of the DC so in this case, the sound was nothing to do with software setting. I had built an AM radio by accident. On Tue, Feb 1, 2022 at 11:44 PM Thaddeus Waldner wrote: > Hi, > > I have a pi4 running the 2.8.2 iso. I have it connected to a Mesa 7i96 and > from there I’m driving small stepper motors with tmc2209 “stepstick” > drivers. I think my step rate at max speed is around 8khz > > I get around +-70us? Max latency. > > I’ve been having an issue with what I think is latency which makes the > steppers sound like they have bad bearings. Raising the servo period to 2ms > made it worse. So I tried a faster servo thread and discovered the issue > pretty much goes away with the servo period at 250us. At that rate I start > running into joint following issues. > > I then learned that with the servo period at 1ms, if I lower the stepper > PID P parameter to 100 instead of the 1000 set by the configuration wizard, > the noise goes away completely and the steppers run buttery smooth. > > I have a few questions: > 1) what are the implications of running the PID stepper with p=100? > 2) why is servo thread jitter even making into the stepper pulse train > with a Mesa Ethernet card and no communication errors? Do I have anything > misconfigured? > > Thanks > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] motherboard that plays well with 7i43
Is there any disadvantage to placing a PCIe-based Parallel port card on a normal PC? On Sun, Jan 30, 2022 at 11:36 AM andy pugh wrote: > On Fri, 28 Jan 2022 at 18:18, dave engvall wrote: > > > I'm looking for users that are actually using a motherboard with a 7i43 > > and can recommend candidate boards from experience. Thanks in advance. > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] motherboard that plays well with 7i43
On Fri, Jan 28, 2022 at 10:18 AM dave engvall wrote: > hi, > I'm looking for users that are actually using a motherboard with a 7i43 > and can recommend candidate boards from experience. Thanks in advance. > A related question I have is "Does choice of mainboard matter?" I'd guess if you are using the USB interface it wouldn't. I was thinking of using the Ethernet interface Mesa board but then read it needs a dedicated Ethernet port in the host PC. This means buying, installing, and configuring an Ethernet card. The PC or any PC today already has half a dozen USB ports. Even a Raspberry Pi has 4 USB ports and might work well for this. But my question is not about using a specific PC, but rather if it matters. -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] 60's vintage position feedback.. (GE Accupins)
On Tue, Jan 25, 2022 at 6:37 PM John Dammeyer wrote: > > So how exactly are you going to count it?Time between the top trace > edge and the bottom trace edge? As you cross the threshold into the next > 0.1" and it mechanically jitters at that point what will you do to not have > counts go wild? > I would guess that something similar to a switch debounce would be used. This can be implemented in software.One simple algorithm is to sample the voltage and wait until you get let's say 5 identical readings. There are other tricks like low pass filtering. But the "N identical" method is easy to do and works well. -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] 60's vintage position feedback.. (GE Accupins)
This On Sun, Jan 23, 2022 at 4:30 PM andy pugh wrote: > On Sat, 22 Jan 2022 at 04:49, Chris Albertson > wrote: > > > > Arduino? Too slow to coune pulses if they happen every 0.0001 inch > unless > > the table moves slow. > > As long as it gets a couple of samples every 0.1" it should be fine. > Between the pins it's an absolute encoder, like a multi-turn resolver. > > Though I would still suggest using one of the M4 or M0 boards from > Adafruit, as the clock rate and floating point performance are just > orders of magnitude higher, for about twice the cost. > That is about what I suggested. "Pico" is a dual core M0. Adafruit has them for $4 each. https://www.adafruit.com/product/4864 So yes, orders of magnitude better but half, not twice the cost. The price is so low that it is little different from free. That for two M0 cores at 125 MHz and 8 of those programmable state machine, also at 125MHz. and 2 MB of flash. Very good documentation. Works with Arduino IDE and runs most Arduio code but you can program it with Python too. > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] rotary table re-engineeering
What you give a foundry is typically the part you want, a pattern. They can advise on the limits of what they can do. For example there is a minimum and maxim thickness and overall size. They can also tell you how much the iron will shrink so you can adjust the pattern Then for you own good you would want to make the pattern so as to minimize the amount of machine work. This also saves you money as you pay the foundry by the pound. So make the parrn have all the curves and shape you need. Wrapping is another issue. if you make it too thin or thick the result will not match the pattern. Tranditionally, patterns where made of wood or maybe wood with bondo over it and sanded and finished sooth. But today you would 3D print the pattern and give them a plastic copy of the part you want. They pack it in sand and make a mold from your pattern. Some can do a process that is like "lost wax" so then you print using a kind of plastic that the melted metal burns up But this is for a "*mini*-mill" you do NOT need cast iron parts. The cutting force is very small. Why not make your adaptor out of plastic? Platic seems like the wrong material because "real mills" are made of steel. I have a CNC converted Mini mill and the conversion parts are printed plastic.It you design, knowing the strength of that material you can do well. I can not measure flex with a dial indcator Plastic is like designing a part with glass. It is very strong and rigid until it shatters. But look at the size of the spindle motor and drive gears on a mini mill. You would shear the teeth off the drive gears before generating enough force to break a plastic adaptor shim. In any case, then after AFTER you know the plastic part works, then you give the foundry the known-to-work part to use as a pattern and they make you one in metal. I've dealt with some manufacturers and while you used to have to ship a pattern, now days you can email them the design file and they can print the pattern and do the design check that looks for violation of thier min and max thickness. There are also places that can do a 3D print in metals like aluminum of even stainless steel. But this is not yet cheap. On Sat, Jan 22, 2022 at 6:26 PM fxkl47BF--- via Emc-users < emc-users@lists.sourceforge.net> wrote: > > Yes, if you are near Windy Hill, then I understand that they do a lot > > of one-offs for vehicle restorers etc. > > > > It's certainly worth considering. > > OK > So from the photos in the ad what would I want to have made? > A big rectangle? > The only "true" surfaces on the back are the dovetails. > I think that's a bit of a wimpy mount. > If I disassemble it I can tell if the unfinished surfaces beside the > dovetails are thick enough to drill for additional supports. > Or is that a lot of overkill. > Just attach to the dovetails and be done with it. > I have a 3D printer. > This is go'n to to be fun :) > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] 60's vintage position feedback.. (GE Accupins)
Arduino? Too slow to coune pulses if they happen every 0.0001 inch unless the table moves slow. Look at the new Raspberry Pi "Pico". First off it only costs $4.Next if you looks, it can be programmed in Python and here I the Key: In addition to the two CPU cores it has 8 "PIO" processors and other can count pulses at about 100 MHz.A "PIO is a very fast hardware state machine and there are 8 of these on the chip. Not bad at 4 bucks. You can read the encoders with nearly zero CPU time, no pooling and no interrupts. The PIO is a way-simple processor. It has 8 instructions and 32 words of memory. but that is enough to handle quadrature. These state machines run at 125 Mhz The other thing about Pico, is the outstandingly good documentation. With 8 of these you can do multiple axis. There are several ways to program Pico. You can use the Arduino IDE and most Ardno stuff "just works". Or you can write in Python. uge the Pico into a USB and the PC thinks it is a thin drive. Sve the Python program to the thum drive and the chip is programmed. On Fri, Jan 21, 2022 at 12:59 AM andy pugh wrote: > On Fri, 21 Jan 2022 at 04:50, Sam Sokolik wrote: > > > > https://www.youtube.com/watch?v=ogUdjX3zerY > > > > The plan is to use the arduino to count the phase difference between the > > input and output. The original control used a 250khz clock frequency to > > count > > You might be able to do a lot better than that if you can find a way > to have the Arduino set timers on interrupts. > (How fast does the Arduino run a polling loop?) > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] motor coolant for water cooled spindles.
Advice from aquariu use. Place EVERYTHING inside the tank if you can, filters and all. Leaks then don't make a mess, broken hoses don't make a mess. Of course, you can not place the spindle motor in the tank but assuming the motor is the only thing not in the tank, there are only two connections that can leak. That said, you would have never noticed the filters were leaking if they were inside the tank. Also, the downside of placing the pump in the tank is the pump will heat the water. But maybe you don't run it enough to matter I was doing 8 or 12-hour runs. On Wed, Jan 19, 2022 at 4:07 PM gene heskett wrote: > On Sunday, January 16, 2022 3:34:37 PM EST gene heskett wrote: > replying to myself to issue a warning and a fix. > > > I just fired up the 6040 to make a pocket for a 3mm sq nut. Noted that > > I put > > half a cup of lysol bathroom spray in it. But haven't disconnected it > > so I can dump the tank in the back yard. Yet. > > > And I still haven't dumped it. I bought an undersink Colligan sediment > filter just because it looked like a higher quality unit than anything > else I'd looked at, a bit less than $50 at tsc. Then I went to the local > hdwe store and bought 2, 1/2" pipe to 1.4" hose barbs and 3/4" bushings > to fit the filter ports. Should have bought brass, these were split > molded, leaving a worn die fin at the junction of the die halves. Took 4 > layers a teflon tape to seal the threads and some permagoo on the barbs > to seal them down to about a drop a minute or less leakage. > > But the screw on filter can might as well not have had an o-ring seal, it > just poured out. Taking it apart again three or 4 times I finally grokked > it had a pressure limit built in and that if it had city water pressure > it would push the o-ring outward in its groove and seal. At 2 or 4 psi, > taint gonna happen. Then I spotted a raised edge on the bottom of the > cap, measured it and it would hit some carefully machined bumps located > outside the o-ring groove, so as it was maybe 5 thou high, I reasoned it > might get a better grip and seal on the o-ring if it was removed, so I > filed it off until the file was touching the rest of the bottom face of > the cap, then smoothed it up with a 600 grit diamond plate. Stop leak > worked. I've got about a drop every 2 or so minutes at the output adapter > > I've unplugged the motor, and left lcnc thinking its running about 4k, so > the pump is running, and all sorts of particulates can be seen in the > filter but I'm far more interested in how clear the mix looks tomorrow. > > Right now it's cloudy and only the ghost of the pump can be seen thru the > plastic tank. But it appears I may have a decent filter that doesn't seem > to restrict the flow. And with 2 spare 5 micron filters I've got maybe > $75 in it. > > Take care and stay well everybody. > > Cheers, Gene Heskett. > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author, 1940) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] motor coolant for water cooled spindles.
I've run these before. Just place a filter in the system. Whatever else you do is secondary. Dirt, bugs spiders and whatever don't matter if there is a filter. If you are not running a production shop that filter will see only a few hours a week of use and last "forever". The simplest possible setup is a bucket or tub with a submersible filter in the bottom of the bucket. Then the output and return lines go to the bottom of the back and if there is a leak it leaks back into the bucket. Then (key point) cover the pumps intake with a few wraps of "filter fluff" and secure it with a zip tie. If you want to streak the water use anything that is marked as SAFE for animals and pets. If it will not kill a cat, it is safe for your pump and motor. I think $3 Vodka would work well. As long as you have the filter zip tied to the intake, noting gets into the system. The other thing I used was a floating aquarium thermometer.This lets you know if the tank is large enough. If the temperature shoots up you need a bigger tank. Summry: (1) use a filter, after that no much else matters (2) use only "pet safe" chemicals but only if you have an actual problem that needs a solution. I keep aquarium tanks filled with just plain water and they remain clear for many years at a time. With no fish, any way-cheap filter will keep the water clear. An air-stone powered filter would work and cost nothing to run 24x7. he airstone also cools the water by turning it over. On Tue, Jan 18, 2022 at 8:27 AM gene heskett wrote: > On Tuesday, January 18, 2022 8:39:09 AM EST Todd Zuercher wrote: > > Automotive applications were predominately iron, aluminum, and some > > plastic with almost no copper for a very long time before different > > color antifreezes became common. > > > One side effect of cooling stuff like a spindle motor comes to mind. In > the vehicle, it quite easily achieves a kill everything temperature, > keeping even an active biological mix reasonably sterile if it doesn't > sit too long. > > Cooling a spindle motor does not generally reach those temps but the > small elevation in heat it does achieve encourages the biological > contaminants to grow faster. So I might setup a cron job to run the pump > for an hour in the wee hours if lcnc isn't running, and it would be easy > enough to dump the return line into a leaky spoon, which if it gets > enough flow would activate a micro-switch to tally enough flow, and have > the cron job send me a change the filter email if, say 5 minutes before > shutdown, the switch isn't closed. Or yell if the switch is closed before > the pump starts because the hole is plugged. It probably won't happen, > but that is what you get when I let my imagination out to play w/o a > chaperone. ;) > > Cheers, Gene Heskett. > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author, 1940) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > > > > _______ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] motor coolant for water cooled spindles.
I would stick with water. it is cheap and has very high specific heat. It is really hard to beat water. I had a water cooling system used to cool the hot side of a thermoelectric module. I used an old 50-gallon aquarium tank as the reservoir and for a pump, I used an aquarium cartridge filter as the pump. I figured that anything coming out of that filter was only a few microns across and would not harm the chiller.Even if the tank filled with green algae none of it would get in the chiller.The water actually remained clear for months. I thought about putting goldfish in the tank. Put then I'd have to maintain the filter more. I really do recommend using a filter. But a big size pleated paper cartridge sized so it can last over a year or two. The tank stays clean Why 50 gallons? I wanted the water temperature to remain constant or at least not change too quickly as the chiller used a PI control loop that assumed a near-constant water temperature. In a recirculating system you have to look at how many watts are being dumped into theater and how fast the tank water heats. This system was designed to run s 8 hours at a time. It would gain only a few degrees in that time Looking back on this project, others with similar equipment just used an "open-loop" system of tap water. The water came out of the tap through the machine then into the flower bed outdoors. The flow rate is low enough in a cooling system not to worry about saving water. On Sun, Jan 16, 2022 at 2:08 PM andy pugh wrote: > On Sun, 16 Jan 2022 at 21:44, gene heskett wrote: > > > I first started out with rv antifreeze, but it spoiled and jelled solid > > I think I would be tempted to try the waterless coolant that they sell > for cars, though that would depend on how much was needed. It's not > cheap. > > https://motocrossactionmag.com/ten-things-you-need-to-know-about-waterless-coolant/ > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] shop air? Not No, but Heck No
This thing looks to be an oversized aquarium pump. A small pump like that has a lot of uses. It might be made for an air brush.At 2 liters per minute it does not move a lot of air but maybe enough to spray a fine mist. What it lacks that you shop air has it an accumulator tank and a pressure switch to turn the motor on and off. For low pressure, you could make a tank from a length of black ABS waste line pipe. I'm about ready to get rid of my big compressor. The only one I use anymore is the portable unit for a nail gun. On Sun, Jan 16, 2022 at 3:24 AM gene heskett wrote: > I found it!! > > What it it? It is an air pump, diaphram type, delivering 10L of air a > minute, at pressures up to 23 psi. Running on 12 volts, 1.2 amps. $6.95 a > copy. > > <https://www.mpja.com/12V-Diaphram-Pump-Air/productinfo/36394+MI/> > > To me trying to justify running my mister on shop air wasting all that > horsepower to make 125 psi air which I have to regulate down to < 5 psi > to feed the mister its air is almost criminal. Not to mention this is > 72db, quite a bit quieter than a 2hp harbor freight compressor. > > Cheers, Gene Heskett. > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author, 1940) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC and Stepper motors.
for that matter) > with Linuxcnc directly. The bldc hal component does the relevant > calculations already. The only slightly tricky part is measuring > phase current. You could use an A/D converter with SPI interface > to the Mesa FPGA, but it is still harder than grabbing analog values > in a microcontroller. > > Personally, I am more interested in the approach you've been > advocating for years, of having a microcontroller per motor handling > commutation and drive, with the Linuxcnc/Mesa setup sending > step/direction pulses or communicating via sserial or Ethercat with > each motor. I've built a number of mosfet bridge boards for 2 or 3-phase > drive of small motors which have been good enough for in-house use, > and it isn't too difficult. Making them bullet-proof for other end-users > to connect to their own motors and power supplies would require some > design hardening. > > The bldc hal component is great for experimenting with driving brushless > motors, though, if you don't need phase current feedback in your control > scheme. > > -- Ralph > > From: Chris Albertson [albertson.ch...@gmail.com] > Sent: Wednesday, January 12, 2022 6:57 PM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC > and Stepper motors. > > CAUTION: This email originated from outside the Walla Walla University > email system. > > > Thanks, this might be of value to someone who wants a VFD for the cost of a > few h-bridges. > > On Wed, Jan 12, 2022 at 6:35 PM Ralph Stirling < > ralph.stirl...@wallawalla.edu> wrote: > > > The bldc.c component computes the two-phase sin/cos > > terms from phase angle, but I don't think it outputs them, > > so a very slight modification would be necessary. The sin/cos > > output values would be connected to PWM inputs in hostmot2, > > which would generate the actual signals for two H-bridges to > > switch the motor phases. The Mesa card generates the high > > speed PWM. A 1kHz servo thread should be sufficient for > > computing the sin/cos values. Here are the relevant lines in > > bldc.c: > > > > sintheta = sin(phase_angle * pi2); > > costheta = cos(phase_angle * pi2); > > A_value = out_abs * costheta; > > B_value = out_abs * (costheta * cos120 + sintheta * > > sin120); > > C_value = out_abs * (costheta * cos120 - sintheta * > > sin120); > > > > The sintheta and costheta are appropriate for two-phase steppers, > > and the A, B, and C_values are for three-phase bldc motors. > > > > The sintheta and costheta values could be multiplied by the output > > of a PID loop to adjust the amplitude based on torque or velocity > > requirements. Bldc connects to an encoder counter to track rotor angle. > > > > -- Ralph > > > > From: Chris Albertson [albertson.ch...@gmail.com] > > Sent: Wednesday, January 12, 2022 5:41 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC > > and Stepper motors. > > > > CAUTION: This email originated from outside the Walla Walla University > > email system. > > > > > > So how would I connect the four wire stepper motor to my PC to allow for > > software commutation and continuous rotation of a stepper? > > > > If this works, it could save people money by eliminating the need for a > VFD > > spindle drive and also > > People are paying quite a lot for these self-contained "closed-loop > > steppers"If this can already be done in software how? > > > > But I have serious doubts a Linuc PC has good enough timing even with an > RT > > kernel. But in theory, how to connect the stepper? > > > > > > On Wed, Jan 12, 2022 at 1:26 AM andy pugh wrote: > > > > > On Wed, 12 Jan 2022 at 01:28, Chris Albertson < > albertson.ch...@gmail.com > > > > > > wrote: > > > > > > > This library would allow the stepper to be commanded > > > > as if it were a serial interfaced servo motor. > > > > > > In principle so does LinuxCNC HAL and the "bldc" HAL component. > > > > > > -- > > > atp > > > "A motorcycle is a bicycle with a pandemonium attachment and is > > > designed for the especial use of mechanical geniuses, daredevils and > > > lunatics." > > > — George Fitch, Atlanta Constitution
Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC and Stepper motors.
Thanks, this might be of value to someone who wants a VFD for the cost of a few h-bridges. On Wed, Jan 12, 2022 at 6:35 PM Ralph Stirling < ralph.stirl...@wallawalla.edu> wrote: > The bldc.c component computes the two-phase sin/cos > terms from phase angle, but I don't think it outputs them, > so a very slight modification would be necessary. The sin/cos > output values would be connected to PWM inputs in hostmot2, > which would generate the actual signals for two H-bridges to > switch the motor phases. The Mesa card generates the high > speed PWM. A 1kHz servo thread should be sufficient for > computing the sin/cos values. Here are the relevant lines in > bldc.c: > > sintheta = sin(phase_angle * pi2); > costheta = cos(phase_angle * pi2); > A_value = out_abs * costheta; > B_value = out_abs * (costheta * cos120 + sintheta * > sin120); > C_value = out_abs * (costheta * cos120 - sintheta * > sin120); > > The sintheta and costheta are appropriate for two-phase steppers, > and the A, B, and C_values are for three-phase bldc motors. > > The sintheta and costheta values could be multiplied by the output > of a PID loop to adjust the amplitude based on torque or velocity > requirements. Bldc connects to an encoder counter to track rotor angle. > > -- Ralph > > From: Chris Albertson [albertson.ch...@gmail.com] > Sent: Wednesday, January 12, 2022 5:41 PM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC > and Stepper motors. > > CAUTION: This email originated from outside the Walla Walla University > email system. > > > So how would I connect the four wire stepper motor to my PC to allow for > software commutation and continuous rotation of a stepper? > > If this works, it could save people money by eliminating the need for a VFD > spindle drive and also > People are paying quite a lot for these self-contained "closed-loop > steppers"If this can already be done in software how? > > But I have serious doubts a Linuc PC has good enough timing even with an RT > kernel. But in theory, how to connect the stepper? > > > On Wed, Jan 12, 2022 at 1:26 AM andy pugh wrote: > > > On Wed, 12 Jan 2022 at 01:28, Chris Albertson > > > wrote: > > > > > This library would allow the stepper to be commanded > > > as if it were a serial interfaced servo motor. > > > > In principle so does LinuxCNC HAL and the "bldc" HAL component. > > > > -- > > atp > > "A motorcycle is a bicycle with a pandemonium attachment and is > > designed for the especial use of mechanical geniuses, daredevils and > > lunatics." > > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Cralph.stirling%40wallawalla.edu%7Cd905773e126a482dd8ab08d9d635e8eb%7Cd958f048e43142779c8debfb75e7aa64%7C0%7C0%7C637776349295328651%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=c2gJ5So48tGLS0ru6%2Fdp6hyrZeGgtYU9A02nJ78jrOU%3Dreserved=0 > > > > > -- > > Chris Albertson > Redondo Beach, California > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Cralph.stirling%40wallawalla.edu%7Cd905773e126a482dd8ab08d9d635e8eb%7Cd958f048e43142779c8debfb75e7aa64%7C0%7C0%7C637776349295328651%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=c2gJ5So48tGLS0ru6%2Fdp6hyrZeGgtYU9A02nJ78jrOU%3Dreserved=0 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC and Stepper motors.
So how would I connect the four wire stepper motor to my PC to allow for software commutation and continuous rotation of a stepper? If this works, it could save people money by eliminating the need for a VFD spindle drive and also People are paying quite a lot for these self-contained "closed-loop steppers"If this can already be done in software how? But I have serious doubts a Linuc PC has good enough timing even with an RT kernel. But in theory, how to connect the stepper? On Wed, Jan 12, 2022 at 1:26 AM andy pugh wrote: > On Wed, 12 Jan 2022 at 01:28, Chris Albertson > wrote: > > > This library would allow the stepper to be commanded > > as if it were a serial interfaced servo motor. > > In principle so does LinuxCNC HAL and the "bldc" HAL component. > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC and Stepper motors.
