Re: [Emc-users] Installing From Debian?
Just to chime in here Rod Webster has been doing a bit of work here with fixing error finishing read signals. The realtek drivers that come with the buster iso are not very good and we get a error finishing read error. This is with a good 4us latency test. The network card is the issue now. So mesa card watch dog hits and throws machine into estop. Real problem. Been working 3 days on it. And just now getting back to machining. 2 solutions. Rod wrote a doc that steps you through how to install bookworm and the new 6.1 RT kernel. That has been running for 3 days on a massive gcode file now with a mesa 7i92m connected and no errors. This is with 2.9linuxcnc. so pretty good solution for 2.9 if you need it. This is on a 2core jt1800 CPU. Only slight issue is the pc runs slower on bookworm in general. But it's only a industrial cheap 2core pc from china. 2nd solution is to use a old Linux mint 19.2 iso I have that someone made. With 2.8 linuxcnc and 4.19rt kernel. That has been running 24hrs with same big file no issues. This is my go to for 8 machines now and it just works Much faster response than bookworm and no issues so far either. And drivers just work on mint lol. Which is nice. But the error finishing read issue is pretty wide spread and a bit of a problem. Anyway back to running machines sing out if you get problems. Hopefully this helps. On Thu, 6 Jul 2023, 13:59 Todd Zuercher, wrote: > Any one installed Debian 12 and running Linuxcnc on a machine? Since > Linuxcnc is now in Debian Stable, I was wondering if anyone cared to voice > their opinions about installing and running Linuxcnc that way. > > I have a machine that I am starting to refit and I am considering trying > to install that way rather than using one of the more obsolete Linuxcnc > ISOs. > > ___ > 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
Re: [Emc-users] Installing From Debian?
I downloaded the debian iso (the big 64bit version) - picked cinnamon as
the desktop environment..
apt-get installed linuxcnc-uspace and was up and running within a few
minutes after copying my configs..
I see the future and it is looking awesome..
great work guys!
sam
On Tue, Jul 11, 2023 at 8:50 AM Todd Zuercher wrote:
> Here is a link to the instructions of how to correct a borked Debian 12
> installation from one of the live amd64 ISOs.
> https://forums.debian.net/viewtopic.php?t=155245
>
> You need to perform 3 command line commands (with root or sudo privileges.)
> # rm
> /etc/{initramfs/post-update.d/,kernel/{postinst.d/,postrm.d/}}z50-raspi-firmware
> # apt purge raspi-firmware
> # rm -rf /boot/firmware
>
> I tested this and it does seem to get things working correctly again. And
> then I could install the real-time or other kernels without issue.
>
> Todd Zuercher
> P. Graham Dunn Inc.
> 630 Henry Street
> Dalton, Ohio 44618
> Phone: (330)828-2105ext. 2031
>
> -Original Message-
> From: Todd Zuercher
> Sent: Monday, July 10, 2023 4:21 PM
> To: Enhanced Machine Controller (EMC)
> Subject: Re: [Emc-users] Installing From Debian?
>
> [EXTERNAL EMAIL] Be sure links are safe.
>
> Karl,
>
> Have you tried installing Debian using one of the Bookworm live ISOs? I
> have, and the Mate version I downloaded on 3/30/2023 is definitely borked.
> I don't remember and didn't take notes of the actual details of the
> problems. (and I've since wiped that install.) It would run live fine,
> install and run fine, install new software from the repos fine, but could
> not install a new kernel via apt-get or Synaptic (same thing really). I
> didn't' bother putting much effort into finding or fixing the problem.
>
> I have it installing again on a spare old harddrive to check out exactly
> what the problems were.
>
> The system I installed on was an older Lenovo desktop with an i5 (not sure
> the exact model.) I don't have a rbpi or any other arm devices.
>
> Todd Zuercher
> P. Graham Dunn Inc.
> 630 Henry Street
> Dalton, Ohio 44618
> Phone: (330)828-2105ext. 2031
>
> -Original Message-
> From: Karl Schmidt
> Sent: Monday, July 10, 2023 2:39 PM
> To: emc-users@lists.sourceforge.net
> Subject: Re: [Emc-users] Installing From Debian?
>
> [EXTERNAL EMAIL] Be sure links are safe.
>
> @Todd -
>
> Not likely a Debian problem - I would have to see his
> /etc/apt/sources.list to know what is going on.
>
> To upgrade you also need something like this in /etc/apt.conf.d/local
>
> APT::Default-Release "/^bookworm(|-security|-updates)$/";
>
> He says this is on an old MacBook (A1278) - that is supposed to be a
> core-i5 - thus amd64.
>
> I don't think the 'Raspberry Pi Desktop' is part of Debian - so his
> sources.list is probably wacked.
>
> Other things to look at:
> # cat /etc/issue
> # uname -a
>
>
> From a normal Debian system with these lines in sources.list:
> deb http://ftp.us.debian.org/debian/ bookworm main contrib non-free deb
> http://deb.debian.org/debian/ bookworm main contrib non-free
> non-free-firmware deb http://security.debian.org/debian-security
> bookworm-security main contrib non-free non-free-firmware deb
> http://deb.debian.org/debian/ bookworm-updates main contrib non-free
> non-free-firmware
>
> # wajig listall linuxcnc
> linuxcnc-doc-de motion controller for CNC machines and robots
> (German documentation)
> linuxcnc-doc-en motion controller for CNC machines and robots
> (English documentation)
> linuxcnc-doc-es controlador de movimiento para máquinas CNC y
> robots (Español).
> linuxcnc-doc-fr motion controller for CNC machines and robots
> (French documentation)
> linuxcnc-doc-zh-cnmotion controller for CNC machines and robots
> (Chinese
> linuxcnc-uspace motion controller for CNC machines and robots
> linuxcnc-uspace-dev PC based motion controller for real-time Linux
>
> The only drawback to running a pure Debian system is it 'tends to be VERY
> conservative' - so the big advantage of running pure Debian is it 'tends to
> be VERY conservative' - something I want for running machine-control or
> public facing systems.
>
> I really want to praise the guys that did all the heavy lifting to get
> linuxcnc into Debian - not a small task - (the software has to survive all
> sorts of probes by the "Lintian" a comprehensive package checker - Debian
> Policy violations and violations of various sub-policies, best practices,
> common mistakes, and problems that maintainers like to catch before
> uploads. )
>
> I'm sure the code will be picked up by the manny