On Tue, Jan 11, 2022 at 12:14 PM Todd Zuercher wrote: > Isn't this how most closed loop stepper drives claim to work? > It might be, but the commercial closed-loop stepper drivers retain the step/dir interface, so they can be commanded only to position but not velocity or torque. This library would allow the stepper to be commanded as if it were a serial interfaced servo motor. The other neat thing about this library is it is open source and designed to use a wide range of cheap and available hardware. If you drill down into their web site they have a procedure and advice for picking the parts and then based on the parts they show how to use the library to make a controller. The documentation is excellent and aims at a reader with only mid-level technical background. This last part is unique for FOC controllers. Usually, this subject is a very complex specialty engineering subject. They made it accessible. I'll try it on the Z-axis of my mini mill. Likely the hardest part will be learning how to interface this with LCNC. > > Todd Zuercher > P. Graham Dunn Inc. > 630 Henry Street > Dalton, Ohio 44618 > Phone: (330)828-2105ext. 2031 > > -Original Message- > From: Chris Albertson > Sent: Tuesday, January 11, 2022 2:06 PM > To: Enhanced Machine Controller (EMC) > Subject: [Emc-users] Field Oriented Control (FOC) algorithm for BLDC and > Stepper motors. > > [EXTERNAL EMAIL] Be sure links are safe. > > Here is a different way to control a stepper motor. This control method > treats the stepper as if it is just a plain old BLDC motor but with 50 > poles and 2 phases. Positional accuracy is determined by an encoder, not > the number of steps as the controller is not doing "steps" but using the > motor as a continuous motion servo. It seems like this could be a > retrofit to a machine that uses steppers to give an increased performance > while keeping the same motors. > > You can read the theory used here: > > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fen.wikipedia.org%2Fwiki%2FVector_control_data=04%7C01%7Ctoddz%40pgrahamdunn.com%7Cc21d1504e33b42a15eaf08d9d535a470%7C5758544c573f47cebee96c3e0806fb43%7C0%7C0%7C637775249503911306%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=CnWkF%2BhmPNQmnNLXcRJePq%2FO1YdMMdCBvrBKVdACd%2Bk%3Dreserved=0(motor) > > What this project offers is not a controller, but Open Source software > that runs on any of a shortlist of microcontrollers and uses a long list of > sensors and motors. The website was a project design walk-through. This is > also, as you can see very low cost, he is using a $5 driver board from eBay. > > What they are saying is that small steppers have better performance and use > less power when driven using their closed-loop algorithm. But it does > require that you place a current sensor (Hall effect or in in-line > resister) on both phases and drive the stepper with a PWM capable pair of > full-H-bridges. What is going on is the control software is measuring > current and forcing it to have the desired waveform against the inductance > and mechanical loads. > > > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyoutu.be%2Fzcb86TRxTxcdata=04%7C01%7Ctoddz%40pgrahamdunn.com%7Cc21d1504e33b42a15eaf08d9d535a470%7C5758544c573f47cebee96c3e0806fb43%7C0%7C0%7C637775249503911306%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=GXOX24h9HBSe3wGa%2BTwmhzw7vMc3E0SGfFUrAfZL8gw%3Dreserved=0 > > Here is the project website > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fsimplefoc.com%2Fdata=04%7C01%7Ctoddz%40pgrahamdunn.com%7Cc21d1504e33b42a15eaf08d9d535a470%7C5758544c573f47cebee96c3e0806fb43%7C0%7C0%7C637775249503911306%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=0xzqDAizMu8CL9EgF0yVHA7pXwJJUf%2BQ72JUBh8Zpu0%3Dreserved=0 > > > > > -- > > Chris Albertson > Redondo Beach, California > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Ctoddz%40pgrahamdunn.com%7Cc21d1504e33b42a15eaf08d9d535a470%7C5758544c573f47cebee96c3e0806fb43%7C0%7C0%7C637775249503921326%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=Rv%2BriLj8Cg5KjS8l6PQ2VWG6K%2FRg4%2BHcIqi0jH2Hvr8%3Dreserved=0 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Field Oriented Control (FOC) algorithm for BLDC and Stepper motors.
Here is a different way to control a stepper motor. This control method treats the stepper as if it is just a plain old BLDC motor but with 50 poles and 2 phases. Positional accuracy is determined by an encoder, not the number of steps as the controller is not doing "steps" but using the motor as a continuous motion servo. It seems like this could be a retrofit to a machine that uses steppers to give an increased performance while keeping the same motors. You can read the theory used here: https://en.wikipedia.org/wiki/Vector_control_(motor) What this project offers is not a controller, but Open Source software that runs on any of a shortlist of microcontrollers and uses a long list of sensors and motors. The website was a project design walk-through. This is also, as you can see very low cost, he is using a $5 driver board from eBay. What they are saying is that small steppers have better performance and use less power when driven using their closed-loop algorithm. But it does require that you place a current sensor (Hall effect or in in-line resister) on both phases and drive the stepper with a PWM capable pair of full-H-bridges. What is going on is the control software is measuring current and forcing it to have the desired waveform against the inductance and mechanical loads. https://youtu.be/zcb86TRxTxc Here is the project website https://simplefoc.com/ -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] OT: mouses
> > > > > Does the space mouse work under linux? > Let's say it does work under Linux. Do any Linux-based CAD systems work with the Space Mouse? ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Looking for a really small motor
I'm looking for a small motor for a mobile robot. I thought I'd ask here on the chance that there are very small size servos used for machine tools that might work for me.I need very good power to weight ratio as this is a battery-powered machine and has 12 to 16 motors.The motors need to be good at holding positions at zero speed. My guess is that there is nothing like this available, and I'll have to adapt a quad-copter motor. I need at least a dozen, eventually two dozen so "cost matters" I'm currently using R/C hobby servos. Modern bus-servos are very good, except for their speed. I'd like something MUCH faster. THese will be used to power the legs of a walking machine and need to be fast enough to respond to a 100 Hz control loop, preferable 1KHz. BTW for those making something like a tool changer these would be great. Features are - 25 kg-cm holding torque - commands are sent over a serial bus. All you need is one serial port and a power supply to drive many servos - you can query it for its current position - you can specify a speed - that can act as position servos over a 240-degree range and have a second "speed" servo mode that does full 360 multiple rotations. They claim 1/3rd degree accuracy. - They self-enforce angle and rate limits so as to not burn themselves up trying to reach an impossible set point. - They are not expensive at $21 each But they are s-l-o-w. About 300 degrees per second is the maximum speed. This is OK for many things but I need something much faster and about the same torque amazon.com/dp/B09CMGN25Q... <https://www.amazon.com/dp/B09CMGN25Q/ref=sspa_dk_hqp_detail_aax_0?psc=1=ZW5jcnlwdGVkUXVhbGlmaWVyPUFCNE83T01ZSDNYNUomZW5jcnlwdGVkSWQ9QTA2MzQ1NDQzOUNJWjVFUlE1SUw0JmVuY3J5cHRlZEFkSWQ9QTAwODUxMzJDR1VVOVI3RjhMVkYmd2lkZ2V0TmFtZT1zcF9ocXBfc2hhcmVkJmFjdGlvbj1jbGlja1JlZGlyZWN0JmRvTm90TG9nQ2xpY2s9dHJ1ZQ==> Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] OT: mouses
I have just a standard Microsoft cordless mouse. It is connected to a Linux PC and is used too many hours a week on CAD (Fusion 360) Which CAD you use might matter because I think some systems allow you to program shortcuts. I think all you really need is one that works smoothly and reliably and has a center scroll wheel. On Mon, Jan 10, 2022 at 9:14 AM fxkl47BF--- via Emc-users < emc-users@lists.sourceforge.net> wrote: > I'm in the market for a "decent" mouse for cad. > There is a passel of them out there. > Many of them are very pricey. > I'm just a casual user. > And a lefty. > Do y'all have any recommendations? > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC bridge.
This is why I said to build the 3D printed arm. You gain a lot of experience with a table top sized robot even if it is not the $100K metal arm you dreamed of. Lots of people can assemble metal parts and buy expensive motors. But purposeful behavior and REALLY hard. What you don't want is a $100K robot that can't do anything. Start with a $150 arm that is made of plastic and uses $10 stepper motors. Get that arm to pick up a pencil from wherever you place it on the table and then put the pencil in a plastic coffee mug. I would say if you can't do this, don't even think about building a robot that can manufacture children's toys or make cars. Too many people try and bypass the design process and jump to the final product. I think it is Youtube that drives this. You can't get a million views if all you do is left a pencil 1 inch off the table, drop it, then pick it up and drop it again 20 times in a row and say "next month I hope to be able to move the pencil to the space over the coffee cup." Doing that would only impress engineers, not the public. On Thu, Jan 6, 2022 at 12:33 AM John Dammeyer wrote: > Anyone looked at this guy and his robot arm etc. Nice to see someone else > makes almost as many mistakes as I do. > https://www.youtube.com/watch?v=T3fl1FIKEQo > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: January-05-22 5:58 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC > bridge. > > > > The current repository is "machinekit-hal". They only work on HAL. > There > > are no CNC apps in the current repository. All the stuff for running CNC > > equipment is removed now.When the developers moved from "machinekit" > to > > "machinekit-hal" they abandoned all the CNC-related stuff. It is all on > > GitHub, go look. > > > > But it does seem to be actively developed although the last update was > > almost two weeks ago. I think someone took a Christmas break. > > > > If I had to guess what Tormach does, it would be to use their own > > closed-source "Path Pilot" app along with HAL from Machine kit. But I'm > > guessing.They are funding the ROS2 work. > > > > Someone asked, "What is ROS2?" It is the planner and motion > controller > > that is now probably the most-used.The part of ROS used for > > manipulators (arms) is called "MoveIt" and a Google search will turn up > > information on MoveIt.Motion planning is harder with a robot arm > > because they don't usually run pre-programmed code. In the CNC world > the > > motion path is designed when the g-code is written. But robot controls > > have to do this in real time as the world around them changes. > > > > When the palletizing robot lifts a box to be placed on a pallet the box > > might be in a different location and of course the pallets are placed > more > > or less randomly wherever the human forklift driver left it. Even a pick > > and place robot has to find the parts which might be loose-packed in a > open > > bin. Moveit is open source and integrated with ROS. ROS is larger > > "Robot Operating System" that includes things for communications, driving > > wheeled robots and other stuff like MoveIt. > > > > Not using MoveIt to move a robot arm is like saying "LCNC is too > complex, > > I'll just write my own code on an Arduino to read g-code and generate > steps > > for my 4-axis mill." You could, maybe, but it would take more than a few > > days. > > > > For those who like looking at interesting robots here is a video from > 2019, > > these have been improved but are not moving boxes in warehouses. You can > > see that the task they are doing can't be pre-programmed. Really this > is > > not much more than a CNC mill. The wheels but the machine in (X,Y and > the > > orange suction cup-thing is like a spindle that moves in Z, it just > > operates in a larger space than a mill table > > https://youtu.be/5iV_hB08Uns > > > > > > > > > > On Wed, Jan 5, 2022 at 11:49 AM Ralph Stirling < > > ralph.stirl...@wallawalla.edu> wrote: > > > > > The newest version of Machinekit separates "emc" from hal, > > > and permits their hal to compile with standard Linuxcnc 2.8 > > > if I recall correctly. Their focus has always been on the > > > hal end of things rather than gui's and cnc machining. > > > > > > I can't find the email post that explained t
Re: [Emc-users] Very good software backlash control demo.
this robot that loads pencil leads it really just a software testing device. A research tool. What is new here s not the idea too use two sensors but 1) use a predictive model that says how the system will flex in the future in s given acceleration is apllied. 2) collect data while the machine is running in so that the above model is up to the millisecond calibration data. Notice the robot starts by slowly placing the lead in that speeds up at the end. The author said this is because the machine runs only as fast as it as confidence on the predictive model. Once you start an operation you coect data and can speed up in real time. To do this on a mill you would need glass encoders physically connected to the table, a rotary encoder on the servo motor. But really, you need sensors any place that can inform your physic model. Again what is new here that LCNC doe not do is attempt to predict in advance the effect of changing the motor torque. A PID loop, ow two nested PID loops can ONLY see error after the fact. PID is reactive. But real world machine always have a delay between applied torque and action because the physical material (even cast iron) is elastic. The reason he used these "crappy" hobby servos and long, thin aluminum extrusions was not because he was dumb or broke. But because H wanted a system with loads of flex and backlash so that he could test his ability to measure flex and backlash. Technically what he is doing is using a Kalman Filter to estimate the position of the robot's hand. KF is mathematically complex is really not hard to understand intuitively.The idea is that if you have many independent source of information you can combine those source and the result is much better then any of them.The Wikipedia article explainsbut is not easy reading, look up to tutorials if you like ut basically the KF use a model to predict the current stat from measurements that were taken in the past. It is not a reactive controller lk PID. On Tue, Jan 4, 2022 at 11:27 AM dave engvall wrote: > Hi, > I ran across an article by one of the servo drive firms that said the > using two sensors/axis improved motion. I think they used a resolver on > the drive and an encoder on the ball screw. While this improves motion > that assumes you have very good ball screws. I think that a better > approach would be to use glass scales and then another sensor on the > ball screw or drive. Right now I have a mill with glass scales on X and > Y with Z being considered. Planned is implementation of stepper-servos > like Gene's setup. First iteration will be stepper-servos plus a DRO > simply to tell me how bad things are. > BTW -- on X using the hand wheel vis a vis an accu-rite 5 um glass scale > was dead on moving X in the positive direction I won't admit how bad > it is on reversal. ;-) Y is damaged but still pretty good going positive > but really bad in Y neg. No scale on Z so no estimate but not perfect. > The X does have a 40 K count/inch in quadrature encoder on it. > Note: Chinese glass scales are now mounted on X and Y since the > Accu-rite scales do not play well with the Chinese DRO. > > Cheers, > Dave > > > > On 1/3/22 6:58 PM, Sam Sokolik wrote: > > This has been done at a bigger scale... (2 feedback loops) > > > > > https://web.archive.org/web/20160222165548/http://jmkasunich.com/cgi-bin/blosxom/shoptask/wichita-trip-02-20-08.html > > > > > http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Combining_Two_Feedback_Devices_On_One_Axis > > > > On Mon, Jan 3, 2022, 8:42 PM Chris Albertson > > wrote: > > > >> CNC is about precision motion control. Here is a new idea where this > >> builder gets 0.05mm accuracy but uses hobby level R/C servo Not only > that, > >> but he connects three of these in series, one to the next to the next so > >> all the backlashes and poor tolerances add together. Then he uses this > to > >> do precise motion. He loads a mechanical pencil with this chain of > cheap > >> parts. > >> > >> What does this means for LCNC? It means that someone has found a > software > >> solution to backlash. What he does is place a quadrature encoder on > both > >> the motor and the output shaft.The difference in encoder reading is > an > >> exact measure of mechanical backlash and effective gear ratio.He can > >> measure the backlash under different conditions and store the > >> measurements. Then he places a cascaded PID controller and Kalman > filter > >> over this hardware. > >> > >> Technically the problem with backlash control via software is the delay > >> from input to output pays poorly with the PID algorithm. He applies a > &g
Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC bridge.
The current repository is "machinekit-hal". They only work on HAL. There are no CNC apps in the current repository. All the stuff for running CNC equipment is removed now.When the developers moved from "machinekit" to "machinekit-hal" they abandoned all the CNC-related stuff. It is all on GitHub, go look. But it does seem to be actively developed although the last update was almost two weeks ago. I think someone took a Christmas break. If I had to guess what Tormach does, it would be to use their own closed-source "Path Pilot" app along with HAL from Machine kit. But I'm guessing.They are funding the ROS2 work. Someone asked, "What is ROS2?" It is the planner and motion controller that is now probably the most-used.The part of ROS used for manipulators (arms) is called "MoveIt" and a Google search will turn up information on MoveIt.Motion planning is harder with a robot arm because they don't usually run pre-programmed code. In the CNC world the motion path is designed when the g-code is written. But robot controls have to do this in real time as the world around them changes. When the palletizing robot lifts a box to be placed on a pallet the box might be in a different location and of course the pallets are placed more or less randomly wherever the human forklift driver left it. Even a pick and place robot has to find the parts which might be loose-packed in a open bin. Moveit is open source and integrated with ROS. ROS is larger "Robot Operating System" that includes things for communications, driving wheeled robots and other stuff like MoveIt. Not using MoveIt to move a robot arm is like saying "LCNC is too complex, I'll just write my own code on an Arduino to read g-code and generate steps for my 4-axis mill." You could, maybe, but it would take more than a few days. For those who like looking at interesting robots here is a video from 2019, these have been improved but are not moving boxes in warehouses. You can see that the task they are doing can't be pre-programmed. Really this is not much more than a CNC mill. The wheels but the machine in (X,Y and the orange suction cup-thing is like a spindle that moves in Z, it just operates in a larger space than a mill table https://youtu.be/5iV_hB08Uns On Wed, Jan 5, 2022 at 11:49 AM Ralph Stirling < ralph.stirl...@wallawalla.edu> wrote: > The newest version of Machinekit separates "emc" from hal, > and permits their hal to compile with standard Linuxcnc 2.8 > if I recall correctly. Their focus has always been on the > hal end of things rather than gui's and cnc machining. > > I can't find the email post that explained this, and the machinekit > documentation is rather sparse, so how one actually installs > things with current LinuxCNC high level code and Machinekit > hal code, I couldn't tell you. > > -- Ralph > > From: Rob C [rclandro...@gmail.com] > Sent: Wednesday, January 5, 2022 10:53 AM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC > bridge. > > CAUTION: This email originated from outside the Walla Walla University > email system. > > > I was also confused by machine kit reference from tormach. I thought > tormach was based on linuxcnc and not the machine kit branch/fork > > although I guess if Tormach are running a 2.7 linuxcnc version I guess you > could say they have their own branch/fork > > I would have thought that since the change to python 3 with linuxcnc that > will be quite a divergent change to linuxcnc that will make it difficult to > integrate any items pushed back to 2.7 > > On Wed, 5 Jan 2022, 18:38 John Dammeyer, wrote: > > > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > > > > > On Wed, Jan 5, 2022 at 3:29 AM Gregg Eshelman via Emc-users < > > > emc-users@lists.sourceforge.net> wrote: > > > > > > > That repeatability is plenty good enough to drop work into an > automatic > > > > holding fixture (which does the final precision positioning) then > pick > > the > > > > pieces out for transfer to another manufacturing stage or into a box. > > > > > > > > > > > > As it turns out, for all the really hard tasks, +/- a quarter-inch > > (0.25") > > > is accurate enough. You only need 0.001 inches for the simple no-brain > > > jobs. So my point is that you can do a LOT with the cheaper 3D > printed > > > plastic version of this robot. > > > > > > I'm surprised there are not more people here working on robots. hey > are > > > very much like CNC tools but more
Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC bridge.
On Wed, Jan 5, 2022 at 3:29 AM Gregg Eshelman via Emc-users < emc-users@lists.sourceforge.net> wrote: > That repeatability is plenty good enough to drop work into an automatic > holding fixture (which does the final precision positioning) then pick the > pieces out for transfer to another manufacturing stage or into a box. > Not many people have a need to load parts into a CNC machine.The big market for robotic manipulators is going to be cleaning toilets and clearing the dishes off dinner tables and placing them in a dishwasher. I'd guess that the domestic market is literally one million times larger than the industrial market. The industrial use of robots is dead-easy. They run on g-code and make precise pre-programmed motions. Clearing dishes is at the other end of the scale of difficulty. But there are intermediate use cases that are not crazy-hard like moving boxes between trucks and warehouses and say picking up trash in a public park. As it turns out, for all the really hard tasks, +/- a quarter-inch (0.25") is accurate enough. You only need 0.001 inches for the simple no-brain jobs. So my point is that you can do a LOT with the cheaper 3D printed plastic version of this robot. I'm surprised there are not more people here working on robots. hey are very much like CNC tools but more interesting. > > >On Tuesday, January 4, 2022, 02:36:42 PM MST, Ralph Stirling < > ralph.stirl...@wallawalla.edu> wrote: > > That AR3 robot is indeed interesting. In the event > that anyone else wants to know the specs on the > robot, here is what I found on page 298 of the manual: > > Reach – 24.75 inches (62.9cm) > Payload – 4.15 lbs (1.9kg) > Repeatability - .2mm > Robot weight (aluminum) – 27lbs (12.25kg) > Enclosure weight – 12.5lbs (5.6kg) > Max Power Consumption – 8.25amp (198 watts) > > Reach and payload are decent, but repeatability is 10x > worse than typical industrial arms. Still a very intriguing > system. Not sure I envy the designer trying to support > buyers of his kits though. There are an awful lot of bits > and pieces and assembly steps. > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC bridge.
You can 3D print this robot too. He has the design files available for download.What sets his robor arm ahead f all the other similar projects (there are MANY) is that he specified very heavy duty taper roller bearing in place of the more common deep groove barings. Once you have a sturdy arm with good bearing, then he connects standard gear-reduced stapper motors as sold at "stepper-online." The printed arm with steppers is good enough for many tasks But with robots, always the bottleneck is software. His software is primitive s you'd be on your own. On Tue, Jan 4, 2022 at 1:37 PM Ralph Stirling wrote: > That AR3 robot is indeed interesting. In the event > that anyone else wants to know the specs on the > robot, here is what I found on page 298 of the manual: > > Reach – 24.75 inches (62.9cm) > Payload – 4.15 lbs (1.9kg) > Repeatability - .2mm > Robot weight (aluminum) – 27lbs (12.25kg) > Enclosure weight – 12.5lbs (5.6kg) > Max Power Consumption – 8.25amp (198 watts) > > Reach and payload are decent, but repeatability is 10x > worse than typical industrial arms. Still a very intriguing > system. Not sure I envy the designer trying to support > buyers of his kits though. There are an awful lot of bits > and pieces and assembly steps. > > -- Ralph > > From: Rob C [rclandro...@gmail.com] > Sent: Tuesday, January 4, 2022 10:25 AM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC > bridge. > > CAUTION: This email originated from outside the Walla Walla University > email system. > > > Interesting looking at the AR2/AR3 > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.anninrobotics.com%2Fdata=04%7C01%7Cralph.stirling%40wallawalla.edu%7Cc2fc608fe31b49905c3e08d9cfaf96e2%7Cd958f048e43142779c8debfb75e7aa64%7C0%7C1%7C637769175325588286%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=1OEAmYT8Al3yGr6i5FuTXiRgXEF2oNOsibOvSqLdrTg%3Dreserved=0 > > On Tue, 4 Jan 2022 at 18:14, andy pugh wrote: > > > On Tue, 4 Jan 2022 at 17:41, Robin Szemeti via Emc-users > > wrote: > > > > > > Out of interest, how is PathPilot related to LinuxCNC? > > > > It's LinuxCNC with their own GUI on top. But maybe v2.7 as I don't > > _think_ that they have the joints-axes changes in. > > > > -- > > atp > > "A motorcycle is a bicycle with a pandemonium attachment and is > > designed for the especial use of mechanical geniuses, daredevils and > > lunatics." > > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Cralph.stirling%40wallawalla.edu%7Cc2fc608fe31b49905c3e08d9cfaf96e2%7Cd958f048e43142779c8debfb75e7aa64%7C0%7C1%7C637769175325588286%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=5vevh1cJOszulDkv2LJY%2B6hhQijKW0z8ko4CcONelAs%3Dreserved=0 > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Cralph.stirling%40wallawalla.edu%7Cc2fc608fe31b49905c3e08d9cfaf96e2%7Cd958f048e43142779c8debfb75e7aa64%7C0%7C1%7C637769175325588286%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=5vevh1cJOszulDkv2LJY%2B6hhQijKW0z8ko4CcONelAs%3Dreserved=0 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Very good software backlash control demo.
I think his method would work for your mill. Yes the tool pushed the table and the table moves. What he does is not just a double loop. He models the movement of the table and predicts how it will move.To do this on a mill you would need encoders on the table and then run a calibration cycle that measure table movement as a function of force. Then as the mill cuts he can update the model parameter in real time by seeing if the predicted table motion matches the actual table motion. In this video he loads a pencil. He says (in comments some place) the reason for the initial slow movement is that the system is measuring flex vs. force applied as it puts the lead in the pencil. after a second or so the data is good enough the system can move faster. You can see the rate of lead insertion gets faster. It is because he has better data. https://youtu.be/GCHXNcpq3OA On Mon, Jan 3, 2022 at 7:02 PM John Dammeyer wrote: > Interesting concept. I notice that although he showed his finger pushing > on the jaws (and text said compensation turned off) we never saw him > disturb the jaws with compensation on. That he can position accurately is > pretty cool. But backlash on my X axis results in the milling cutter > pulling or pushing the table under heavy cuts. > > The only real solution is ball screws or double nut ACME. > > Now for a robot arm that does pick and place of work "on to"/"off of" the > mill or a conveyor belt maybe it's not an issue? > John > > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: January-03-22 6:40 PM > > To: Enhanced Machine Controller (EMC) > > Subject: [Emc-users] Very good software backlash control demo. > > > > CNC is about precision motion control. Here is a new idea where this > > builder gets 0.05mm accuracy but uses hobby level R/C servo Not only > that, > > but he connects three of these in series, one to the next to the next so > > all the backlashes and poor tolerances add together. Then he uses this > to > > do precise motion. He loads a mechanical pencil with this chain of cheap > > parts. > > > > What does this means for LCNC? It means that someone has found a > software > > solution to backlash. What he does is place a quadrature encoder on > both > > the motor and the output shaft.The difference in encoder reading is > an > > exact measure of mechanical backlash and effective gear ratio.He can > > measure the backlash under different conditions and store the > > measurements. Then he places a cascaded PID controller and Kalman filter > > over this hardware. > > > > Technically the problem with backlash control via software is the delay > > from input to output pays poorly with the PID algorithm. He applies a > > predictive model. > > > > Checkup this video. It is unimpressive if you have a $100,000 CNC > milling > > machine, but he is using a linked chain of hobby servos. The novel idea > is > > his software. It is on github, you can read it. > > https://youtu.be/gq-P39rfRqU > > > > He explains it here. Notice in the video he shows the backlash. The > > gear-slop is at the 1/4 inch level but using his software backlash > > correction you can see the results in the dial indicator is about 0.05mm > > (or about 0.002 in American units) Not bad given the truly horrible > > mechanics. > > https://youtu.be/SioCwvR_PYY > > > > Why do I care? I have a robot-dog leg here on my workbench using hobby > > servos, let's say performance could be improved. But the anti-lash > > technique might be used on a real milling machine. Maybe one of the > > experts here could look and see if it could be applied?I will use > parts > > of his idea on my dog-bot. > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Very good software backlash control demo.
CNC is about precision motion control. Here is a new idea where this builder gets 0.05mm accuracy but uses hobby level R/C servo Not only that, but he connects three of these in series, one to the next to the next so all the backlashes and poor tolerances add together. Then he uses this to do precise motion. He loads a mechanical pencil with this chain of cheap parts. What does this means for LCNC? It means that someone has found a software solution to backlash. What he does is place a quadrature encoder on both the motor and the output shaft.The difference in encoder reading is an exact measure of mechanical backlash and effective gear ratio.He can measure the backlash under different conditions and store the measurements. Then he places a cascaded PID controller and Kalman filter over this hardware. Technically the problem with backlash control via software is the delay from input to output pays poorly with the PID algorithm. He applies a predictive model. Checkup this video. It is unimpressive if you have a $100,000 CNC milling machine, but he is using a linked chain of hobby servos. The novel idea is his software. It is on github, you can read it. https://youtu.be/gq-P39rfRqU He explains it here. Notice in the video he shows the backlash. The gear-slop is at the 1/4 inch level but using his software backlash correction you can see the results in the dial indicator is about 0.05mm (or about 0.002 in American units) Not bad given the truly horrible mechanics. https://youtu.be/SioCwvR_PYY Why do I care? I have a robot-dog leg here on my workbench using hobby servos, let's say performance could be improved. But the anti-lash technique might be used on a real milling machine. Maybe one of the experts here could look and see if it could be applied?I will use parts of his idea on my dog-bot. -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
Large ring bearings are not crazy-expensive. I was thinking about using some in a robot leg design. This one could work for your size reduction drive. Use them in pairs spaced as far apart as possible. They are deep groove so they can take some axial load. With wide enough spacing you don't get much play. This company, "uxcell", makes cheap but good enough bearings and sells on Amazon, eBay and other places like that. amazon.com/uxcell-6816-2RS-Bearing-80x100x10mm- <https://www.amazon.com/uxcell-6816-2RS-Bearing-80x100x10mm-Bearings/dp/B07RQ4RXDR/ref=sr_1_3?crid=21LZ75QCVNKGM=uxcell+ball+bearing+100mm=1641234560=industrial=uxcell+bal+earing+100mm%2Cindustrial%2C111=1-3> I generally press-fit bearing onto the plastic and have to always "post process" the plastic until the bearing slips on with slight hand pressure. I have a lathe but I'm lazy and mostly just sandpaper the plastic until the bearing fits. but if you turn the plastic you get better concentric fit. For cheap hardware I now just search for "uxcell" and find screws, nuts and bearings, precision ground rods and what not, all "decent Chinese quality". Also when you use big ring bearing, threat them as structural member in the design. A plastic part is very weak but a plastic part with a big stainless steel ring attached with CA glue is almost as strong as a big stainless steel ring. I discovered "hardware as structure" a few weeks ago as a fix for broken plastic parts. So I re-printed the part but with long internally thread M3 or M4 holes all the way through then I fill the hole with CA glue and put in a long screw. I call it "steel reinforced plastic".Like rebar in concrete. On Mon, Jan 3, 2022 at 12:57 AM John Dammeyer wrote: > For fun I created Todd's version with 0.5module gears. The OD of the > assembly is now 60mm compared to 150mm. It would now be possible to > choose a far east harmonic drive tapered roller assembly to hold the driven > gear in place. Although they are expensive. > > Not sure what type of bearings, if any would be used for the planetary > gear clusters. I suppose one could press in bronze sleeves. I guess it > all depends on the duty cycle and target application. If it's a 6 axis > robot arm used for tool changing and placement/removal of milled parts then > it's not running continuously. Might well for the home or small shop be > more than adequate. > > This example still has 67.333:1 reduction but if the fit is well set > then the backlash is essentially zero which is different from normal > planetary reduction drives. Not sure how important that is for a 6 axis > robot arm. > > In either case, using a 4th axis to create all three spur gears at the > same time (one set of 40T and one set of 41 T) and then some sort of > broaching/indexing to create the two ring gears the real issue is the outer > bearing of the 101T output ring gear. But overall not that difficult to > create once you have the 0.5 module shaper style cutter. > > John > > > _______ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Tormach writing open sorce ROS2/Machiekit/LCNC bridge.
Has anyone else noticed the activity on Tormach's Git Hub repository. There seem to be multiple commits each day to https://github.com/tormach/hal_ros_control/tree/foxy-devel In a nutshell, they are making HAL a ROS2 node.From a LNCN user's view point this means machine tool can become part of a larger automated robotic system and from the robotic engineer's view this means he can use the HAL infrastructure to control robot parts. Interesting because it is Tormach not some hobby user working part-time. It is in Tormach's github. I can imagine a work cell where robots load and unload milling machines or move parts between machines and then pack the finished parts in boxes. Then palletize the boxes and load them onto self-driving trucks that carry them to a "lights out" factory where they make robots. We all become redundant. More seriously, this about doubles what each group can do. And Tormach seems to be investing real money in this. -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Choice of CNC conversions
No, RT patches are not a waste of time. But using something as big as a Linux based PC to drive a machine is not going to be popular today and will be even less so in 10 years. 40 years ago the PC is the only affordable computer that had enough compute power. This is not longer true. we now have 32-bit microcontrollers and FPGAs. al for under $20. Look at a typical hobby R/C servo. The cheapest one on Amazon is $5 (The MG996). Inside there is a digital motor controller that measures the input pulse with in microseconds and reads an analog voltage from the position encoder. It uses a PID control loop to adjust the motor's current via an H-bridge to keep the output shaft at the commanded set point. The entire device, computer. motor, gears and all, sells for $5 on Amazon. That $5 servo Is reliably measuring at the sub uSec level. Something a PC can't do. In the 80's these servos were analog because you would need something like an IBM PC to implement a PID loop. Nothing wrong with using LCNC, I use it. But we were talking about why a person starting out would pass on it. It is just not what people would do today. Today, I buy something, I expect to turn it on and then maybe download the phone app that let's me control it.I don't expect to edit text files to make by video doorbell work I'm not suggesting anyone do anything, just explaining why LCNC is not taking over the world. People will go for the $300 box that is plug and play 80% of the time. People want support and a warenty. On Fri, Dec 24, 2021 at 11:07 AM Sam Sokolik wrote: > Hmm.. so the realtime patches being integrated into Linux kernel is > worthless old technology? They are waisting their time? (This is > unrelated to LinuxCNC) > > Currently LinuxCNC is being added to debian testing. It will soon be part > of the debian repository. > > On Fri, Dec 24, 2021, 12:54 PM Chris Albertson > wrote: > > > The problem that is limiting LCNC's wider use is that it is a very old > > design. It is definitely not what anyone would design today. And it is > > not what moderned users expect or want. Documentation i=will not change > > what it is. > > > > Today, if this were being built again from scratch it would > > > > Run on any computer and not require some special real-time version of > > Linux. The user interface would be written in some portable way so it > > could be accessed even on a iPad or Andriod tablet or from macOS or > > Windows. This is possible. I proved it to myself just a few days > > ago. I have a 12 DOF robot here that is being driven by a Raspberry Pi > > and the user interface is web-based or X11 based and in theory, should > work > > on other platforms. > > > > Should not need a real-time OS on the computer. The real-time stuff (al > > of it) goes in hardware, Leaving only not-t=real-time tasks to the > > PC/Mac/iPhone > > > > It would configure 100% with no need to edit a single file by hand. > > > > It would have a conversational system so that a user could do simple > things > > with no need to write g-code. > > > > People care less about if it is free then if it acts like the above. > > > > What I would do is design some kind of real-time module. Perhaps that > > would be made of Mesa cards with different firmware or of > microcontrollers > > like "Teensy" and each of these could handle some number of axies. Maybe > > four. Then you use multiple of these to drive a larger machine of a > > robot. > > > > The 1980's was 40 years ago. Yes it really has been that long. LCNC is > > using 1980s software technology and people today are expecting the 21st > > century and mostly getting what they expect. Think of a basic 3D > > printer. It is no different from a milling machine just mechanically > > lighter weight. The whole thing, g_code interpreter and all is a cheap > > package with a self-contained controller One does not need to hunt > > dumpsters for antique desktop PCs and then install specialist OSes on > > them. The controller is built-in and pre-programmed. > > > > That said. I use LCNC because it does what I want and uses the 40 years > > old (maybe 50 years old now) technology I'm familiar with. Yes it is > that > > Old. I was a computer science student back when this wascutting edge and > > I'm retired now. > > > > Big goals for any new system should be > > 1) cross platform, especially mobile device friendly > > 2) zero file editing (zero, not just a small number) > > 3) modular, you can swap out parts and add parts as requirements change. > > 4) today we have "The Cloud" It could live at Amazon or
Re: [Emc-users] Choice of CNC conversions
e would cover all the usual suspects of DIY CNC projects like router, > laser, plasma, mill, lathe builds/conversion/retrofits. > These docs may provide infos for the size sensitive components for the > reader to adapt. > > Another (complementary) approach would be to add to the docs a library of > well crafted howtos about the setup of the various subsystems like motion, > spindle/torch, coolant, limits, e-stop... Some kind of decision tree could > be provided to lead the implementor who'd find the appropriate support doc > for each choice he'd make. > > For those of us that like to tinker with machines anyway, LinuxCNC is > already just great... For the rest of the world that'd better have a > machine that they can use to make stuff, we need to provide setups that > "just work" in a way or another. > > Willing to work on this after docs migration and french translation is > done. I'm all in to bring LinuxCNC goodness to the masses without putting > the burden on the devs who I'd rather have working on fancier things for > the future... > > As I may be unable to build an actual machine for the foreseable future. I > plan to start building docs using vismach, then move to tabletop/lab setups > with small real components, etc... > > All these reference docs could have a category and a dedicated thread in > the forum to hold discussions, requests and criticisms. > > TY > Jérémie > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
On Wed, Dec 22, 2021 at 1:27 AM John Dammeyer wrote: > > > I'm going to try and reprint the cup with random layer changes to try and > get rid of the bump. Should be able to mount it into the existing box and > lid and try it out. > > > Are you looking for things like this before you print or waiting to try yhr parts. Cura, of course will display a preview showing how the plastic is laid down. I always look layer by layer at this before deciding to save the g-code. Then be sure to "export the project" and save the setting use so you can do it again. But I find if a RELLY want to see how the print will be made before it is printed I look at the layer by larger plots using "octoprint". Octoprint will analyze and plot the g-code Finally I never expect 3D prints to be perfect. With my timing pulleys I'd put them in the lath and us a borring bar to make the holes clean perfectly round before glusing them to steel hubs. then mount eh pully in the lath and do a clean-up of the eterior. I would nt suggest doing this on oter 3D printing lists but here I assume overone own some machine tools. So put the part on a lathe. Same for screw holes and such, I ream them to corect size r install threaded brass inserts. -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
On Mon, Dec 20, 2021 at 5:35 PM gene heskett wrote: > On Monday, December 20, 2021 7:03:21 PM EST John Dammeyer wrote: > > Perhaps save it as a step file or iges. I can't read that scad file. > > I don't think it can output those non-free formats. > STEP is an ISO standard and IGES is a U.S. National Bureau of Standards format. Of the two, STEP is more widely universal as "everyone" can read STEP files. Also, as said, FreeCAD can import and run SCAD and then can do things like add fillets and report or even, generate code for milling. I think SCAD is great for making parts that can be algorithmically defined. But I'd hate to have to make an organic shape like a automotive body or a power tool. o it's good to be able to move designs between tools. STEP is the best for that. Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
I got the idea to post process the gears with heat when I was experimenting with brass inserts. They make this tool for installing inserts. One end is turned to fit the insert the other is turned to exactly fit my temperature control solder iron. I set the iron for 180C and the insert is pretty much exactly at 180C after 5 seconds on the tool.I thought I could make a brass gear shaft that fits my Hakko solder station. Even if I used the gadget as a hand tool I could improve a gear like in your photo. The Hakko solder iron is fairly powerful it reacts to cooling really fast and controls the tip temperature well. But I don't know how large of a brass gear it could heat Cover the gear with that ceramic insulation and maybe a big gear. I covered the bottom side of my 3D printer's build plate with a larger of ceramic insulation and thin foil tape. It now heats up in seconds because it is not radiating heat from the bottom. This is an easy upgrade. I'm thinking different from many here because when I want a part I want dozens of them, not just one. Each little walking robot has 12 motors and needs 12 reduction systems and 12 encoders I bailed on the reduction design as I just can make it small enough and used R/C hobby servos. As you can see there is room for improvement. But then this leg cost less than $50 complete as shown, drive motors and all. https://youtu.be/I93HApHSB0Q I need at least 8 of these legs. So I start with this and in the background try to find a way to use BLDC motors and reduction drives.I really need some reducers that are about 50mm overall diameter at most and at least about 50 Kg-cm torque. I think I'm looking at steel parts. On Mon, Dec 20, 2021 at 2:15 PM Matthew Herd wrote: > So I imagine it is possible to post-process a thermal plastic gear, but at > that point, maybe it’d be better to set up a gear skiving machine to make > the flex rings. If you can get the cutter and angle a rotary axis with > respect to the spindle, you should be able to do it without much hassle. > Has anyone considered that route? Obviously it requires a rotary axis to > make a rotary axis, but it really doesn’t seem any more insurmountable and > the results would probably be much more precise. > > Matt > > > On Dec 20, 2021, at 4:56 PM, Chris Albertson > wrote: > > > > Is it possible to post process a thermal plastic gear? Lets say you > took > > a high-precision metal gear and heated it to 180C and the rolled it over > > the printed gear with the correct center to center distance. You would > > need to build a test fixture to do this but might be worth it. > > > > lately I've been experimenting with brass thread inserts. I have 1/2 > > dozen different types and printed test blocks with different hole > diameters > > and I've tried the soldering iron at different temperatures. The best > > results are really good with the M3 size screw failing before the nut. > > The worst case is they just pull out easily with pliers. > > > > The hard part seems to be repeatability and if the hole is parallel or at > > right angles to the layers. Printer setting and part design seems to > matter > > a lot also.I've got a walking-dog type robot and I need to convert it > > all over to threaded inserts, about 80 places. I find it helps to think > if > > each holes gets larger or smaller then design when it is printed. and > this > > depends on ho the hole is connected to the rest of the part. By walls or > > sheets or infill.. Engineering is fun... > > > > > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
Is it possible to post process a thermal plastic gear? Lets say you took a high-precision metal gear and heated it to 180C and the rolled it over the printed gear with the correct center to center distance. You would need to build a test fixture to do this but might be worth it. lately I've been experimenting with brass thread inserts. I have 1/2 dozen different types and printed test blocks with different hole diameters and I've tried the soldering iron at different temperatures. The best results are really good with the M3 size screw failing before the nut. The worst case is they just pull out easily with pliers. The hard part seems to be repeatability and if the hole is parallel or at right angles to the layers. Printer setting and part design seems to matter a lot also.I've got a walking-dog type robot and I need to convert it all over to threaded inserts, about 80 places. I find it helps to think if each holes gets larger or smaller then design when it is printed. and this depends on ho the hole is connected to the rest of the part. By walls or sheets or infill.. Engineering is fun... On Mon, Dec 20, 2021 at 1:05 PM John Dammeyer wrote: > As an experiment, now that I've modified the AlibreCAD Python script to > use Module or Pitch Diameter and learned how to cut internal spline teeth I > created two parts with double the original harmonic drive: 140T and 142T > with 0.5module teeth. > > The cup ID dropped to 65mm from 80mm for the 1.25module 70T. I've printed > it out of PLA again since the yellow I'm using is easier to look at for > details compared to the shiny PETG black. > > Printing the cup went well with clean teeth. The ring gear is still > finishing and I'm not impressed. For whatever reason it's not looking good. > > As for profiles, once you go above 135T the standard gear cutter wheels > are marked 135 to infinity where infinity means an infinite pitch diameter > also known as a rack profile. So if one wants to use standard gear cutter > tooling the number of teeth in the cup/gear assembly must be over 135. > > To test that theory I recreated the original 70/72T design using AlibreCAD > which then generated the involute tooth profile which we've been told isn't > as effective as pure triangle or slight sine curve on the tooth slope. > > Now where before the cup/gear combination would bottom out before making > full side contact this one is really tight. Maybe too tight. But mix the > two with the cup made from the original STEP file profile and the AlibreCAD > ring and the fit is perfect. More testing on that when I find some more > 3mm screws of the proper length. > > Meanwhile once the 3D print of the 0.5 module ring gear is complete I'll > see if it's possible to clean up the teeth. The thing is that the teeth > are smaller. So the cup doesn't have to flex nearly as much as it does > with the 1.25module cup. Perhaps if the 1.25module cup also had 140T and > the larger diameter it also wouldn't flex as sharply. And therefore less > chance of cracking. > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: December-20-21 10:22 AM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Harmonic Drive > > > > On Fri, Dec 17, 2021 at 6:05 PM Sam Sokolik wrote: > > > > > Hmm - seems we are making things a bit more complicated than they need > to > > > be.. Why didn't the simple flex gear work ok? You need to make the > > > system with fine enough splines that the flexure is at a minimum... > You > > > can lower the backlash to a minimum by applying a small amount of > > > pressure to the system. > > > > > > > The problem with small splines is that he is making this with printed > > plastic. There is a minimum feature size that works. Seriously you > want > > to stay above module 0.5 and bigger if you can. Mod 1.0 is more > > realistic There are two reasons. (1) plastic is not very strong unless > > you make the parts big and (2) the printer makes parts with dimensional > > tolerances of about 0.4 mm so if your parts have 1mm features you can > > expect 50% dimensional errors. That said, modulo 1.0 gears work really > > well. Smaler ones have a short life. > > > > Finally ANY flexing at all is the death of printed parts. They fail by > the > > failure of interlayer adhesion. Think of printed plastic like it is soft > > pine wood. When making gears from yellow pine, the direction of the > grain > > matters a lot and no one would think of making module 0.5 gears with > pine. > > > > I've been able to build an entire CNC conversion kit for a mini-mill from >
Re: [Emc-users] Harmonic Drive
On Mon, Dec 20, 2021 at 11:23 AM Andrew wrote: > > > Has anyone tried DLP/SLA printed flex splines? > Thinking more along those lines, There are printers that can print metal parts. They are expensive but if you only need one or two parts Shapeways can print them for you. They offer to print in 316L stainless steel. There are other even more exotic printers you can buy time on. SpaceX prints the fuel injectors for their rocket engines. The printer costs literally a million dollars, but you only need one part made so no need to buy the printer. > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
Yes, he used maybe 30 pounds of metal parts. And I only saw a few degrees of rotation in the video, run it for 100 hours and see what happens. Flex is what kills the plastic so after 100 hours it will have had a few million cycles. But still it is 99% made of metal. I want to see a 100% plastic unit stand up to 100 hours of use. It can be done but not if the plastic parts flex. On Mon, Dec 20, 2021 at 10:39 AM Sam Sokolik wrote: > Did you watch this video til the end? I time stamped it. > > https://youtu.be/eW1GGI55Epc?t=878 > > sam > > > > On Mon, Dec 20, 2021 at 12:26 PM Chris Albertson < > albertson.ch...@gmail.com> > wrote: > > > On Fri, Dec 17, 2021 at 6:05 PM Sam Sokolik wrote: > > > > > Hmm - seems we are making things a bit more complicated than they need > to > > > be.. Why didn't the simple flex gear work ok? You need to make the > > > system with fine enough splines that the flexure is at a minimum... > > You > > > can lower the backlash to a minimum by applying a small amount of > > > pressure to the system. > > > > > > > The problem with small splines is that he is making this with printed > > plastic. There is a minimum feature size that works. Seriously you > want > > to stay above module 0.5 and bigger if you can. Mod 1.0 is more > > realistic There are two reasons. (1) plastic is not very strong unless > > you make the parts big and (2) the printer makes parts with dimensional > > tolerances of about 0.4 mm so if your parts have 1mm features you can > > expect 50% dimensional errors. That said, modulo 1.0 gears work really > > well. Smaler ones have a short life. > > > > Finally ANY flexing at all is the death of printed parts. They fail by > the > > failure of interlayer adhesion. Think of printed plastic like it is soft > > pine wood. When making gears from yellow pine, the direction of the > grain > > matters a lot and no one would think of making module 0.5 gears with > pine. > > > > I've been able to build an entire CNC conversion kit for a mini-mill from > > printed plastic. The stuff is VERY rigid and strong if you make the > parts > > large enough > > > > > > > > > > > > > > > > > > > > sam > > > > > > On Fri, Dec 17, 2021 at 7:46 PM John Dammeyer > > > wrote: > > > > > > > Alright. So it spins really freely but keeping the planetary cluster > > > > centred is a bit of an issue. So the attached photo shows roughly > > (very > > > > roughly) what I had in mind. > > > > > > > > With the right coupler between the motor and the gear cluster the > > > > planetary assembly should stay in the same place. There really isn't > > any > > > > axial load. There is a place for a 40mmx19mm bearing that I happen > to > > > have > > > > on hand. > > > > > > > > But what about the driven ring gear? In one of the model shops > today I > > > > saw a package of 100 5.5mm stainless steel balls they sell for > putting > > > into > > > > paint bottles to help mix them up. > > > > > > > > Seemed and ideal size for a bearing race between the two ring gears. > > > Kind > > > > of like what Gene H. stated he did on his setup. (Pictures?) So the > > > Blue > > > > Gear is fixed. The Green Rotates. The lazy susan like bearing > between > > > > handles axial pressure in the direction of the blue gear. But really > > the > > > > only thing holding up the green ring gear is the planetary assembly. > > > > > > > > Say we wanted to mount a face plate or chuck onto the green gear. > I'm > > > > open to ideas on how to stabilize that. Sketches would be great. > > > > > > > > Methinks a part that surrounds the blue gear and extends over the > green > > > > and they also have a bearing race between. Or we make a similar > > bearing > > > > mount to the rear one for the front with the gear cluster holding > this > > in > > > > place extended out to the bearing. Now the faceplate is attached to > > > this. > > > > But a lot more axial twist on the small bearing in the middle. > > > > > > > > Thanks > > > > John > > > > > > > > ___ > > > > Emc-users mailing list > > > > Emc-users@lists.sourceforge.net >
Re: [Emc-users] Harmonic Drive
On Fri, Dec 17, 2021 at 6:05 PM Sam Sokolik wrote: > Hmm - seems we are making things a bit more complicated than they need to > be.. Why didn't the simple flex gear work ok? You need to make the > system with fine enough splines that the flexure is at a minimum...You > can lower the backlash to a minimum by applying a small amount of > pressure to the system. > The problem with small splines is that he is making this with printed plastic. There is a minimum feature size that works. Seriously you want to stay above module 0.5 and bigger if you can. Mod 1.0 is more realistic There are two reasons. (1) plastic is not very strong unless you make the parts big and (2) the printer makes parts with dimensional tolerances of about 0.4 mm so if your parts have 1mm features you can expect 50% dimensional errors. That said, modulo 1.0 gears work really well. Smaler ones have a short life. Finally ANY flexing at all is the death of printed parts. They fail by the failure of interlayer adhesion. Think of printed plastic like it is soft pine wood. When making gears from yellow pine, the direction of the grain matters a lot and no one would think of making module 0.5 gears with pine. I've been able to build an entire CNC conversion kit for a mini-mill from printed plastic. The stuff is VERY rigid and strong if you make the parts large enough > > sam > > On Fri, Dec 17, 2021 at 7:46 PM John Dammeyer > wrote: > > > Alright. So it spins really freely but keeping the planetary cluster > > centred is a bit of an issue. So the attached photo shows roughly (very > > roughly) what I had in mind. > > > > With the right coupler between the motor and the gear cluster the > > planetary assembly should stay in the same place. There really isn't any > > axial load. There is a place for a 40mmx19mm bearing that I happen to > have > > on hand. > > > > But what about the driven ring gear? In one of the model shops today I > > saw a package of 100 5.5mm stainless steel balls they sell for putting > into > > paint bottles to help mix them up. > > > > Seemed and ideal size for a bearing race between the two ring gears. > Kind > > of like what Gene H. stated he did on his setup. (Pictures?) So the > Blue > > Gear is fixed. The Green Rotates. The lazy susan like bearing between > > handles axial pressure in the direction of the blue gear. But really the > > only thing holding up the green ring gear is the planetary assembly. > > > > Say we wanted to mount a face plate or chuck onto the green gear. I'm > > open to ideas on how to stabilize that. Sketches would be great. > > > > Methinks a part that surrounds the blue gear and extends over the green > > and they also have a bearing race between. Or we make a similar bearing > > mount to the rear one for the front with the gear cluster holding this in > > place extended out to the bearing. Now the faceplate is attached to > this. > > But a lot more axial twist on the small bearing in the middle. > > > > Thanks > > John > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
> > > single gear double tall. But all of the others would have different > shifts. For a 3 planet the shift would be +1/3 tooth for one set > > and > > > -1/3 for the other set. This set up has exactly 1 tooth per spider > rev reduction. > > > > > > I dropped the identical ring size version when I discovered that > making the pinions 1 tooth different made up for the 1tooth > > > difference in pitch diameter of the rings, because I didn't like how > the fudged module rings were meshing with the pinions, and I > > > couldn't change the tooth profile to make it work better with the > software I was using. > > > > > > Todd Zuercher > > > P. Graham Dunn Inc. > > > 630 Henry Street > > > Dalton, Ohio 44618 > > > Phone: (330)828-2105ext. 2031 > > > > > > -Original Message- > > > From: andy pugh > > > Sent: Friday, December 17, 2021 10:43 AM > > > To: Enhanced Machine Controller (EMC) > > > > Subject: Re: [Emc-users] Harmonic Drive > > > > > > [EXTERNAL EMAIL] Be sure links are safe. > > > > > > On Fri, 17 Dec 2021 at 15:27, Todd Zuercher > wrote: > > > > > > > > Andy, Initially I thought what you are saying about the pitch > diameters needing to be the same was true. And that would be the > > > case if all the pinions had the same tooth count and size. But When I > was modeling it I found that if all of the module was the same > > > for all the gears, and adding one tooth to the output set, and the > output pinions, their pitch diameters still meshed perfectly. > > > > > > I think that whilst that will work in theory, getting it assembled > right would be a problem. > > > > > > With two ring gears with the same PCD but +1 teeth on one gear you can > have pinion pairs with equal teeth and a fixed phase > > offset. > > > These can be assembled in any rotational orientation (as long as they > are in the right place relative to each other) > > > > > > Also, with such an arrangement you can drive the pinions with a small > sun gear, for a second level of reduction. You just have to > > > make sure that the sun gear toothcount is divisible by the number of > plantets, and also that you drive the unshifted half of each > > > pinion (I think) > > > > > > > > > -- > > > atp > > > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils > > > and lunatics." > > > - George Fitch, Atlanta Constitution Newspaper, 1912 > > > > > > > > > ___ > > > Emc-users mailing list > > > Emc-users@lists.sourceforge.net > > > > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc- > > > > > usersdata=04%7C01%7Ctoddz%40pgrahamdunn.com > %7C4842da1f9f934e91441e08d9c1741eef%7C5758544c573f47cebee96 > > > > > > c3e0806fb43%7C0%7C0%7C637753527048039153%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJB > > > > > > TiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=jxrNeQMSLeIFw8UIaEbmsfbb61M1abPjoVKFmwCIRP8%3Dreserved=0 > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] making post square.
On Sun, Dec 12, 2021 at 12:39 PM Gene Heskett wrote: > On Sun, 12 Dec, 2021 at 2:54 PM, Chris Albertson < > albertson.ch...@gmail.com> wrote: > > > To: enhanced machine controller (emc) > I'm surprised that people don't use kinematics to remove the residual > alignment errors that can't be mechanically removed. Of course, you can't > remove tool orientation errors this way. If the z-axis is a few > milliradians "off" the tool will also be "off" but kinematics can > compensate for the X, Y error associated with variable Z movement. > > That said, I can't measure this kind of low-level stuff so I don't worry > about it. > > You should be able to lead the Z- column over at 45 degrees and still hit > X, Y point spot on. (it would work only for a spherical tool) > > > > If it was moveable Chris, but its not, the column is fixed, but the head > can be No, you never want to do that. My point was that there could be an absurd amount of error and you should still be able to work with it by telling the kinematics system the column meets the Y axis at 89.9 degrees. I'd think we are talking arc minutes at most here for a real milling machine. But there are home-made CNC machine made with rails mounted to a plywood base. BTW, even with the 0.1-degree error, you still have good accuracy with realistic Z-axis moves. > > rotated 90 degrees either way and I have several times. That works quite > well. > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] making post square.
I'm surprised that people don't use kinematics to remove the residual alignment errors that can't be mechanically removed. Of course, you can't remove tool orientation errors this way. If the z-axis is a few milliradians "off" the tool will also be "off" but kinematics can compensate for the X, Y error associated with variable Z movement. That said, I can't measure this kind of low-level stuff so I don't worry about it. You should be able to lead the Z- column over at 45 degrees and still hit X, Y point spot on. (it would work only for a spherical tool) On Sun, Dec 12, 2021 at 9:47 AM Gene Heskett wrote: > On Sun, 12 Dec, 2021 at 12:13 PM, Chris Albertson < > albertson.ch...@gmail.com> wrote: > > > To: enhanced machine controller (emc) > Would "non-trivial kinematics" correct for non-square mechanical axis? It > seems to me this is the way to correct the residual errors that all > real-world machines have. > > > > IDK, Chris, no one has mentioned it so its not been investigated. I will > look it up. > > Thanks, Chris. > > > Cheers, Gene > > -- > > Chris Albertson > Redondo Beach, California > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net<mailto:Emc-users@lists.sourceforge.net> > https://lists.sourceforge.net/lists/listinfo/emc-users > . > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] making post square.
Would "non-trivial kinematics" correct for non-square mechanical axis? It seems to me this is the way to correct the residual errors that all real-world machines have. On Sun, Dec 12, 2021 at 8:40 AM Gene Heskett wrote: > On Sun, 12 Dec, 2021 at 6:09 AM, me wrote: > > > To: emc-users@lists.sourceforge.net > Greetings all; > > > I need access to the Z position that is G53 referenced as I need the > absolute > position on the post assuming the home switch abut 40 thou below the top > is at 0.0. > > > Which hal pin or signal gives me that value? > > No answer so far, so let me ask a different way since I cannot watch the > #5XXX > > values with a hal meter to identify the var I need. > > > Is joint.N.cmd-pos the stable value I need? > > > Not changing with co-ord mapping or touch-offs, or even metric/imperial > measurement switches. > > > I am currently using it for Z pos reference as the pos input to a lincurve > module > whose output is directed to the X offset module "wired" up in hal to comp > bed > wear on the Sheldon, and that seems to be working well as it needs about 13 > thou of comp right in front of the chuck. It can now turn within a thou, > for at > > least 15" from the chuck. > > > > My GO704's post is not quite plumb, the boltholes that attach it to the > base aren't > big enough to allow it to be pulled plumb, so I either need to ream the > holes. or > > fix it with an offset in the X path, preferably based on a zero point at > the home > switches closure point. By inserting a tiny bit of Z motion in order to > not bend > a drill bit against the side of the hole as it descends. This bent drill > is quite > > visible. > > > > Ideal situation would having this joint.N.pos,cmd be 0.0 at the home > > switch closure. > > > Then I want, not to use a lincurve module, but a scale module with a very > > small scale setp'd into it to develop a correction offset that might be > only > 50 thou in the height of the head travel. The head is currently well > trammed > to the table and a small X movement as Z descends could correct it, > drilling > and tapping round holes by moving X to make it stay plumbed as Z descends. > > > I have even broken taps with the non plumb sideways push at the bottom > of a hole. And obviously sloppy threads result. > > > > One more step IOW in my attempts to make a silk purse out of an $1100 sow's > ear mill. > > > > Thank you. > > Cheers, Gene > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net<mailto:Emc-users@lists.sourceforge.net> > https://lists.sourceforge.net/lists/listinfo/emc-users > . > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
Yes, 16 mm bearing can't help. I think what would be best is to use two bearing and make them half the diameter of the inside of the cup. That would be the largest possible size that would still fit. That said, why not make a large size bearing from a small one by slipping a plastic disk over you 16mm bearing to make it's effective u Once I thought of slipping a disk over a bearing to make the bearing half the diameter of the cup. Then I think "Why can't the diameter of the plastic bearing overlap. Cut groves in each disk? (really now it is a plastic "tire" mounted on a bearing) You might get a "tire" diameter that is 60% of the cup diameter and still use smaller skateboard bearings On Wed, Dec 1, 2021 at 9:37 PM John Dammeyer wrote: > Broke the first cup. It's possible that a few times I didn't insert it > right because the holes didn't line up well so I might well have begun the > crack which then escalated. If it's inserted off axis it's hard to get it > into position and if it even turns is very rough. Get it right and it's > quite smooth. > > There are several issues. First of all a tiny vertical piece of the root > of the tooth broke off. I think that happened first. Then the pressure of > the bearings sheared the cup along one of the print layers. > > So first weak spot of course the print layers and the plastic probably > never really bonding past a thin line. I used 0.2mm printing. > > My smaller 16mm bearing is also a much higher load point than the larger > ones used in the belt versions. Even if I'd made the full disk with 10 > bearings I don't expect it would have made much of a difference because > they just can't get as close together as a real harmonic drive with ball > bearings which are also distributing the load over the entire outer race. > If anything a larger group of small bearings as close as possible to each > other would more accurately duplicate this. > > www.autoartisans.com/harmonicdrive/HarmonicDrive-11.jpg > > This type of flex cup if not made of metal would be better with a cross > hatch carbon fibre in flexible epoxy. > > John > > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
The Step file I sent is fully functional when imported to Fusion360. There are a few changes I would make too. If the problem is the cup flexing, Maybe try some accordion pleats, a few broad sine-like curves in the cup. Why are the wals cylindrical? Another idea is to remove the bottom from the cup so it is just a tube and not a cup then cut a square wave patters on the bottom and drive ot with a disk with an interlocking pattern. Then there is no more bending on the booton radius. On Wed, Dec 1, 2021 at 3:38 PM John Dammeyer wrote: > Chris sent me a zip with the project step file. This one I was able to > load and save. The individual part files were not editable. > > However, once I saved that imported step file as the entire assembly > AlibreCAD saved it as a series of assemblies and part files; 88 in all > including the original step file. The assembly does not have any > constraints so once you pull the pieces apart it's hard to put them back > without adding in constraints. > > Anyway, these are editable. > If anyone wants a zip with the .AD_ASM and .AD_PRT files I'll pass it on. > At this point you load Harmonic Drive Project v45.AD_ASM and can edit the > individual part files although they can't be saved from within the part > editing. Only inside the assembly. > > But separately they can be edited. Probably a bug in Alibre. They don't > call it Read Only or Read/Write but locked and unlocked. > > John > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: December-01-21 2:59 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Harmonic Drive > > > > If it is too much work, ask me and I will re-export the entire model from > > Fusion360 to STEP, and I'm sure there will be no read-only as Fusion360 > > does not do that. > > > > I also want to re-organize it, like at least placing all the nuts and > > screws in a "hardware" component so that can all be turned off and on at > > once rather than doing 100+ clicks > > > > On Wed, Dec 1, 2021 at 2:53 PM John Dammeyer > wrote: > > > > > How about. Now it works. I can load the assembly but the embedded > read > > > only characteristic is just plain rude. So the work around with > Alibre is > > > as follows: > > > > > > Import the .step file. > > > Select one of the elements like "Housing" and right click to open in > > > separate window. > > > Doesn't matter what you do to the drawing you can't change it but it > can > > > be exported as an .IGES file. > > > So now I have "Housing.igs" and can import that and then save as the > > > "Housing.AD_PRT" > > > > > > It's still a clumsy step file type of import that doesn't have the same > > > ease of use editing that parts built with Alibre have but at least it's > > > editable and some of the very poor features of the design can be > changed. > > > > > > John > > > > > > > > > > -Original Message- > > > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > > > Sent: December-01-21 1:18 PM > > > > To: Enhanced Machine Controller (EMC) > > > > Subject: Re: [Emc-users] Harmonic Drive > > > > > > > > Thanks for the link. But I noticed there was a STEP file provided. > > > I've > > > > imported it into Fusion360. It imports well. No need to convert > the STL > > > > files. > > > > > > > > Here is a view-only public link to the Fusion360 model. > > > > https://a360.co/3phrfI6 > > > > I've made no edits to the above. simply did the import. > > > > > > > > The model could use some reorganization. It is just a linear list of > > > parts > > > > with no hierarchy but unlike STL files the exact shape of the parts > is > > > > preserved. > > > > > > > > Thanks again. > > > > > > > > Chris > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Wed, Dec 1, 2021 at 9:33 AM John Dammeyer > > > > wrote: > > > > > > > > > Chris, > > > > > > > > > > This was my starting point. The step files aren't available in > that > > > > > link. The STLs have been converted although some badly. Still > > > worki
Re: [Emc-users] Harmonic Drive
If it is too much work, ask me and I will re-export the entire model from Fusion360 to STEP, and I'm sure there will be no read-only as Fusion360 does not do that. I also want to re-organize it, like at least placing all the nuts and screws in a "hardware" component so that can all be turned off and on at once rather than doing 100+ clicks On Wed, Dec 1, 2021 at 2:53 PM John Dammeyer wrote: > How about. Now it works. I can load the assembly but the embedded read > only characteristic is just plain rude. So the work around with Alibre is > as follows: > > Import the .step file. > Select one of the elements like "Housing" and right click to open in > separate window. > Doesn't matter what you do to the drawing you can't change it but it can > be exported as an .IGES file. > So now I have "Housing.igs" and can import that and then save as the > "Housing.AD_PRT" > > It's still a clumsy step file type of import that doesn't have the same > ease of use editing that parts built with Alibre have but at least it's > editable and some of the very poor features of the design can be changed. > > John > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: December-01-21 1:18 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Harmonic Drive > > > > Thanks for the link. But I noticed there was a STEP file provided. > I've > > imported it into Fusion360. It imports well. No need to convert the STL > > files. > > > > Here is a view-only public link to the Fusion360 model. > > https://a360.co/3phrfI6 > > I've made no edits to the above. simply did the import. > > > > The model could use some reorganization. It is just a linear list of > parts > > with no hierarchy but unlike STL files the exact shape of the parts is > > preserved. > > > > Thanks again. > > > > Chris > > > > > > > > > > > > > > > > > > > > > > On Wed, Dec 1, 2021 at 9:33 AM John Dammeyer > wrote: > > > > > Chris, > > > > > > This was my starting point. The step files aren't available in that > > > link. The STLs have been converted although some badly. Still > working on > > > getting better step files from STL. > > > > > > > > > > https://howtomechatronics.com/how-it-works/what-is-strain-wave-gear-harmonic-drive-a-perfect-gear-set-for-robotics- > > applications/ > > > > > > I don't have the larger 47x35 bearings yet so I subbed in a 47x20. > > > Without two bearings there will always be a bit of wiggle there. I > found > > > only two of the 5x16x5 in my random bearing box. I have 10 of those on > > > order too. For now I recreated the wave generator to just use two > > > bearings. > > > > > > I have a tube of 11mm bearings that when doubled up (or maybe not) can > be > > > put closer together creating less of a point source bend that the > single > > > bearing does. That might fatigue the plastic less. > > > > > > Also turning the hub that goes through the bearing out of metal and > making > > > the companion mounting plate for the cup out of metal would allow a > coupler > > > shaft through from the wave generator disk to the centerline of the > drive > > > bearing. This would likely also clean up any wobble from thin printed > > > plastic. > > > > > > John > > > > > > > > > > > > > -Original Message- > > > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > > > Sent: December-01-21 9:13 AM > > > > To: Enhanced Machine Controller (EMC) > > > > Subject: Re: [Emc-users] Harmonic Drive > > > > > > > > John, > > > > > > > > Looks like I'm finding this thread late. Where are the design files > for > > > > this harmonic drive? > > > > > > > > On Tue, Nov 30, 2021 at 11:22 PM John Dammeyer < > jo...@autoartisans.com> > > > > wrote: > > > > > > > > > I've added a photo. > > > > > www.autoartisans.com/harmonicdrive/HarmonicDrive-8.jpg > > > > > > > > > > The new wave generator pieces puts the bearings 1mm further out. > Was > > > also > > > > > able to bring it a bit closer to the motor so the bearings now sit > in > > > the > > > > > middle of the fixed spline gear. > >
Re: [Emc-users] Harmonic Drive
Thanks for the link. But I noticed there was a STEP file provided. I've imported it into Fusion360. It imports well. No need to convert the STL files. Here is a view-only public link to the Fusion360 model. https://a360.co/3phrfI6 I've made no edits to the above. simply did the import. The model could use some reorganization. It is just a linear list of parts with no hierarchy but unlike STL files the exact shape of the parts is preserved. Thanks again. Chris On Wed, Dec 1, 2021 at 9:33 AM John Dammeyer wrote: > Chris, > > This was my starting point. The step files aren't available in that > link. The STLs have been converted although some badly. Still working on > getting better step files from STL. > > > https://howtomechatronics.com/how-it-works/what-is-strain-wave-gear-harmonic-drive-a-perfect-gear-set-for-robotics-applications/ > > I don't have the larger 47x35 bearings yet so I subbed in a 47x20. > Without two bearings there will always be a bit of wiggle there. I found > only two of the 5x16x5 in my random bearing box. I have 10 of those on > order too. For now I recreated the wave generator to just use two > bearings. > > I have a tube of 11mm bearings that when doubled up (or maybe not) can be > put closer together creating less of a point source bend that the single > bearing does. That might fatigue the plastic less. > > Also turning the hub that goes through the bearing out of metal and making > the companion mounting plate for the cup out of metal would allow a coupler > shaft through from the wave generator disk to the centerline of the drive > bearing. This would likely also clean up any wobble from thin printed > plastic. > > John > > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: December-01-21 9:13 AM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Harmonic Drive > > > > John, > > > > Looks like I'm finding this thread late. Where are the design files for > > this harmonic drive? > > > > On Tue, Nov 30, 2021 at 11:22 PM John Dammeyer > > wrote: > > > > > I've added a photo. > > > www.autoartisans.com/harmonicdrive/HarmonicDrive-8.jpg > > > > > > The new wave generator pieces puts the bearings 1mm further out. Was > also > > > able to bring it a bit closer to the motor so the bearings now sit in > the > > > middle of the fixed spline gear. > > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] to JD
The best stuff is PTFE based, "Superlube" for $5 on Amazon. This is a "food-grade" lube which means it is ok to use on food prep equipment (not to eat) and is non-reactive. It is what "everyone" recommends for 3D printed gearboxes. amazon.com/Super-Lube-21030-Synthetic-Grease <https://www.amazon.com/Super-Lube-21030-Synthetic-Grease/dp/B000XBH9HI/ref=asc_df_B000XBH9HI/?tag=hyprod-20=df0=167152075853==g=2257681426504032291c===9031026=pla-31599188=1> Teflon is good stuff. It has low friction and is 100% non-reactive.I don't know what a carrier is but if the stuff is "food grade" i'd guess it is mostly harmless. On Wed, Dec 1, 2021 at 9:22 AM Gene Heskett wrote: > I'll also add some white lithium grease and then run it for a number of > hours. I suspect , > > > That petroleum based grease will craze the plastic, Sam. You need a > vegetable based grease or oil, crisco is cheaper than the "special" stuff > for plastic, which I've used on the bb's in miine. > > > OTOH the special stuff is IMO a bit thicker than crisco, I bought a small > can on it for $14 from > some guy in China. > > > Still on webmail (spit) excuse the mess. 7 damned bullseye installs later > I still don't have > the tools to make a proper email agent work.. > > > > Take care and stay well John. > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
John, Looks like I'm finding this thread late. Where are the design files for this harmonic drive? On Tue, Nov 30, 2021 at 11:22 PM John Dammeyer wrote: > I've added a photo. > www.autoartisans.com/harmonicdrive/HarmonicDrive-8.jpg > > The new wave generator pieces puts the bearings 1mm further out. Was also > able to bring it a bit closer to the motor so the bearings now sit in the > middle of the fixed spline gear. > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Harmonic Drive
That article is one of the best descriptions of a harmonic drive.But the problem is that 3D printed plastic is not strong enough to make a practical system.Note that the author says his drive can supply 3 Nm of torque. A stepper motor can do that already. On Sat, Nov 27, 2021 at 8:05 PM John Dammeyer wrote: > For those who have 3D printed the gear portions of the harmonic drive what > plastic did you find worked the best? > > Thanks > John > > > -Original Message- > > From: John Dammeyer [mailto:jo...@autoartisans.com] > > Sent: November-26-21 11:18 PM > > To: Enhanced Machine Controller (EMC) > > Subject: [Emc-users] Harmonic Drive > > > > List has been quiet lately on the harmonic drive subject. I stumbled on > this site. > > > > > https://howtomechatronics.com/how-it-works/what-is-strain-wave-gear-harmonic-drive-a-perfect-gear-set-for-robotics- > > applications > > > > John > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > _______ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] LinuxCNC with Robot Arm Setup
Checking and stopping is not good enough. A robot has to be able to "try" multiple solutions to the problem before it executes the movement plan. For example: It has been told to fetch a bottle of beer from the 'fridge and it finds the milk is in front of the beer. So it makes a possible plan to lift the milk, place it on the counter then lift the beer, place it on the counter and then replace the milk. Can this plan work? It needs to know in advance if a limit of torque or angle is reached on any of the six joints. Maybe the wrist motor can't keep the milk level? If so then the robot explores removing the pickle jar and then sliding the milk jug. Machine tools work the same way but only the planning is done by the CAM system or the human g-code programmer. In either case, an optimal solution is found from the hundreds of less good solutions long before LCNC gets the g-code file.It is not different with a robot. The "guy" who writes the g-code has to plan so as to prevent a crash. The other problem unique to robot arms is that collision avoidance must take into account the object that is grasped by the arm. You do not want the beer bottle to hit the pickle jar and drag the pickles to the kitchen floor. But all this is done. We don't need to use"physics engines". We can use "MoveIt." BTW, maybe we can see that robots will NOT be taking over jobs so soon. Today even fetching a beer is a very hard problem that is not yet fully solved. Loading a dishwasher or cleaning a toilet is an IMPOSSIBLE task. All blue collar jobs are very secure for a LONG time. . On Thu, Nov 18, 2021 at 10:16 AM Bari wrote: > On 11/18/21 00:54, Chris Albertson wrote: > > > > > I'm working on some 4+ axis CAM that if the machine and the work holding > >> devices are included the physics engine will check for collisions to > >> prevent crashes. > >> > A physics engine could be added to the simulator in LCNC to check for > collisions before they happen. > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] LinuxCNC with Robot Arm Setup
Andy, In the case of a robot arm you need to send (x, y, z, w, x, y) where w, x, y are rotational orientation. QUESTION: Can we also send the speeds for each target point (x', y', z', w', x', y'). What about acceleration like x'' ? Correct me, But I think this could be done using LCNC's trivial kinematics if the inverse kinematics is done in the AI-based front end. Then the location and speeds are in the motor's natural units in steps or encoder counts or whatever. Of course, LCNC would move in a straight line but the front end would send the points at something like 20 points per second. On Thu, Nov 18, 2021 at 5:50 AM andy pugh wrote: > On Thu, 18 Nov 2021 at 02:48, Ian Charnas wrote: > > > I am indeed planning on using computer vision to set targets for the > robot > > arm to move to. > > As a first stage this can be done using only whatever vision system > you use and LinuxCNC. > > It is easy enough to send an "MDI" command to LinuxCNC that will move > the end-point to a specified set of XYZ (BC) coordinates. > But that will be a straight-line move from the current position. > Anything more complex requires a planner, but initially you can ignore > that layer. > > An MDI command such as "F100 G1 X10 Y20 Z30" can be sent to LinuxCNC > in a number of ways, including: > http://linuxcnc.org/docs/2.8/html/man/man1/linuxcncrsh.1.html > or > http://linuxcnc.org/docs/2.8/html/config/python-interface.html > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] LinuxCNC with Robot Arm Setup
On Wed, Nov 17, 2021 at 8:38 PM Bari wrote: > Check out this thread on the forums "ROS LinuxCNC Link " > > https://forum.linuxcnc.org/38-general-linuxcnc-questions/42008-ros-linuxcnc-link Does anyone know if this was ever completed and made to work? My idea would be to implement a ROS topic to HAL pin bridge. I would write a ROS node that of course runs as a normal user-space process and also write one HAL task.These would be connected by pair of queues. Then any ROS node on the network could read or write any HAL pin. I don't know what kind of performance it could achieve. I'm working on some 4+ axis CAM that if the machine and the work holding > devices are included the physics engine will check for collisions to > prevent crashes. > I assume this is unrelated to the above. Do you have anything working yet to show us? -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] LinuxCNC with Robot Arm Setup
Asfor communicating between computers, you will find that one of RS' main purposes is to move data between nodes that might be (or not be) on different computers. you can even move the nodes around at run time. Just about everything in your proposed system would end up being a ROS node,even the user space app that talks to LCNC ROS uses a publish/subscribe kind of system. So I imagine some low level driver is publishingjoint angles from encoders and the video system needs to know where the location and orientation of the camera(s) so it subscribes to a topic or rather it uses a service that does spacial transforms to move pixel (x,y) to real word coordinates. Where these processie run is transparent. They can be in a robot mounted Pi4 oraLinuxserver ina server room. It is ROS' job to get dtafrom publisher to subscriber. We don't care if it is UDP or not. Mostly ROS2 now uses shared memory ifthe dat is on the samemachine LCNC has a similar concept. Data is written to a "pin" or a vertical wire and then is read from the Pin and the writers and reader don't have to knowabot each other. An arm as you describe is not a beginner project. I suggest your first robot be just a bare stepper motor on a work bench with the shaft up and some tape on it so you can see it rotate. Then use a gamepad (X-Box controller) to move the motor shaft. the gamepad is connected to a ROS node and then into LCNC. Get that to work and you learned about 80% of what you need and have the world's most complex remote controlled motor. Next get simulated armto working Gazebo and finally connect to the real arm Going for the final goal in one jump is not going to work, no one is that smart to make it all work the first time. Finally. Not many people here care about robots. They are machinists. But, please when you get the ROS/LCNC interface to work be sure and tell us. Such an interface would make LCNC usfullfor many more users In the end I bet you will decide to write a user-space HAL program that is also a ROS node. It will subscribe to ROS topics and also publish ROS topics., and at the same time, it will read and write to collection of HAL pins.Seems like a good idea. But the problem is I double a HALprogram can alsobe a ROS node as HAL is run periodically in real-time and ROS noodes like toblock and waitfor data.You will need to implement two programs and pair of queues in shared memory. On Wed, Nov 17, 2021 at 6:48 PM Ian Charnas wrote: > Chris, many thanks for the reply. > > I am indeed planning on using computer vision to set targets for the robot > arm to move to. Thanks for sharing the link to MoveIt. One solution could > be to have my computer vision code send the waypoints I want to MoveIt, > which would plan the path, and then send the path to LinuxCNC. > > In this scenario, I'm imagining I'd run the computer vision code and MoveIt > on a higher end workstation with a GPU (I can write gpu-accelerated > computer vision code), and then send the results perhaps over gigabit > ethernet using UDP (since it's faster than RS485 or CAN) to a separate > computer running LinuxCNC in real time. The second computer doesn't need to > be as powerful in terms of CPU and GPU which is nice. > > I had no idea about the learning curve on MoveIt. I looked around and found > that the authors just published a nice set of video tutorials > <https://www.youtube.com/channel/UCF7Yy57ZE2WNYeeXKEu8JQA/videos> on > MoveIt > 2 so I'll start there. > > I really appreciate the kind and helpful responses here, y'all are great! > Ian > > On Wed, Nov 17, 2021 at 8:54 PM Chris Albertson > > wrote: > > > On Wed, Nov 17, 2021 at 3:07 PM theman whosoldtheworld < > > bleachk...@gmail.com> > > wrote: > > > > > wow movelt . and there are some ethercat connection on movelt? > > thanks a > > > lot > > > > > > > As you plow through the stacks of documentation, you find that MoveIt > > outputs a "joint trajectory" Which is a set of messages with (time, x, > x', > > x'') The hardware interface is up to the user. MoveIt stops short of > the > > device drivers. > > > > This is why I said a hybrid with LCNC might be interesting. LCNCis goodat > > low levelmotor control andhas the Mesa FPGA ecosystem that work very > well. > > > > MoveIt is complex and has a really steep learning curve. It is built on > > top of ROS and ROS has a steep learning curve. I'd guess the stack is a > > couple orders of magnitude more complex than LCNC. It all runs on Linux > > and if you are not a Linux software developer then you have ever more to > > learn. It takes most new people about a year to come up to speed.A > > Movit-based 6-DOF robot arm is a tough introduction. You ar
Re: [Emc-users] LinuxCNC with Robot Arm Setup
On Wed, Nov 17, 2021 at 3:07 PM theman whosoldtheworld wrote: > wow movelt . and there are some ethercat connection on movelt? thanks a > lot > As you plow through the stacks of documentation, you find that MoveIt outputs a "joint trajectory" Which is a set of messages with (time, x, x', x'') The hardware interface is up to the user. MoveIt stops short of the device drivers. This is why I said a hybrid with LCNC might be interesting. LCNCis goodat low levelmotor control andhas the Mesa FPGA ecosystem that work very well. MoveIt is complex and has a really steep learning curve. It is built on top of ROS and ROS has a steep learning curve. I'd guess the stack is a couple orders of magnitude more complex than LCNC. It all runs on Linux and if you are not a Linux software developer then you have ever more to learn. It takes most new people about a year to come up to speed.A Movit-based 6-DOF robot arm is a tough introduction. You are best off starting with basic ROS tutorials and making system that can print "Hello World" to the screen. Also note the current transition from ROS-1 to ROS-2. New projects really should be using ROS-2. So check that you are reading the correct tutorials. Many are not labeled 1 or 2 because there was no need back when only ROS-1 existed.So look for explicit labeling as "2". > > Il giorno mer 17 nov 2021 alle ore 18:22 John Dammeyer < > jo...@autoartisans.com> ha scritto: > > > Very cool. Thanks for the link. > > John > > > > > -Original Message- > > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > > Sent: November-17-21 8:45 AM > > > To: Enhanced Machine Controller (EMC) > > > Subject: Re: [Emc-users] LinuxCNC with Robot Arm Setup > > > > > > If you are taking the right approach or not depends on the tasks you > want > > > this robot arm to reform. Are they entirely scripted such that you can > > > write G-code to control the arm? Or is this arm using a camera and > > vision > > > system to pick up randomly placed objects? > > > > > > If the former, LCNC will work fine and self-collision is a > > > non-issue because you programmed the G-code so as not to let that > happen. > > > But if the arm's path through space is determined in real-time and is > not > > > pre-programmed, LCNC is not the best tool, and you should be looking at > > > "MoveIt". https://moveit.ros.org/ > > > > > > But I think it is possible to build a hybrid system that uses BOTH > > software > > > systems. There is a part of LCNC that does motion control and has > > > programmatic API and this could be used to build an arm controller. > > MoveIt > > > then computes a trajectory. > > > > > > For LCNC experts not familiar with MoveIt... MoveIt outputs a series > of > > > "target points". for each motor. A joint has the following information > > > > > >- Time this target point shall be reached > > >- The location of the point > > >- The velocity > > >- The acceleration > > > > > > Question. Given that you have a user space process running that has > that > > > above information how to best pass this to LCNC so that LCNC can output > > the > > > stepper pluses or whatever is needed to drive the hardware. > > > > > > On Tue, Nov 16, 2021 at 9:12 PM Ian Charnas > > wrote: > > > > > > > Hi everyone. I'm new to LinuxCNC but I'm an engineer with a > background > > in > > > > software and cnc machining so I'm hoping I'll learn quickly. > > > > > > > > I'm interested in setting up LinuxCNC to run a robot arm. From what I > > can > > > > tell, the wise approach is to follow this guide > > > > <http://linuxcnc.org/docs/2.8/html/motion/dh-parameters.html> to > > setting > > > > up > > > > my modified DH parameters, and to use this guide > > > > <http://linuxcnc.org/docs/2.8/html/gui/vismach.html> to setting up a > > > > visual > > > > representation of the arm - either by importing STL files or by > drawing > > > > graphics primitives. > > > > > > > > My first question: am I heading down a wise path or are there other > > > > better options that I'm unaware of? > > > > > > > > My second question: Can LinuxCNC prevent the robot arm from crashing > > into > > > > itself, and if so how does it know when a crash wo
Re: [Emc-users] LinuxCNC with Robot Arm Setup
What LCNC lacks is high-level motion planning. LCNC's version of motion planning is means planing the speed of all the axis and keeping them coordinated so that the end effector(cutting tool) traces the specified path. What LCNC does not do is design the path. For example if the dripper is holding a metal pipe, LCNC will not be able to figure out it needs to rotet the pipr to a vertical position to avoid colidingwith some object. LCNC will but be able to figure out that it needs to rotate the hand so the fingers are parallel to the object that is to be grapsed.In the machine tool universe allthis is left towhat they call "CAM" software. In short CAM (or a human programmer) figures out the tool path and then CNC causes the motors to follow to prescribed path. In a hybrid system MoveIt would figure out the arm needs to reacharound the pickle jar in the 'fridge" to get your beer andthat the fingers need to gobehind thebeer and the thumb in front and the line from finger to thumb passing through the beer bottle. LCNC could be used to translate all those real-work velocities to stepper motor pulses. That said, many industrial robots run a fixed-preprogrammed path and are nothing more than a 6 axis machine tools running a program written in G-code. LCNC is a perfect fit to that. But LCNC would be of no help if the job is to grab the beer in the fridge and plan a motion path to extract to bottle without knocking over other objects. On Wed, Nov 17, 2021 at 2:13 AM andy pugh wrote: > On Wed, 17 Nov 2021 at 05:12, Ian Charnas wrote: > > > My first question: am I heading down a wise path or are there other > > better options that I'm unaware of? > > I think that is a reasonable approach. > You might want to think about whether you want to use genserkins or > pumpkins, or some other option. > Genserkins is very general and mathematical, but might not always find > a solution because of that. > Pumakins is only for Puma style robots, and might work better (or worse...) > Experiment with the two simulator configs included with LinuxCNC. See > which you find easiest to break. > (sim/axis/vismach/puma) > From the Readme: > "puma_cube.ini -- Config file for a PUMA type robot using the > puma-specific kinematics module pumakins. This configuration provides > a startup gcode file ([DISPLAY]OPEN)FILE) that sets usable coordinate > offsets to trace a cube outline. Velocity, acceleration and positional > limits are set large for convenience and are not representative of a > real machine. > > puma.ini -- Config file for a PUMA type robot using the puma-specific > kinematics module pumakins ($ man pumakins). Positional limits are not > enforced. System coordinate offsets need to be set by user. > > puma560.ini -- Config file for a PUMA 560 type robot using the > generalized serial kinematics of the genserkins module ($ man > genserkins). Genserkins uses an interactive solution for inverse > kinematics and may require that [JOINT_n]HOME locations are consistent > with [TRAJ]HOME settings." > > > My second question: Can LinuxCNC prevent the robot arm from crashing into > > itself, and if so how does it know when a crash would occur? Does it use > > the geometry of each link that I provide in vismach? > > Unfortunately LinuxCNC does not know how to prevent a robot from > colliding with itself. The Vismach representation is purely a one-way > display. > It's an interesting thought, though, that perhaps there is capability > in the OpenGL libraries used to perform collision detection during the > preview phase. > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > — George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] LinuxCNC with Robot Arm Setup
If you are taking the right approach or not depends on the tasks you want this robot arm to reform. Are they entirely scripted such that you can write G-code to control the arm? Or is this arm using a camera and vision system to pick up randomly placed objects? If the former, LCNC will work fine and self-collision is a non-issue because you programmed the G-code so as not to let that happen. But if the arm's path through space is determined in real-time and is not pre-programmed, LCNC is not the best tool, and you should be looking at "MoveIt". https://moveit.ros.org/ But I think it is possible to build a hybrid system that uses BOTH software systems. There is a part of LCNC that does motion control and has programmatic API and this could be used to build an arm controller. MoveIt then computes a trajectory. For LCNC experts not familiar with MoveIt... MoveIt outputs a series of "target points". for each motor. A joint has the following information - Time this target point shall be reached - The location of the point - The velocity - The acceleration Question. Given that you have a user space process running that has that above information how to best pass this to LCNC so that LCNC can output the stepper pluses or whatever is needed to drive the hardware. On Tue, Nov 16, 2021 at 9:12 PM Ian Charnas wrote: > Hi everyone. I'm new to LinuxCNC but I'm an engineer with a background in > software and cnc machining so I'm hoping I'll learn quickly. > > I'm interested in setting up LinuxCNC to run a robot arm. From what I can > tell, the wise approach is to follow this guide > <http://linuxcnc.org/docs/2.8/html/motion/dh-parameters.html> to setting > up > my modified DH parameters, and to use this guide > <http://linuxcnc.org/docs/2.8/html/gui/vismach.html> to setting up a > visual > representation of the arm - either by importing STL files or by drawing > graphics primitives. > > My first question: am I heading down a wise path or are there other > better options that I'm unaware of? > > My second question: Can LinuxCNC prevent the robot arm from crashing into > itself, and if so how does it know when a crash would occur? Does it use > the geometry of each link that I provide in vismach? > > many thanks for any tips and advice, > Ian > > -- > Ian Charnas | Engineer | iancharnas.com <http://www.iancharnas.com> > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] USB to Ethernet
You can work out the delay of a cut-through switch at 100M bits per second. Comes to a handful of microseconds. But does it matter if the delay is always the same? It is variability that matters. We are about talking nanoseconds. That said, I'd not want to use a network that was also being used by a few others to stream a video. But even then, a "managed" switch would fix it for you. Try it and measure the results. My bet is that you can have several LCNC computers and machine tools on the same Eithernet switch. On Tue, Nov 16, 2021 at 10:53 AM johnd wrote: > I was under the impression that the Mesa required dedicated Ethernet so > UDP packets wouldn't be delayed. JohnSent from my Samsung S10 > Original message ----From: Chris Albertson < > albertson.ch...@gmail.com> Date: 2021-11-16 9:51 a.m. (GMT-08:00) To: > "Enhanced Machine Controller (EMC)" > Subject: Re: [Emc-users] USB to Ethernet Do you mean that the notebook PC > has Ethernet, but it is in use, and youwant to add a second Ethernet > port? That is not the way to go. Buy anEthernet Switch. Or perhaps > your notebook is already connected to aswitch, then just plug the Mesa card > into the switch. Not only is this acheaper solution it is more efficient > of resource in the PC and Mesa can'tknow the difference.> And the one > Ethernet port on the machine is connected to the MESA board.> I don't have > reliable WiFi so a USB to Ethernet would be one way to get> into it. > Although working with the probe I'd rather be right there with my> hand on > the ESTOP.>> John>>>>> ___> > Emc-users mailing list> Emc-users@lists.sourceforge.net> > https://lists.sourceforge.net/lists/listinfo/emc-users>-- Chris > AlbertsonRedondo Beach, > California___Emc-users mailing > listEmc-users@lists.sourceforge.nethttps:// > lists.sourceforge.net/lists/listinfo/emc-users > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] USB to Ethernet
Do you mean that the notebook PC has Ethernet, but it is in use, and you want to add a second Ethernet port? That is not the way to go. Buy an Ethernet Switch. Or perhaps your notebook is already connected to a switch, then just plug the Mesa card into the switch. Not only is this a cheaper solution it is more efficient of resource in the PC and Mesa can't know the difference. > And the one Ethernet port on the machine is connected to the MESA board. > I don't have reliable WiFi so a USB to Ethernet would be one way to get > into it. Although working with the probe I'd rather be right there with my > hand on the ESTOP. > > John > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Prox sensors and wiring
I bought some of the proximity detectors for my 3D printer to measure the bed.I tested them on my mill by moving them untillthey tripped, backing off and doing it again to see how repeatable they are.Even the cheap no-=brand unit where as good as my ability to measure them. Set up a test using a dial indicator. The absolute trip point was some random distance, all you care about is how repeatable they are. The one I liked is mounted in M12x1 threads. I bought am M12x1 tap to make a mounting block The usual connector is called an "Aviation Plug". they come in different diameters and with different amounts of pins. The smaller size is enough for signals and the larger ones can handle motor current.They come in different quality levels too. The cheap Chinese ones are "good enough" but the American ones are precision made and have rubber gaskets. Prices are about $3.50 for the cheap ones (on eBay or Amazon) and about $35 for the best quality ones (at Mc Master Carr and the like). The cheap plugs are made of something like chromed zinc and look cheap but work surprisingly well. The best one are made of machined an green-primmered aluminum and have the quality you'd expect for an $80 per mating par part. They are also sold with engineering-grade plastic housing On Wed, Nov 3, 2021 at 10:01 PM Ralph Stirling < ralph.stirl...@wallawalla.edu> wrote: > I have finally started stripping out the control cabinet on my cnc mill in > preparation for my retrofit. The brushed servos and mechanical limit and > home switches were wired up with crimped "bullet" quick connect pins. I'm > replacing the servos with brushless servos, and am considering replacing > the mechanical switches with inductive prox sensors (pnp, nc type). All > the old wires are sticky with coolant and metal chips. > > So, I am interested to hear what other lcnc retrofitters have found works > well for modest priced coolant proof connectors (3 or 4 pin), and opinions > on cheap Chinese prox sensors (since the name brand ones are so expensive, > even on ebay). An example is: > https://www.ebay.com/itm/US-5Pcs-NC-PNP-LJ18A3-8-Z-AY-Inductive-Proximity-Sensor-Switch-DC6V-36V-/143861840692 > > Photos (rather unorganized) of the retrofit are at: > https://photos.app.goo.gl/yBSRVf3QAVUK39PC7 > > Thanks, > -- Ralph > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Requesting a short history review, if you'd be so kind.
Gene is getting good results with his 3-phase steppers and no PID control in LCNC because the 3-phase steppers have internal PID position control loops in the motor driver electronics. These motors have an encoder wheel for feedback. So these steppers are basically the same as a servo with feedback implemented on the servo driver. >From my experience with motion control (robots), the BEST place to run the control loop is in hardware attached to the motor. Typically, this is a microcontroller There, the loop can run as fast as it needs to and the timing jitter is a non-issue. On Thu, Oct 28, 2021 at 11:14 AM Sam Sokolik wrote: > This is a quote from peter on the forum.. > > 'I have done some experimentation and have found that you can get better > stepgen following performance on systems with moderate jitter by using a > PID loop for position control instead of the stepgen drivers built-in > position control loop. > The problem with the built in driver position control is that its tends to > overcorrect > for measured position errors cause by sampling jitter.' > > On Thu, Oct 28, 2021 at 12:35 PM Gene Heskett > wrote: > > > On Thursday 28 October 2021 11:45:49 Sam Sokolik wrote: > > > > > All my machines - steppers and servos are running pid.I think > > > Peter forgot to mention - The reason PID is recommended is because it > > > handles latency spikes way better. We have seen situations where a > > > latency spike confuses the hell out of position mode stepgens. > > > (ymmv) > > > > But thats why we run preempt-rt kernels Sam. Even on the pi, the maximum > > spike is 200 microseconds while FF is initializing. W/O FF, 4 to 50 > > u-secs. And the big lathe has never noticed that. The little mill, > > before it commited suicide by spitting out all the balls of the y nut, > > was running sw steppers, and they did notice occasionally. It also > > severely limited that machines rapids, to around 8 or 9" a minute. > > Putting a 5i25 in it and it ran about 50 ipm w/o any stutters. > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] github problem?
t; > >
> > > ___
> > > Emc-users mailing list
> > > Emc-users@lists.sourceforge.net
> > > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
> > Peter Wallace
> > Mesa Electronics
> >
> > (\__/)
> > (='.'=) This is Bunny. Copy and paste bunny into your
> > (")_(") signature to help him gain world domination.
> >
> >
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
>
>
> Cheers, Gene Heskett.
> --
> "There are four boxes to be used in defense of liberty:
> soap, ballot, jury, and ammo. Please use in that order."
> -Ed Howdershelt (Author, 1940)
> If we desire respect for the law, we must first make the law respectable.
> - Louis D. Brandeis
> Genes Web page <http://geneslinuxbox.net:6309/gene>
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
--
Chris Albertson
Redondo Beach, California
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] github problem?
You should not need a password to clone a publicly available repository. Could you post a link to the exact place that is not working? Or maybe netter, cut and paste the command you use and the resulting error message. On Tue, Oct 26, 2021 at 11:46 AM Gene Heskett wrote: > Greetings all; > > Trying to use my email and and a passwd to clone a github repo, I just > ran into a big can of worms. According to the error msg, passwds were > canceled on Aug 13, and they are now useing a personal token. no clue > given as to how to acquire that. > > I also used the normal passwd to reclone LinuxCNC without a problem on > Sept 28th, well after the error msg said it should have been blocked. > > Can anyone clairify what is going on? > > Thanks all. > > Cheers, Gene Heskett. > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author, 1940) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] 1973 cnc retro tech.. How to switch between metric and English
3D printed geardsworks well for threading and cost less then $1 to make (assuming you have a printer) Yes gears are hard to design but all the gear companies have CAD files on their websites. I just go to McMaster Carr and download the CAD file, do some trivial mods to make the hubs and face width what I want and then print it. There is a Youtube video where a guy makes a printed gear and tried to intentionally break it by making very deep cuts and he can't break the plastic gear. These gears have only small loads compared to the spindle. Not bad for under $1 and an hour of your time. On Tue, Oct 26, 2021 at 5:47 AM Eric Keller wrote: > I wanted to modify my lathe to cut metric threads. But it requires 7 > gears. Of course none of them are commercially available. I don't really > want to be in the business of hobbing gears, so I'm going to an electronic > gearbox instead. The stock setup was a timing belt, so hooking up a motor > is a lot easier than cutting gears. I am not positive, but I think you > could switch back and forth with the gear train setup. > Eric Keller > Boalsburg Pennsylvania > > > On Mon, Oct 25, 2021 at 10:02 PM Sam Sokolik wrote: > > > We are finally getting around to converting the 73 > > cincinnati milacron lathe. > > > > https://www.youtube.com/watch?v=-J7EmnRZXjw > > > > At the end of the video I show the switch that switches between 2 > > differently geared resolvers. > > > > On the back of the hydraulic servo is a gearbox. first is a 2.5:1 > > increaser that makes it so the english resolver spins 5 times per inch > (.2" > > of axis movement - 2TPI lead screw) then there is a set of gears 127:50 > > (2.54:1) for the resolver used for metric. Than that resolver runs at > 2mm > > per rev. Amazing how it was cheaper to do this in hardware than in the > > control (ttl ic's) > > > > https://photos.app.goo.gl/yezaTdnNWZHg9Dm4A > > > > https://photos.app.goo.gl/QNDzVJmnD9HWjs8LA > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Panasonic serial encoder interface
Just curious, what will this robot do when it is finished? On Fri, Oct 22, 2021 at 10:20 AM Ralph Stirling < ralph.stirl...@wallawalla.edu> wrote: > They are in a Denso robot, and the robot controller directly > reads the encoders. Several fpga's on the main board. We > don't have the big cable that connects the controller to the > arm, unfortunately, but may try to cobble together one > channel to see if we can capture communication between > the controller and the encoder. All the batteries were dead, > so all stored info is gone. > > -- Ralph > > From: Todd Zuercher [to...@pgrahamdunn.com] > Sent: Friday, October 22, 2021 6:44 AM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Panasonic serial encoder interface > > Are you using the Panasonic servo drives that go with the motors and > encoders? How are you commanding them? Depending on the command interface > you may not need to deal with reading the encoders directly or even at > all. The drives will do that for you and maybe even send out a position > signal linuxcnc is capable of reading for feedback. > > Todd Zuercher > P. Graham Dunn Inc. > 630 Henry Street > Dalton, Ohio 44618 > Phone: (330)828-2105ext. 2031 > > -Original Message- > From: Ralph Stirling > Sent: Thursday, October 21, 2021 11:14 PM > To: emc-users@lists.sourceforge.net > Subject: [Emc-users] Panasonic serial encoder interface > > [EXTERNAL EMAIL] Be sure links are safe. > > We are digging into a 2002-vintage Denso 6-axis robot, with a Linuxcnc > conversion in mind. The joint motors on this robot appear to have > Panasonic encoders with a four wire interface and a fair amount of > circuitry inside each encoder. I suspect that these are the same as the > ones in this EEVblog thread: > > > https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.eevblog.com%2Fforum%2Fprojects%2Finterfaceprotocol-to-encoder-on-panasonic-servo-motor%2Fdata=04%7C01%7Cralph.stirling%40wallawalla.edu%7C665f7f0c23a742ef04e708d9956f225a%7Cd958f048e43142779c8debfb75e7aa64%7C0%7C1%7C637705126827244953%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=RYymtPoDjeH8QREN4liUgmrumfSUAtmFaHB3rvb4Bmg%3Dreserved=0 > > A rather terse comment at the end of that thread suggests that these are > 2.5Mbaud async RS485 communications: > >2.5MBps RZ 8N1 RS485 >0x52 comes from drive - seems encoder reset. >changes afterwards to 0x2a - read encoder command > > We already determined that ours are using RS485, so this seems to confirm > we are looking at right things. Anybody ever tried interfacing such > encoders to LCNC? > > Thanks, > -- Ralph > > ___ > Emc-users mailing list > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Fwd: Rogue Index Pulses
Differential drive is always the best solution, look at Ethernet. It can push 100 megabit over a kilometer using unshielded cable. But single ended signals can be noise resistant if the wire uses resistive termination. The opto-isolator is proving resistive termination. On Wed, Oct 13, 2021 at 6:43 PM John Dammeyer wrote: > Hi Chris, > When I developed my ELS for 1PPR one of the criteria was that the index > vane or slot had to be wide enough so that it would generate a pulse at > least about 5 or so 50uS ticks wide and at least again that wide for the > non index period. That the tick count would then be at least about 500 uS > per rev. That's a frequency of 2000 Hz in minutes 120,000 RPM. Very > unlikely an old surplus lathe can run that fast. > > A lathe like a small Unimat DB-200 could run 8100 RPM which is a 135 RPS > or a period of about 7.4mS. If the index pulse slot is 250uS the non index > pulse time is 0.00715mS or about 143 ticks. So it's pretty easy to expect > the index pulse to be active for 3 ticks or 150uS. Most noise is much > larger than that. > > For a system where there are spurious index pulses or noise, and the > HPCL2631 solution I proposed wasn't possible, I'd be tempted to add a > 1-shot pulse extender right at the encoder to create a longer index pulse. > Once again debouncing that in software (or hardware with a counter that > verifies low for X clocks) creates a reliable index. > > Since Peter's encoder had push pull outputs and a drive capability of at > least 10mA it made sense to go high speed optical. If it wasn't and had > simple TTL output then I'd have suggested adding high speed differential > drivers like the AM26LS31CN which has 4 channels. See attached schematic. > > A differential receiver at the other end and using twisted pair 100Ohm > impedance wire with 100 ohm resistors at both ends creates a nice balanced > transmission line more immune to interference. > > John Dammeyer > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: October-13-21 6:10 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Fwd: Fwd: Rogue Index Pulses > > > > You are right, but under all conditions, the resistor connecting the wire > > to ground through a diode has lower impedance than the same wire > terminated > > at a transistor gate as was the case before. The proof is that it works > > now. > > > > The other thing was thinking is that it should be simple to use software > to > > filter out a ghost pulse from an encoder. After all, we know that it can > > only happen once per revolution. We could look at the counter that > racks > > A/B quadrature pulses and know it is is time for an index. Software > > could do this even with no configuration by spinning the shaft. In fact > > verifying that the index pulse always happens at the expected time is a > > great why to error-check that no counts were dropped or added. It would > > be a great self-dignostic. > > > > On Wed, Oct 13, 2021 at 4:11 PM John Dammeyer > > wrote: > > > > > To fair we need to use the word low resistance path to ground. The > > > impedance of the wiring at 300kHz where electrical noise might exist > from a > > > VFD may be considerably higher. > > > > > > Think of the basic electronics of an inductor and capacitor in > parallel. > > > At DC the resistance and impedance is the coil resistance to ground. > At > > > ultra high frequencies the reactance of the capacitor is likely so low > that > > > it's also a short circuit to ground. And in-between somewhere at some > > > frequency the pair are in resonance and the 'impendance' can be very > high. > > > > > > The wire from the encoder to the electronics has inductance, > especially if > > > coiled up and there is capacitance per foot/meter rating on various > > > cables. It may or may not be important depending on the noise > frequency. > > > > > > John Dammeyer > > > > > > > -Original Message- > > > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > > > Sent: October-13-21 3:51 PM > > > > To: phodg...@uk22.net; Enhanced Machine Controller (EMC) > > > > Subject: Re: [Emc-users] Fwd: Fwd: Rogue Index Pulses > > > > > > > > Then this proves the cause. The isolators provide a low impedance > path > > > to > > > > ground. It is just current-limiting resistor and a forward-biased > diode. > > > > With this low impedan
Re: [Emc-users] Fwd: Fwd: Rogue Index Pulses
You are right, but under all conditions, the resistor connecting the wire to ground through a diode has lower impedance than the same wire terminated at a transistor gate as was the case before. The proof is that it works now. The other thing was thinking is that it should be simple to use software to filter out a ghost pulse from an encoder. After all, we know that it can only happen once per revolution. We could look at the counter that racks A/B quadrature pulses and know it is is time for an index. Software could do this even with no configuration by spinning the shaft. In fact verifying that the index pulse always happens at the expected time is a great why to error-check that no counts were dropped or added. It would be a great self-dignostic. On Wed, Oct 13, 2021 at 4:11 PM John Dammeyer wrote: > To fair we need to use the word low resistance path to ground. The > impedance of the wiring at 300kHz where electrical noise might exist from a > VFD may be considerably higher. > > Think of the basic electronics of an inductor and capacitor in parallel. > At DC the resistance and impedance is the coil resistance to ground.At > ultra high frequencies the reactance of the capacitor is likely so low that > it's also a short circuit to ground. And in-between somewhere at some > frequency the pair are in resonance and the 'impendance' can be very high. > > The wire from the encoder to the electronics has inductance, especially if > coiled up and there is capacitance per foot/meter rating on various > cables. It may or may not be important depending on the noise frequency. > > John Dammeyer > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: October-13-21 3:51 PM > > To: phodg...@uk22.net; Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Fwd: Fwd: Rogue Index Pulses > > > > Then this proves the cause. The isolators provide a low impedance path > to > > ground. It is just current-limiting resistor and a forward-biased diode. > > With this low impedance to ground there is no way for EMI to cause the > > voltage on the wire to raise. > > > > On Wed, Oct 13, 2021 at 1:40 PM Peter Hodgson < > peterjohnhodg...@gmail.com> > > wrote: > > > > > > > > Well! > > > > > > A quick update for those that are interested. > > > > > > The HPCL2631's won the postal race and arrived in the mail today. > > > > > > I made up a little 'interface' board to mount them on with the pull > ups, > > > caps, etc. and fitted that board this evenening. > > > > > > After an initial check on the index pulse I ran my G76 program (in air) > > > for the M47 x 3.0 thread, that as giving me so much trouble, whilst > > > watching Halscope for the encoder index and I didn't see one ghost > > > pulse. I did see one that was missing over the 7 mins but I don't think > > > a missing pulse is going to give me as much trouble as ghost pulse and > > > anyway that could have been a software glitch. > > > > > > Too late to cut metal now but I will tomorrow. > > > > > > Looking good though! > > > > > > Pete > > > > > > > > > > > > On 13/10/2021 19:12, John Dammeyer wrote: > > > > I'll agree with Andy here. I have a dual trace older digital scope > and > > > a 4 trace digital that can even decode CAN, SPI and I2C messaging. I > > > bought a small Digilant PC development kit version for $400 or so to > get > > > UART decoding for a PAN-TILT system that was misbehaving. The add on > > > module for the scope to do that was $1200 so a no brainer. Haven't > used it > > > since we solved the serial port issues 3 years ago. > > > > > > > > But my 4 trace I use all the time. The 2 trace has spent time out in > > > the shop by the mill but was useless for determining what the issue was > > > with the out of spec DC servo that only lost steps in one direction and > > > never the other. The G-54 zero crept in one direction resulting in the > > > center drill peck holes being in one place and then a bit later the > drill > > > bit for those holes not lining up. > > > > > > > > In a noisy shop environment just getting the scope to see real data > as > > > opposed to the noise it picks up may be impossible. I saw that at a > > > Siemen's plant in Regensberg Germany back in the early 90's. We > scoped a > > > CAN bus and could barely see the signal on the differential pair yet > the > >
Re: [Emc-users] Fwd: Fwd: Rogue Index Pulses
Then this proves the cause. The isolators provide a low impedance path to ground. It is just current-limiting resistor and a forward-biased diode. With this low impedance to ground there is no way for EMI to cause the voltage on the wire to raise. On Wed, Oct 13, 2021 at 1:40 PM Peter Hodgson wrote: > > Well! > > A quick update for those that are interested. > > The HPCL2631's won the postal race and arrived in the mail today. > > I made up a little 'interface' board to mount them on with the pull ups, > caps, etc. and fitted that board this evenening. > > After an initial check on the index pulse I ran my G76 program (in air) > for the M47 x 3.0 thread, that as giving me so much trouble, whilst > watching Halscope for the encoder index and I didn't see one ghost > pulse. I did see one that was missing over the 7 mins but I don't think > a missing pulse is going to give me as much trouble as ghost pulse and > anyway that could have been a software glitch. > > Too late to cut metal now but I will tomorrow. > > Looking good though! > > Pete > > > > On 13/10/2021 19:12, John Dammeyer wrote: > > I'll agree with Andy here. I have a dual trace older digital scope and > a 4 trace digital that can even decode CAN, SPI and I2C messaging. I > bought a small Digilant PC development kit version for $400 or so to get > UART decoding for a PAN-TILT system that was misbehaving. The add on > module for the scope to do that was $1200 so a no brainer. Haven't used it > since we solved the serial port issues 3 years ago. > > > > But my 4 trace I use all the time. The 2 trace has spent time out in > the shop by the mill but was useless for determining what the issue was > with the out of spec DC servo that only lost steps in one direction and > never the other. The G-54 zero crept in one direction resulting in the > center drill peck holes being in one place and then a bit later the drill > bit for those holes not lining up. > > > > In a noisy shop environment just getting the scope to see real data as > opposed to the noise it picks up may be impossible. I saw that at a > Siemen's plant in Regensberg Germany back in the early 90's. We scoped a > CAN bus and could barely see the signal on the differential pair yet the > system operated flawlessly. Just couldn't find the right ground or > whatever... > > > > So the money spent on a scope may not be worth anything. > > > > John > > > > > >> -Original Message- > >> From: andy pugh [mailto:bodge...@gmail.com] > >> Sent: October-13-21 3:22 AM > >> To:phodg...@uk22.net; Enhanced Machine Controller (EMC) > >> Subject: Re: [Emc-users] Fwd: Rogue Index Pulses > >> > >> On Wed, 13 Oct 2021 at 11:13, Peter Hodgson > wrote: > >> > >>> I'm watching a couple of Oscilloscopes on eBay so might take the > plunge. > >> I am going to go against the consensus and suggest that if you don't > >> already have an oscilloscope, and don't know how to use one, then you > >> might not find it much help. > >> > >> I certainly got nowhere trying to use one to track down false > >> triggering of limit switches. And I have owned a 'scope for decades > >> (though not used it more than a few times a year) > >> > >> -- > >> atp > >> "A motorcycle is a bicycle with a pandemonium attachment and is > >> designed for the especial use of mechanical geniuses, daredevils and > >> lunatics." > >> � George Fitch, Atlanta Constitution Newspaper, 1912 > >> > >> > >> ___ > >> Emc-users mailing list > >> Emc-users@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-users > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Rogue Index Pulses
Yes, experimental setup does require some skill. Sometimes you even have to build a test connection to the scope with a connector on it. Then on the scope there are setting for "DO coupled" or "AC coupled and for placing in low pass filters and trigger level and on the digital scope even more settings with menus three levels deep Logic analyzers are another good tool. These can have 8 or 16 onput channels and can have very complex triggering and recording, Mixed signal scopes combine both. Some good logic analyzers can be had for only $10. People with no budget can make-do with make-shift things like setting up a iPhone camera to video-record a 1970's vintage analog scope.Then you can scroll back and look. Many times the key is being smart in how you rig the experiments. For example measuring inductance with a scope takes some thinking and a plan. The nest subject after buying a scope is scope-probes. These are not just simple wires with clips. Almost always they ate 10 to 1 attenuator but I have some 100:1 that switch to 1000:1 Probes are a whole field of study in themselves. On Wed, Oct 13, 2021 at 11:16 AM John Dammeyer wrote: > I'll agree with Andy here. I have a dual trace older digital scope and a > 4 trace digital that can even decode CAN, SPI and I2C messaging. I bought > a small Digilant PC development kit version for $400 or so to get UART > decoding for a PAN-TILT system that was misbehaving. The add on module for > the scope to do that was $1200 so a no brainer. Haven't used it since we > solved the serial port issues 3 years ago. > > But my 4 trace I use all the time. The 2 trace has spent time out in the > shop by the mill but was useless for determining what the issue was with > the out of spec DC servo that only lost steps in one direction and never > the other. The G-54 zero crept in one direction resulting in the center > drill peck holes being in one place and then a bit later the drill bit for > those holes not lining up. > > In a noisy shop environment just getting the scope to see real data as > opposed to the noise it picks up may be impossible. I saw that at a > Siemen's plant in Regensberg Germany back in the early 90's. We scoped a > CAN bus and could barely see the signal on the differential pair yet the > system operated flawlessly. Just couldn't find the right ground or > whatever... > > So the money spent on a scope may not be worth anything. > > John > > > > -Original Message- > > From: andy pugh [mailto:bodge...@gmail.com] > > Sent: October-13-21 3:22 AM > > To: phodg...@uk22.net; Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Fwd: Rogue Index Pulses > > > > On Wed, 13 Oct 2021 at 11:13, Peter Hodgson > wrote: > > > > > I'm watching a couple of Oscilloscopes on eBay so might take the > plunge. > > > > I am going to go against the consensus and suggest that if you don't > > already have an oscilloscope, and don't know how to use one, then you > > might not find it much help. > > > > I certainly got nowhere trying to use one to track down false > > triggering of limit switches. And I have owned a 'scope for decades > > (though not used it more than a few times a year) > > > > -- > > atp > > "A motorcycle is a bicycle with a pandemonium attachment and is > > designed for the especial use of mechanical geniuses, daredevils and > > lunatics." > > � George Fitch, Atlanta Constitution Newspaper, 1912 > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Rogue Index Pulses
cabling. The stepper > >>> drivers control the motor current by turning themselves on and off > >>> at an ultrasonic frequeny we don't hear. If you can find "starquad" > >>> cabing in a gage heavy enough it doesn't run warm at the motors > >>> current. It is actualy the gold standard microphone cable, a top > >>> quality microphone cable available in several gages, all VERY > >>> flexible, get the lowest gage number Suzan has. 22 gage IIRC. Ground > >>> the shielding drain wire at the star bolt, trim and insulate it at > >>> the motor end. > >>> > >>>> It seems I have three options from here. > >>>> > >>>> 1) Change to the HPCL2631 opto isolators. > >>>> > >>>> 2) Change to the 74HC14 buffer. I think I will need resistance > >>>> dividers with this as the max input in the datasheet suggests 6v so > >>>> I will need to drop the 12v encoder signal to <6v (?) > >>>> > >>>> 3) Try the existing 74HC4050 buffer with resistance dividers. > >>>> > >>>> I’ve got some components on order so I guess whichever turns up > >>>> first will be the first I’ll try. > >>>> > >>>> Seems like I also need to find myself some sort of oscilloscope on > >>>> eBay! > >>> Digital storage will show you stuff that cannot be seen on an > >>> analogue scope. Definitely worth the extra sheckles. Some of this > >>> stuff is at 100 or more megahertz. And tends to be very dim on a > >>> analog scope. I actually have 3, a 30 yo Hitachi 100 mhz dual trace > >>> analog, much better than a tek of the same vintage, a 5 yo digital > >>> with the same specs you can get for $300 or so today, and I just > >>> bought Siglents best, a 4 trace, 350 mhz digital sampler. It also > >>> costs a down payment on a new small car. Inheritances are handy. > >>> > >>>> I’ll keep you posted. > >>> Please. > >>> > >>>> Cheers, > >>>> > >>>> Pete > >>>> > >>>>> On 11 Oct 2021, at 21:11, Gene Heskett > >>>>> wrote: > >>>>> > >>>>> On Monday 11 October 2021 15:49:43 Chris Albertson wrote: > >>>>>> You could be correct. High impedance is a recipe for noise. > >>>>>> I had suggested a resistive divider just because it is simpler. > >>>>>> But you are right about providing a ground path. A divider > >>>>>> certainly would do that. If an opto is really needed then use a > >>>>>> high-value resistor to ground to keep the line from floating and > >>>>>> bleed off static. > >>>>>> I also don't like the idea of grounding the shield on the encoder > >>>>>> end as it makes it impossible to know the path from encoder > >>>>>> housing back to true Earth ground. It is "unanalyzable" (if such > >>>>>> a word exists) Running the shield to star ground point makes it > >>>>>> easy to verify it is correct. > >>>>> +100 Chris. Run a separate ground to the encoder from the star > >>>>> bolt, and connect the cables overall shield ONLY to that bolt. If > >>>>> that encoder uses the shield as its ground connection, toss it in > >>>>> the out bin, and get one that does have a separate ground wie > >>>>> going into it which is isolated from the metalic case. I would > >>>>> also verify that the encoder has a good ground to its metalic > >>>>> housing. Painted brackets are a recipe for failure. As are > >>>>> metallic shaft couplers. The elastomeric coupler that came with my > >>>>> omron, failed a year ago, and the coupling is now a couple layers > >>>>> of heat shrink with the inside layer of thermal glue. If it fails, > >>>>> replace it with a fresh copy. 50 cents maybe. > >>>>> > >>>>> Cheers, Gene Heskett. > >>>>> -- > >>>>> "There are four boxes to be used in defense of liberty: > >>>>> soap, ballot, jury, and ammo. Please use in that order." > >>>>> -Ed Howdershelt (Author, 1940) > >>>>> If we desire respect for the law, we must first make the law > >>>>> respectable. - Louis D. Brandeis > >>>>> Genes Web page <http://geneslinuxbox.net:6309/gene> > >>>>> > >>>>> > >>>>> ___ > >>>>> Emc-users mailing list > >>>>> Emc-users@lists.sourceforge.net > >>>>> https://lists.sourceforge.net/lists/listinfo/emc-users > >>>> ___ > >>>> Emc-users mailing list > >>>> Emc-users@lists.sourceforge.net > >>>> https://lists.sourceforge.net/lists/listinfo/emc-users > >>> Cheers, Gene Heskett. > >> ___ > >> Emc-users mailing list > >> Emc-users@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-users > > > > Cheers, Gene Heskett. > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Replacing a handle.
If you can afford it, a part like this can be 3D printed in stainless steel. I think SpaceX 3D prints the injectors used on their Raptor rocket engines in some exotic material like Inconel. Printer that can do that are expensive but you can use a service. But I think the point of this handle is an exercise, not to make a handle. Given that you have a CNC mill how to you best make an exact replica? Think of this as a homework assignment. Here is what I would do... Take the CAD file and orient it so the two drill holes are on the vertical axis. (I think? the crank and handle holes are parallel) The in the CAD system add a metal pedestal The elevates the crank. On the bottom of the pedestal are two blind threaded holes, perhaps M5 or M6 size. Start by making the hole on the backside of the billet. Fip the billet and screw it to the table from the bottom. Mill the crank top and sides and drill the holes. Flip the crank and use the two holes you just drilled to screw the part to the table, pedestal facing up. You need spacers under the crank to make the drilled holes vertical Using an end mill finish the side of the crank that is now facing up. This will turn the entire pedestal into chips. For both cuts the top surface is flat but horizontal, so maybe a ball mill is required, at least for finishing A second method is to design the part with bridges to the billet. You cut the part, flip the billet and cutfrom the other side and then there is a crank suspended inside the scrap and connected to it by maybe three bridges. then you cut them with a saw and the part falls out. Finally, you smooth over the saw cuts.THis is like casting where you have to saw off some parts. On Wed, Oct 13, 2021 at 8:17 AM Bruce Layne wrote: > > > On 10/13/21 12:02 AM, John Dammeyer wrote: > > Chrome plated pot metal and if that can crack and break a 3D printed > handle would definitely. > > An FDM 3D printed crank arm would almost certainly break, regardless of > the material used. > > An MSLA (resin) 3D printed crank arm almost certainly would not break. > I would fix this problem by creating the part in FreeCAD and printing it > on a resin 3D printer using an "ABS-like" structural resin. > > A part's strength can be measured in many ways, but in general, MSLA > resin printed 3D parts created using a structural polyurethane resin are > stronger than aluminum but not as strong as steel. The polyurethane is > as durable as a hockey puck and polyurethane is very resistant to > cracking, unlike aluminum or the pot metal original part. > > Even for home gamers, 3D printers are not just filament based. Resin goo > MSLA 3D printers are now fairly cheap, and the resolution is almost as > good as injection molding. The parts are very strong and the material > cost is around US$.03 per gram. Part strength is isotropic so layer > delamination isn't a problem when designing parts as it is with FDM 3D > printing. The biggest MSLA downside is the isopropyl alcohol part > cleaning and UV post curing, but it's not that bad. > > Someone needs to make an upgraded MSLA printer that automates the post > processing operations. > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Replacing a handle.
Objvoiuly you could redesign the handle to use some parallel sides and it would be easy to make. But I assume this is an exercise Repair would be the easy way to go. Mill off the broken fork side and use a plate to replace it taach the plate with small screws. 3D printing would be easy and you might even send the design files to an outfit who can print in metal. They can print stainless steel or even gold. So printed does not have to be plastic. But if you were to mill this, the trick is to make the clamp tools. For example if you make a taper then the handle (and the taper) fit in a milling vice. You would end up making a set of custom tapered clamp shims. The shims can be machined or printed I would just repair this. On Tue, Oct 12, 2021 at 8:36 PM John Dammeyer wrote: > This has nothing to do with LCNC specifically. Really more a how to use > LCNC. > > The attached picture is of a broken tripod crank handle that clearly was a > casting. > > I can't figure out how I'd hold the raw stock to machine it. The slot > would be easy. But the angles and tapers and curves go in three different > dimensions. > > Maybe a stub on the end held in a 4th axis? > > Any suggestions are welcome. > John > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Rogue Index Pulses
The trouble with guessing that it might be noise on the AC mains power or radiate noise from motor cables or a ground loop is that when you get lucky and the problem stops you never know it is was because by luck you happened to coil some cable differently or that you actually fixed it. Without seeing the root cause you can't know that the problem will not come back. 'Sopes are useful and Gen is right, The Chinese digital ones are good and worth themoney. The older vintage analog scopes are good too if you do not want to pay so much.Having one vs. not having one matters a thousand times more that what kind you have. Just do make sure the bandwidth is at a bare minimum 50 MHz, but 100 is better. That said a logic analyzer is almost as good as a scope and costs about $10. These things can only work on logic-level signals. They allow you to plot up to 8 or more signals. You can look at any signal AFTER it has been shifted to a logical level. THese do nothelp yu see noise, they only record high/low or 1/0 levels but still very usfull. There are MANY Saleaeclones on eBay. On Mon, Oct 11, 2021 at 6:54 PM Gene Heskett wrote: > On Monday 11 October 2021 17:32:31 Peter Hodgson wrote: > > > Thanks all for your continued support. > > > > I’ve now separated all the earth grounding to individual cables going > > to one bolt in the steel control enclosure which is then connected > > directly to AC earth wire from the 240v outlet. I’ve also made a > > polycarbonate mounting bracket for the encoder housing so it’s > > insulated from the machine frame and terminated it’s screen to the > > same earth star bolt at the control panel end. > > > > I’m happy to have this now as best practice but it wasn’t the solution > > for the ghost pulses. > > > > Interestingly, I noticed that the earth cable coming from the spindle > > VFD screen made the encoder signal extremely noisy when it was in > > close proximity to the encoder cable. > > That smells like a ground loop. If disconnected at the star bolt, it > should be an open circuit, to ground or anything else. > > > Also, just for your info, the > > stepper motors or drivers create a lot of ‘white noise’ on the shop > > radio when they are holding or running so I guess they are chucking > > out a lot of high frequency noise. > > They do. > > I generally run my motor cables in shielded cabling. The stepper drivers > control the motor current by turning themselves on and off at an > ultrasonic frequeny we don't hear. If you can find "starquad" cabing in > a gage heavy enough it doesn't run warm at the motors current. It is > actualy the gold standard microphone cable, a top quality microphone > cable available in several gages, all VERY flexible, get the lowest gage > number Suzan has. 22 gage IIRC. Ground the shielding drain wire at the > star bolt, trim and insulate it at the motor end. > > > > It seems I have three options from here. > > > > 1) Change to the HPCL2631 opto isolators. > > > > 2) Change to the 74HC14 buffer. I think I will need resistance > > dividers with this as the max input in the datasheet suggests 6v so I > > will need to drop the 12v encoder signal to <6v (?) > > > > 3) Try the existing 74HC4050 buffer with resistance dividers. > > > > I’ve got some components on order so I guess whichever turns up first > > will be the first I’ll try. > > > > Seems like I also need to find myself some sort of oscilloscope on > > eBay! > > Digital storage will show you stuff that cannot be seen on an analogue > scope. Definitely worth the extra sheckles. Some of this stuff is at 100 > or more megahertz. And tends to be very dim on a analog scope. I > actually have 3, a 30 yo Hitachi 100 mhz dual trace analog, much better > than a tek of the same vintage, a 5 yo digital with the same specs you > can get for $300 or so today, and I just bought Siglents best, a 4 > trace, 350 mhz digital sampler. It also costs a down payment on a new > small car. Inheritances are handy. > > > I’ll keep you posted. > > Please. > > > Cheers, > > > > Pete > > > > > On 11 Oct 2021, at 21:11, Gene Heskett wrote: > > > > > > On Monday 11 October 2021 15:49:43 Chris Albertson wrote: > > >> You could be correct. High impedance is a recipe for noise. > > >> I had suggested a resistive divider just because it is simpler. > > >> But you are right about providing a ground path. A divider > > >> certainly would do that. If an opto is really needed then use a > > >> high-value resistor to ground to keep the line from floating and > > >> bleed
Re: [Emc-users] Fwd: Rogue Index Pulses
You could be correct. High impedance is a recipe for noise. I had suggested a resistive divider just because it is simpler. But you are right about providing a ground path. A divider certainly would do that. If an opto is really needed then use a high-value resistor to ground to keep the line from floating and bleed off static. I also don't like the idea of grounding the shield on the encoder end as it makes it impossible to know the path from encoder housing back to true Earth ground. It is "unanalyzable" (if such a word exists) Running the shield to star ground point makes it easy to verify it is correct. On Mon, Oct 11, 2021 at 12:29 PM John Dammeyer wrote: > I still believe the real problem is a very very high impedance circuit > that easily can capture electrostatic interference that occasionally is > high enough to cross the threshold of a logic 1 simulating an extra index > pulse. > > My favourite reference book for learning about this is: > Electromagnetic Compatibility Engineering but Henry W. Ott publishd by > Wiley. (c) 2009 ISBN 978-0-470-18930-6 > > https://www.google.com/search?client=firefox-b-d=ISBN+978-0-470-18930-6 > > Spend 6 months treating this as a EE course and you will learn what you > need from systems down to PC board design. > > One could use a resistor divider circuit with the existing hardware once > you know what the voltage level of the PushPull High is. > > If it's say 11V from the encoder powered at 12V for example then > 11V/0.01mA is 1100 Ohms. The CMOS threshold on the level translator was > 4.5V I believe so 1100 Ohms - 450 Ohms = 650 Ohms. One might even try 620 > Ohm and 470 Ohm as those are both standard 5% resistor values. > > Therefore replace the 1K resistor with the encoder index output connected > to a 620 ohm and 470 Ohm in series to ground. Tap off that junction so the > CMOS level translator sees the value across the 450 Ohm and ground which > should change between 4.5V and likely 0V or very close to that. > > Now you have a low resistance circuit (although high frequency impedance > might still be large) depending on capacitance and inductance of the cable. > > I still prefer the HCPL2631 approach but this is an experiment to try. > The point is to make the electrostatically coupled electrical noise > floating around find the easiest path to ground resulting in the lowest > impressed voltage on the wire. > > John Dammeyer > > > > -Original Message- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: October-11-21 11:07 AM > > To: phodg...@uk22.net; Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Fwd: Rogue Index Pulses > > > > One HUGE problem I see is that the halscope does not show the actual > > encoder pulses. The encoder has THREE outputs and each is a 50% duty > > cycle square wave and the A and B waves are 90 degrees out of phase. > > Halscope is not showing the encoder output. I suspect it is showing > > highly processed version of the encoder data and I suspect the processing > > is wrong. Can you make it show the actual input signals? > > > > Old analog scopes are dirt cheap. Even good ones made by HP or Tektronix > > are on eBay for around $100 > > > > The way to debug this is to take it apart and verify the parts all work > > then connect some of them together and verify the assembly works and > > continue until the entire system is assembled. This is how > > professional engineers work. In my experience building the full up > system > > then applying power rarely works except in very simple systems. Test > each > > part independently and then again after assembly each assembly step > > > > So for example, I'd place the encoder with a DC motor (or chuck it in a > > drill) and look at its output. Then attach the level converter and look > > again. Then put the encoder on the real spindle and look again. Make > one > > change at a time and test.Having a real scope means you don't need to > > have a computer to test. > > > > At this point we don't know where the error is. The encoder itself could > > be the problem.But I doubt that. May gues in now a software/config > > issue.I would keep that gues until I was able to see the raw encoder > > data. > > > > > > > > On Mon, Oct 11, 2021 at 2:19 AM Peter Hodgson < > peterjohnhodg...@gmail.com> > > wrote: > > > > > Hi Chris, > > > > > > I've added the encoder part number to the schmeatic and also added > > > datasheets and 'Halscope' screen shot to the webpage: > > > > > > > &
Re: [Emc-users] Fwd: Rogue Index Pulses
One HUGE problem I see is that the halscope does not show the actual encoder pulses. The encoder has THREE outputs and each is a 50% duty cycle square wave and the A and B waves are 90 degrees out of phase. Halscope is not showing the encoder output. I suspect it is showing highly processed version of the encoder data and I suspect the processing is wrong. Can you make it show the actual input signals? Old analog scopes are dirt cheap. Even good ones made by HP or Tektronix are on eBay for around $100 The way to debug this is to take it apart and verify the parts all work then connect some of them together and verify the assembly works and continue until the entire system is assembled. This is how professional engineers work. In my experience building the full up system then applying power rarely works except in very simple systems. Test each part independently and then again after assembly each assembly step So for example, I'd place the encoder with a DC motor (or chuck it in a drill) and look at its output. Then attach the level converter and look again. Then put the encoder on the real spindle and look again. Make one change at a time and test.Having a real scope means you don't need to have a computer to test. At this point we don't know where the error is. The encoder itself could be the problem.But I doubt that. May gues in now a software/config issue.I would keep that gues until I was able to see the raw encoder data. On Mon, Oct 11, 2021 at 2:19 AM Peter Hodgson wrote: > Hi Chris, > > I've added the encoder part number to the schmeatic and also added > datasheets and 'Halscope' screen shot to the webpage: > > > https://www.purbrookengineering.com/index.php?option=com_content=article=11 > > I don't have an oscilliscope so am relying on Halscope at the moment. > > Thanks, > > Pete > > > On 11/10/2021 02:58, Chris Albertson wrote: > > sorry there was no part number on the encoder. I figured it was the > generic > > Omron clone that is 5 to 12 volts. But there are two kinds that other > is > > 12 to 24. volts. See in the link below > > https://www.ia.omron.com/products/family/486/specification.html > > > > Is the encoder "open collector" or voltage output. If open colector > pull > > the outputs up to 5 volts. > > > > Placing an encoder part number of the schematic would help. > > > > Can you post a picture of "ghost pulses", if using an analog scope just > use > > you phone camera, most digitalscopes can save the scree to USB drive. > > Noise like this is best eliminated at the source > > > > > > On Sun, Oct 10, 2021 at 1:04 PM Peter Hodgson < > peterjohnhodg...@gmail.com> > > wrote: > > > >> Hi Chris, > >> > >> The encoder datasheet gives a voltage range of 10-30vdc so I don’t think > >> direct connection with 5vdc would be an option. > >> > >> Would a direct connection via voltage dividers eliminate the ‘ghost > >> pulses’ being picked up by the Pico board? > >> > >> The encoder won’t be running above 3,000 rpm. > >> > >> Pete > >> > >>> On 10 Oct 2021, at 19:00, Chris Albertson > >> wrote: > >>> Is your math right? Yes, I get 83 KHz for a 500 line encoder at > 10,000 > >>> RPM. Yes that is < 1% of the isolator's rated speed (of 10 MHz) > >>> > >>> But the schematic had an RC low pass filter in the encoder output that > >>> would have prevented the system from working above about 1000 RPM. I > see > >>> this is gone now. > >>> > >>> But why are these isolators even needed? Why not simply operate the > >>> encoder at 5 volts? > >>> > >>> Or if you must run the encoder at 12 volts, you can do level > translation > >>> with a resistor voltage divider. > >>> > >>> There is nothing yo isolate. The entire system runs on a common ground > >> and > >>> there is no high power devices in the schematic and I assume the entire > >>> system is in one building with no long (100 meter) cables > >>> > >>> If you are worried about accidents blowing up the controller, use > diodes > >> to > >>> shunt any transients. But really there are no inductive loads > >>> > >>> > >>> > >>> > >>> > >>> On Sun, Oct 10, 2021 at 10:25 AM John Dammeyer > > >>> wrote: > >>> > >>>>> From: Gene Heskett [mailto:ghesk...@shentel.net] > >>>> > >>>> > &g
Re: [Emc-users] Fwd: Rogue Index Pulses
sorry there was no part number on the encoder. I figured it was the generic Omron clone that is 5 to 12 volts. But there are two kinds that other is 12 to 24. volts. See in the link below https://www.ia.omron.com/products/family/486/specification.html Is the encoder "open collector" or voltage output. If open colector pull the outputs up to 5 volts. Placing an encoder part number of the schematic would help. Can you post a picture of "ghost pulses", if using an analog scope just use you phone camera, most digitalscopes can save the scree to USB drive. Noise like this is best eliminated at the source On Sun, Oct 10, 2021 at 1:04 PM Peter Hodgson wrote: > Hi Chris, > > The encoder datasheet gives a voltage range of 10-30vdc so I don’t think > direct connection with 5vdc would be an option. > > Would a direct connection via voltage dividers eliminate the ‘ghost > pulses’ being picked up by the Pico board? > > The encoder won’t be running above 3,000 rpm. > > Pete > > > On 10 Oct 2021, at 19:00, Chris Albertson > wrote: > > > > Is your math right? Yes, I get 83 KHz for a 500 line encoder at 10,000 > > RPM. Yes that is < 1% of the isolator's rated speed (of 10 MHz) > > > > But the schematic had an RC low pass filter in the encoder output that > > would have prevented the system from working above about 1000 RPM. I see > > this is gone now. > > > > But why are these isolators even needed? Why not simply operate the > > encoder at 5 volts? > > > > Or if you must run the encoder at 12 volts, you can do level translation > > with a resistor voltage divider. > > > > There is nothing yo isolate. The entire system runs on a common ground > and > > there is no high power devices in the schematic and I assume the entire > > system is in one building with no long (100 meter) cables > > > > If you are worried about accidents blowing up the controller, use diodes > to > > shunt any transients. But really there are no inductive loads > > > > > > > > > > > > On Sun, Oct 10, 2021 at 10:25 AM John Dammeyer > > wrote: > > > >>> From: Gene Heskett [mailto:ghesk...@shentel.net] > >> > >> > >> > >>> If the opto's are fast enough, you should be > >>> fixed. A 500 ppr encoder s/b fine. A 1024 or 1000 requires a faster > >>> opto. > >>> > >> > >> HI Gene, > >> Comparing the cheap far east BoB optos with the HCPL2621 is apples and > >> oranges. > >> > >> A 2500 line encoder still only creates 2500 pulses. The quadrature just > >> looks at two lines and 2 edges for 10,000 edges. But the max speed of > his > >> encoder is 10,000 RPM which in RPS is 166.7 and would result in 417kHz > >> which is 4% of the opto max 10 Mbps capabilities if the encoder was > 2500 > >> line. > >> > >> But his is 500 line so at 166.7 RPS is 83kHz which is 0.8% of opto > >> capabilities. > >> > >> > >> John > >> > >> > >> > >> > >> ___ > >> Emc-users mailing list > >> Emc-users@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-users > >> > > > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Rogue Index Pulses
Is your math right? Yes, I get 83 KHz for a 500 line encoder at 10,000 RPM. Yes that is < 1% of the isolator's rated speed (of 10 MHz) But the schematic had an RC low pass filter in the encoder output that would have prevented the system from working above about 1000 RPM. I see this is gone now. But why are these isolators even needed? Why not simply operate the encoder at 5 volts? Or if you must run the encoder at 12 volts, you can do level translation with a resistor voltage divider. There is nothing yo isolate. The entire system runs on a common ground and there is no high power devices in the schematic and I assume the entire system is in one building with no long (100 meter) cables If you are worried about accidents blowing up the controller, use diodes to shunt any transients. But really there are no inductive loads On Sun, Oct 10, 2021 at 10:25 AM John Dammeyer wrote: > > From: Gene Heskett [mailto:ghesk...@shentel.net] > > > > > If the opto's are fast enough, you should be > > fixed. A 500 ppr encoder s/b fine. A 1024 or 1000 requires a faster > > opto. > > > > HI Gene, > Comparing the cheap far east BoB optos with the HCPL2621 is apples and > oranges. > > A 2500 line encoder still only creates 2500 pulses. The quadrature just > looks at two lines and 2 edges for 10,000 edges. But the max speed of his > encoder is 10,000 RPM which in RPS is 166.7 and would result in 417kHz > which is 4% of the opto max 10 Mbps capabilities if the encoder was 2500 > line. > > But his is 500 line so at 166.7 RPS is 83kHz which is 0.8% of opto > capabilities. > > > John > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Rogue Index Pulses
Using the opto ioslators as voltage translators should work. Note the data sheet has the words "not recommended for new design" on it. There is likely a newer part. But maybe you have these on hand already? These isolators will do 10 Mbits/second, likely fast enough. The bottleneck is the RC filter made using the 1K resistor and 0.1 uF cap (time constant of 0.0001 second) One other idea. Can you run the encoder from a 5 volt power supply? Then there is no need for level conversion At 1,000 RPM and 500 pulses per revolution you get 8333 pulses per second. Your low pass filter might be a little too aggressive. But I don't know the max shaft RPM. But do check this. On Sat, Oct 9, 2021 at 12:39 PM Gene Heskett wrote: > On Saturday 09 October 2021 15:22:29 Peter Hodgson wrote: > > > Hi John, > > > > Thanks for sharing the drawing. Yes, I would like to get things > > properly drawn when I get the time. > > > > In the mean time... would you mind having a quick look at the > > revised schematic on the following link to check I have placed the > > capacitors you suggested in the correct place when using the HCPL2631 > > ?? > > > > https://www.purbrookengineering.com/index.php?option=com_content= > >article=11 > > > > > > I notice on your drawing you have 'Earth' and FGND. What's the > > difference? > > > > Pete > > > That looks better, as long as the opto's are fast enough. Verify that > with a o-scope. And the halscope is not fast enough, use a real one. But > please don't use yellow on white for the signal wires, they disappear in > the pdf. :( > > > On 09/10/2021 18:18, John Dammeyer wrote: > > > Hi Peter, > > > I find it helps to draw a wiring diagram of where things are wired. > > > That’s different from a schematic which just shows how things are > > > connected. > > > > > > For example, I've attached the old schematic of my mill wiring > > > (before I removed the two HP-UHU drives and wired in the quadrature > > > encoder for the spindle into the second Chinese BoB). Note to self: > > > Really must update now that I have AC servos on everything. > > > > > > I have colour codes on there for wiring. But it does not show the > > > actual wiring paths. For example the 24V comes out of the Power > > > Supply and splits in two directions. One to the DIN Rail for the > > > MiniBoB->STMBL drive, the STMBL drive and 24V-16V UHU power supply. > > > From there another DIN rail set of terminal blocks to run power out > > > to the Bergerda Spindle Interface and now the Bergerda interface > > > boards. > > > > > > But the drawing doesn't actually show the DIN rails or wire lengths > > > so there's no way to see if the 24V system has ground loops etc. > > > Not from the Schematic. > > > > > > So I still have to create a new 'mechanical' wiring diagram that > > > shows where rather than how things are connected. > > > > > > John > > > > > >> -Original Message- > > >> From: Peter Hodgson [mailto:peterjohnhodg...@gmail.com] > > >> Sent: October-09-21 1:38 AM > > >> To: Enhanced Machine Controller (EMC) > > >> Subject: Re: [Emc-users] Fwd: Rogue Index Pulses > > >> > > >> Thanks Chris, yes, I will change the shielding ground to the > > >> controller end and isolate the case. > > >> > > >> Is it best practice to run the 5v and 12v �grounds� back to the > > >> original 24v power supply 0v? > > >> > > >>> On 9 Oct 2021, at 00:53, Chris Albertson > > >>> wrote: > > >>> > > >>> ?I'll say what Gene said. I'd bet on a groubing issue. The > > >>> schmatic has an error, a groubd wire ismissing and also shows > > >>> random crossing grounds. Run everything back to the power supply > > >>> minus post. > > >>> > > >>> Also you be much better if the sheild was grounded to the > > >>> controller end. Never us a machine and it's mounting bolts as > > >>> grounds. > > >>> > > >>> > > >>> -- > > >>> > > >>> Chris Albertson > > >>> Redondo Beach, California > > >>> > > >>> ___ > > >>> Emc-users mailing list > > >>> Emc-users@lists.sourceforge.net > > >>> https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Fwd: Rogue Index Pulses
I'll say what Gene said. I'd bet on a groubing issue. The schmatic has an error, a groubd wire ismissing and also shows random crossing grounds. Run everything back to the power supply minus post. Also you be much better if the sheild was grounded to the controller end. Never us a machine and it's mounting bolts as grounds. -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] VFD Recomendation
ed for 20 years so I'm > > getting rusty. And despite the fact that I have now outlived all 3 > > wives, and 3 of the children the first one gave me, I still think I've > > had a good ride for what will be 87 years on 10/04. > > > > [...] > > > > -- > > "There are four boxes to be used in defense of liberty: > > soap, ballot, jury, and ammo. Please use in that order." > > -Ed Howdershelt (Author, 1940) > > If we desire respect for the law, we must first make the law respectable. > > - Louis D. Brandeis > > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > > -- > Addressee is the intended audience. > If you are not the addressee then my consent is not given for you to read > this email furthermore it is my wish you would close this without saving or > reading, and cease and desist from saving or opening my private > correspondence. > Thank you for honoring my wish. > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] more dc servo questions
t duration that there is no > detectable heating of the caps during the few milliseconds of the > overvoltage duration. And its been doing it for 5+ years with zero > failures. From 3k revs fwd to 3k revs reverse is 350 to 400 > milliseconds. > > > Regards, > > > > John > > > Thanks John. > > Cheers, Gene Heskett > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] more dc servo questions
SIn waves are there own derivatives and integrals, except for phase. So a sine wave acceleration works if the gaol is a sine wive velocity curve. But I assume the goal here is maximum change in velocity. You get that by running the motor is full reverse. I doubt we are seing back EMF because the worm cn not be back driven by a wheel. My guess about the root cause if the power supply current spike is a shoot-through in the double h-brige because the direction owas reserved instantly with no though to the gate driver rise and fall times. (I actualy hope it is this because the fix then is easy and can be done on software. If I'm wrong th fix is harder. On Wed, Sep 22, 2021 at 11:51 AM dave engvall wrote: > Hi, > > Fully recognizing that nothing is impossible to the person that doesn't > have to make it work. With that proviso I will open > mouth and insert foot. To wit: a sine for acceleration should give a > more gentle startup and approach to end point. Of course implementation > is left as an exercise for the student. ;-) > Now I'll shut up and go away. > > Dave > > On 9/22/21 11:28 AM, Chris Albertson wrote: > > So you are driving it with something like this? > > amazon.com/HiLetgo-BTS7960-Driver... > > < > https://www.amazon.com/HiLetgo-BTS7960-Driver-Arduino-Current/dp/B00WSN98DC/ref=asc_df_B00WSN98DC/?tag=hyprod-20=df0=198090929431==g=9099973595926304313c===9031026=pla-403676040447=1 > > > > > > One possible mistake is if even for a short time the both enable inputs > are > > active you will get "shoot through" and short the power supply.Can > you > > verify this NEVER haens using a scope or logic analyser? Do not > > bother with a software "hal scope" You need to look at that is actually > on > > the wires with a device that samples the actual metal pins. Real pins > have > > riseand fall times and they can't intersect. > > > > I would also just use one PWM generator and connet it to both LPWM and > > RPWM.A safer why to handle L-EN and R_EN is to make 100% certain that > > they go through a state where BOTH ARE OFF. Don't flip them instantly as > > there is a finite rise and decay time. There needs to be soome number of > > microseconds where both are off. > > > > I bet LCNC flips both L-EN and R_EN on eht same software cycle. If so > then > > the supply is shorted via the controller board. > > > > There boards are cheap and another way to fix this is to used a boaed > with > > a "forward/Reverse" pin rather then two pins sothere is no chance of > > enabling both pins. > > > > Or using a "smart" controller with a serial interface where the speed and > > direction are sent as a digital command message. Then the acceletaion > and > > such is handled by the controller. > > > > The sounds like an short in the h-bridge to me but I could be wrong. You > > can test this with a digital scope. on the control pins. > > > > On Wed, Sep 22, 2021 at 10:21 AM Gene Heskett > wrote: > > > >> On Wednesday 22 September 2021 11:44:51 John Figie wrote: > >> > >>> Gene, > >>> > >>> I would like to better understand your problem and have been thinking > >>> about this. I have some questions. > >>> > >>>>> I don't think the fact that there is a worm gear matters. The > >>>>> problem, I bet is the large inertia of the system. > >>> Hmm I am not sure about the inertia. I think if you have a large gear > >>> reduction then the inertia reflected across the gears should appear > >>> low from the motors point of view. > >>> > >>>>> Aside from proper tuning of the PID gains you could change the > >>>>> system to use a nested or "cascade" PID. THis allows the velocity > >>>>> setpoint to be controlled by the position error > >>>>> see the section "cascade" in the wiki article > >>>>> https://en.wikipedia.org/wiki/PID_controller > >>>>> > >>>>> I think you want a very fast loop i=for the inner PID. LIkely it > >>>>> wouldbe in external hardwarelike a microcontroller or FPGA. > >>>>> (does MESA > >>>> do > >>>> > >>>>> the PID algorithm in the FPGA? It should.) > >>>>> > >>>>> > >>>>> > >>>>> On Sun, Sep 19, 2021 at 3:39 PM Gene Heskett > >>>>> > >>>> wrote: > >>>>>>
Re: [Emc-users] more dc servo questions
has its own version. > > Null in this context is when the encoder output equals commanded > position. > > > > >> arcseconds early. I can't allow it to use reverse to stop as the > > > >> motor seems to be a near short circuit then, crowbarring the > > > >> power supply, > > > > How do you prevent the PWM from reversing the voltage on the motor? Is > > the motor driven from a PWM in only one polarity? How is the PWM and > > the switches arranged? > > Control is by a LCNC PID, fed by the lcnc encoders position output > feeding the PID feedback, with a pwmgen running in mode 2 where it has > two pulse width modulated outputs, one fwd, one reverse, which are fed > to a BTS 7960 board containing 2 of the Infineon half bridges for a full > bridge control. This IC is rated at 45 volts, 43 amps. Fed by a 24 volt > supply good for nearly 20 amps, intended for automotive seat position > controls, it runs this motor with zero heating. But if it overshoots the > commanded position and goes into reverse to bring the motor back by more > than a 1 or 2 % drive while the motor is still coasting fwd, it will > crowbar the psu, getting it hot instantly, causeing the psu to do a shut > down until it has cooled for 2 or 3 minutes. Even then, no heat in the > BTS IC's. > > The net result is that I must limit its cuising speed and accelleration > in order for it to coast to a stop using its own friction while avoiding > the use of reverse to stop, usually a quite small fraction of a degree > early. And that is about 10 to 15% of the speed it can move with the > full 24 volts applied. > > > Regards, > > > > John Figie > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > Cheers, Gene Heskett > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] more dc servo questions
On Wed, Sep 22, 2021 at 8:49 AM John Figie wrote: > Gene, > > I would like to better understand your problem and have been thinking about > this. I have some questions. > > > > > > > > I don't think the fact that there is a worm gear matters. The > problem, I > > > bet is the large inertia of the system. > > > Hmm I am not sure about the inertia. I think if you have a large gear > reduction then the inertia reflected across the gears should appear low > from the motors point of view. > An ideal gear reduction is a torque multipliers lets say a 4:1 system. But X/4 is still large if X is large and small if X is small. THe mass in the system still maters. > > > > > So you have a 2 worm gear reduction. Where in this system is the encoder > mounted? What are the encoder counts per rev of the encoder? > What are the motor characteristics? > I also suspect this is the root of the problem.It might be misdesigned. That said, while I know how to calculate forces in a perfect system (like a belt reduction) I'm stumped when it comes to worm gears because they are asymetric in that they can not be back driven due to friction. I don't know how to calculate the load on the motor when the motor is opposing ditection of movment. There might be (?) large frictional loads. We also don't know how the driver electronics works. > > Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] more dc servo questions
Every time I have implemented PID based motor control the PID loop runs on a low-cost microcontroller and then the main computer sends the PID a series of set-points. Micros work well as most of them nowadays have hardware quadrature decoders and can handle over a million lines per second. The $4 Raspberry Pi Pico is at least an order of magnitude faster than needed for this use case. You don't need an FPGA as a $4 micro is already overkill. That said, LCNC just uses very powerful computers so it should be able to run nested PID loops. If you make velocity (*velocity*, not acceleration or motor current) a function of the position error then the motor wil slow down as it gets closer. There is also likely a good way to use feed forward with just a single PID loop. On Mon, Sep 20, 2021 at 9:05 AM dave engvall wrote: > Hi all, > To the best of my knowledge any version of the lcnc pid is in software. > OTOH mesa does offer SOFTDMC which does run in/on mesa fpga products. > However, that breaks the core concept of multiple vendors and real ease > of user modification of the software although there are a few in this group > that could fiddle the code for an fpga. In addition softdmc needs all > the support written to wrap around it; interp, task, etc. Years ago I > head a rumor tht someone had tried to do it and abandoned the effort. If > I recall correctly it does offer something besides trap accel, controls > jerk and runs on a 50 us clock. > The manual is available online. > Dave > > On 9/19/21 5:12 PM, Chris Albertson wrote: > > Is the PID controller running on the PC under LCNC or is this an external > > PID controller you built? > > > > What are the numbers, the loop period, motor speed and numer of encoder > > lines. and so on? > > > > I don't think the fact that there is a worm gear matters. The problem, I > > bet is the large inertia of the system. > > > > Aside from proper tuning of the PID gains you could change the system to > > use a nested or "cascade" PID. THis allows the velocity setpoint to be > > controlled by the position error > > see the section "cascade" in the wiki article > > https://en.wikipedia.org/wiki/PID_controller > > > > I think you want a very fast loop i=for the inner PID. LIkely it > > wouldbe in external hardwarelike a microcontroller or FPGA. (does MESA > do > > the PID algorithm in the FPGA? It should.) > > > > > > > > On Sun, Sep 19, 2021 at 3:39 PM Gene Heskett > wrote: > > > >> Greetings all; > >> As most of you know, I built a servo from scratch for a BS-1. > >> > >> But I am not at all happy with its performance. > >> It is a motor with a worm output, driving the worm of the bs-1. And it > >> has an A/B quad encoder in it. > >> > >> But I must rather severely limit its run speed because the PID doesn't > >> see the null coming near fast enough to slow it and stop a couple > >> arcseconds early. I can't allow it to use reverse to stop as the motor > >> seems to be a near short circuit then, crowbarring the power supply, > >> causing a 2 to 3 minute dead time for the 450 watt psu to cool, so I > >> must limit its speed to disallow its use of reverse to accomplish the > >> stop. With 2 worms in series the scale for a 360 degree full turn is of > >> coarse huge. If it could be throttled to a gradual stop by anticipating > >> the coming null, I could probably run it 2 to 4x faster for a cruising > >> speed. > >> > >> So my question is, which of the pid inputs would if raised, better > >> anticipate the approaching null, slowing it over the last 5 angular > >> minutes of a turn, ideally to a stop within an arc-second of the command > >> from the gcode? > >> > >> Thanks all. > >> > >> Cheers, Gene Heskett > >> -- > >> "There are four boxes to be used in defense of liberty: > >> soap, ballot, jury, and ammo. Please use in that order." > >> -Ed Howdershelt (Author) > >> If we desire respect for the law, we must first make the law > respectable. > >> - Louis D. Brandeis > >> Genes Web page <http://geneslinuxbox.net:6309/gene> > >> > >> > >> ___ > >> Emc-users mailing list > >> Emc-users@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/emc-users > >> > > > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] more dc servo questions
Is the PID controller running on the PC under LCNC or is this an external PID controller you built? What are the numbers, the loop period, motor speed and numer of encoder lines. and so on? I don't think the fact that there is a worm gear matters. The problem, I bet is the large inertia of the system. Aside from proper tuning of the PID gains you could change the system to use a nested or "cascade" PID. THis allows the velocity setpoint to be controlled by the position error see the section "cascade" in the wiki article https://en.wikipedia.org/wiki/PID_controller I think you want a very fast loop i=for the inner PID. LIkely it wouldbe in external hardwarelike a microcontroller or FPGA. (does MESA do the PID algorithm in the FPGA? It should.) On Sun, Sep 19, 2021 at 3:39 PM Gene Heskett wrote: > Greetings all; > As most of you know, I built a servo from scratch for a BS-1. > > But I am not at all happy with its performance. > It is a motor with a worm output, driving the worm of the bs-1. And it > has an A/B quad encoder in it. > > But I must rather severely limit its run speed because the PID doesn't > see the null coming near fast enough to slow it and stop a couple > arcseconds early. I can't allow it to use reverse to stop as the motor > seems to be a near short circuit then, crowbarring the power supply, > causing a 2 to 3 minute dead time for the 450 watt psu to cool, so I > must limit its speed to disallow its use of reverse to accomplish the > stop. With 2 worms in series the scale for a 360 degree full turn is of > coarse huge. If it could be throttled to a gradual stop by anticipating > the coming null, I could probably run it 2 to 4x faster for a cruising > speed. > > So my question is, which of the pid inputs would if raised, better > anticipate the approaching null, slowing it over the last 5 angular > minutes of a turn, ideally to a stop within an arc-second of the command > from the gcode? > > Thanks all. > > Cheers, Gene Heskett > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > Genes Web page <http://geneslinuxbox.net:6309/gene> > > > _______ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Minimally printed rotary. Initial torque testing..
Yes, I agree with the words below. I better way to measure torque is to make a drumwrape a cable around it and lift a bucket. Run the drum at different PPMs and load the bucket until the drum slows from it's programmed RPM. THis will allow you to plot an RPM vs torque curve. What you have with a torque wrench is only one point on that curve. For a stepper motor this is the "best case" point on the curve, all other points would be lower. Static torque is important but mostly these drives are meant to move. Along to same line. Measuring longevity under zero load is not as interesting as if measured at say 50% of max torque and running it in both directions. I know, testing takes time. That said, I'm watching these experiments closely because I want to make some of these reduction drives myself. I meed at least 100:1 reduction in a very smallspace. I want to use a motor like one of these (but likely notthis exact one) amazon.com/Hobbypower-1000kv-Brushless... <https://www.amazon.com/Hobbypower-1000kv-Brushless-Multicopter-Quadcopter/dp/B00E7LG85O> The "1000 KV" means 1000 RPM per volt and it comes with a digital controller. I Would mount the motor almost inside the flex cut of the harmonic drive and maek an effective 10KV motor system On Mon, Sep 13, 2021 at 10:26 AM John Dammeyer wrote: > Nice update. > My math works out as follows: > 35 lb-ft is 35 * 12"/ft * 16 oz/lb = 6720 oz-in. > You said the ratio is 100:1 so that means to get 6720 you only input 67.2 > oz-in. That's about 1/3 of what you suspect your motor can do. > > Something is wrong with this picture then. Especially since you were > effectively using static torque rather than the much lower torque when the > motor is turning at say 400 RPM. With micro-stepping you also get a > reduction in full step torque because the max current is now 0.707 x full > torque. And without closed loop encoder feedback you can quite easily have > to micro-step a number of times before you see physical motion. > > None of that matters other than that you have shown your mechanics can > handle 35 lb-ft which is all in all pretty cool ! > > John > > > > > > -Original Message- > > From: Sam Sokolik [mailto:samco...@gmail.com] > > Sent: September-12-21 5:50 PM > > To: Enhanced Machine Controller (EMC) > > Subject: [Emc-users] Minimally printed rotary. Initial torque testing.. > > > > Small update > > > > https://youtu.be/eW1GGI55Epc > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] jerk control
In the general case there may be no closed form solution so numeric integration is the only possible solution.I don't think there is any other way to do it other than numeric integration except to require the user to supply a function for closed form integrals. But for a 3-axis mill that uses trivial kinematics there is an easy closed form integral. So the problem is very different depending on what kind of machine you want to control.For people here, almost all are using trivial kinematics. The problem is also VERY different based on another decision. Do you want only simple jerk limiting or do you want an optimal path? In other words, do you simply take your foot off the gas to avoid going over the speed limit or do you want to drive to work in the absolute minimum time but without speeding. Those are different problems. The first one is easy to do, the second takes quite a lot of effort. Jerk limited, optimal paths for non trivial kinematics is a very hard problem. Putting a simple jerk limit on a simple machine is easy, Marlin does this on my 3D printer and it runs on an Aruino. On Sun, Aug 29, 2021 at 10:07 AM Curtis Dutton wrote: > I'm not sure about all this yet, brainstorming here... > > After looking at TinyG code for handling jerk limitation in the joint > control it appears that they are using the forward physics equations and > numerically integrating to avoid violations. > > Since numerical integration needs to be used for jerk equations (as well as > snap crackle and pop...) The forward kinematic equations will need to be > fed into the integrator for non trivial kinematics. > > The integrator should be pluggable as well. We should find an open source > library for this as numerical integration methods are always advancing and > high performance integration is not easy to implement. > > > > > > On Sun, Aug 29, 2021 at 11:55 AM Andy Pugh wrote: > > > > > > > > On 29 Aug 2021, at 11:40, Alexander Brock > > wrote: > > > > > > The idea can be implemented in C and for simple cases like 1D funktions > > > it should be fairly straight-forward. > > > > My point is that the kinematics modules already exist, and not all of > > them are under our control. Various users’ machines are out there with > > their own custom kinematics. > > Also, AFAIK if they are kernel modules then they have to be written in C. > > > > Hence my suggestion of calling the kinematics function repeatedly to > > perform simple numerical differentiation. > > > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] jerk control
The software in the link below is very widely used in industry. It is used inside MoveIt which in turn is driving many complex machines. https://moveit.ros.org On Fri, Aug 27, 2021 at 11:52 AM Curtis Dutton wrote: > That thread leads to this. https://githubhelp.com/pantor/ruckig > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] jerk control
I think the reason for a jerk limit is to limit changes in the force applied to the machine's structure.A high constant acceleration applies a constant force and hence constant deflection. "A constant deflection" is a fancy way to way "no motion". But any jerk means the force and deflection is non-constant, a fancy way to so the machine's frame is moving. This matters on 3D printers and other lightly constructed machines a lot more than on a CNC mill. On Mon, Aug 23, 2021 at 1:45 PM Todd Zuercher wrote: > But you need to look at why you have your jerk limit is set to what it is > in the first place. Is it not there so that the motor (joint) can > accelerate in a controlled fashion as quickly as possible within its > capabilities smoothly? Therefore the vector sum is also the vector sum of > the combined joints capabilities, so shouldn't that also be the same amount > higher? > > Todd Zuercher > P. Graham Dunn Inc. > 630 Henry Street > Dalton, Ohio 44618 > Phone: (330)828-2105ext. 2031 > > -Original Message- > From: Chris Albertson > Sent: Monday, August 23, 2021 3:39 PM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] jerk control > > [EXTERNAL EMAIL] Be sure links are safe. > > On Mon, Aug 23, 2021 at 7:16 AM Todd Zuercher > wrote: > > > To my lay persons eyes I would think it would be enough to jerk limit > > in joint space. The limiting in Cartesian space would then take care > > of itself. > > > > Let's work a simple example to see. Assume jerk is limited to 1 meter per > second cubed. Assume the machine is just 2D, X and Y pen plotter. If > both axes are running at the same speed and the current X and Y jerk is at > the 1m/s^3 limit then the pen (Cartesian space) has jerk equal to 1.414 > (square root of two) which is over the limit. > > Controllering only joint limits can't work if the cartesian motion can be > expressed as the vector sum of joint motions because the sum of two vacters > can have greater magnitude then either of the two joint vectors > > Simply not bumping into a limit is not really hard but it becomes a hard > problem when you want to move as fast as possible, but no faster. You can't > do it without doing a full simulation and numeric optimization. > > > > > I think this is true on a 3-axis machine where all the axes are linear and > orthogonal. Basically wherever "trivial kinematics" applies. > > > > > > > > As to the jogging question, does it matter? Why would jogging have to > > have the same acceleration limits as planned motion? Set the jogging > > limits to something safe and conservative that won't matter if they > > are jerk limited. > > > > Todd Zuercher > > P. Graham Dunn Inc. > > 630 Henry Street > > Dalton, Ohio 44618 > > Phone: (330)828-2105ext. 2031 > > > > -Original Message- > > From: andy pugh > > Sent: Monday, August 23, 2021 4:50 AM > > To: Enhanced Machine Controller (EMC) > > > > Subject: Re: [Emc-users] jerk control > > > > [EXTERNAL EMAIL] Be sure links are safe. > > > > On Mon, 23 Aug 2021 at 04:40, Chris Albertson > > > > wrote: > > > > > Actually for a machine tool, why not run the simulation off-line and > > > use as much time and computer power as it takes. > > > > Feed-override? > > > > Do you allow infinite jerk on abort? You might think that is an easy > > question, except that continuous jog is implemented as a move to the > > limit that is aborted on key release. > > > > And, do we need to jerk-limit in joint space or cartesian space, or both? > > > > -- > > atp > > "A motorcycle is a bicycle with a pandemonium attachment and is > > designed for the especial use of mechanical geniuses, daredevils and > lunatics." > > - George Fitch, Atlanta Constitution Newspaper, 1912 > > > > > > ___ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > > > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flist > > s.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Cto > > ddz%40pgrahamdunn.com%7Cdd7544f442b14a481c5d08d9666dccd5%7C5758544c573 > > f47cebee96c3e0806fb43%7C0%7C0%7C63765382037937%7CUnknown%7CTWFpbGZ > > sb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3 > > D%7C3000sdata=2OcIVDtgOdrHp0UmetoWxlAJ3mdvi%2FoT5hRpOH3Oc4Q%3D > > p;reserved=0 > > > > > > _______ > > Emc-user
Re: [Emc-users] jerk control
On Mon, Aug 23, 2021 at 7:16 AM Todd Zuercher wrote: > To my lay persons eyes I would think it would be enough to jerk limit in > joint space. The limiting in Cartesian space would then take care of > itself. > Let's work a simple example to see. Assume jerk is limited to 1 meter per second cubed. Assume the machine is just 2D, X and Y pen plotter. If both axes are running at the same speed and the current X and Y jerk is at the 1m/s^3 limit then the pen (Cartesian space) has jerk equal to 1.414 (square root of two) which is over the limit. Controllering only joint limits can't work if the cartesian motion can be expressed as the vector sum of joint motions because the sum of two vacters can have greater magnitude then either of the two joint vectors Simply not bumping into a limit is not really hard but it becomes a hard problem when you want to move as fast as possible, but no faster. You can't do it without doing a full simulation and numeric optimization. I think this is true on a 3-axis machine where all the axes are linear and orthogonal. Basically wherever "trivial kinematics" applies. > > As to the jogging question, does it matter? Why would jogging have to > have the same acceleration limits as planned motion? Set the jogging > limits to something safe and conservative that won't matter if they are > jerk limited. > > Todd Zuercher > P. Graham Dunn Inc. > 630 Henry Street > Dalton, Ohio 44618 > Phone: (330)828-2105ext. 2031 > > -Original Message- > From: andy pugh > Sent: Monday, August 23, 2021 4:50 AM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] jerk control > > [EXTERNAL EMAIL] Be sure links are safe. > > On Mon, 23 Aug 2021 at 04:40, Chris Albertson > wrote: > > > Actually for a machine tool, why not run the simulation off-line and > > use as much time and computer power as it takes. > > Feed-override? > > Do you allow infinite jerk on abort? You might think that is an easy > question, except that continuous jog is implemented as a move to the limit > that is aborted on key release. > > And, do we need to jerk-limit in joint space or cartesian space, or both? > > -- > atp > "A motorcycle is a bicycle with a pandemonium attachment and is designed > for the especial use of mechanical geniuses, daredevils and lunatics." > - George Fitch, Atlanta Constitution Newspaper, 1912 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Femc-usersdata=04%7C01%7Ctoddz%40pgrahamdunn.com%7Caaabdbd2b59a4e0ad88708d966134695%7C5758544c573f47cebee96c3e0806fb43%7C0%7C0%7C637653055682515186%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000sdata=iD2Nxd21YC5aKx8hByW2ScEBuMjM9%2FXJCwo05FHb%2BLU%3Dreserved=0 > > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
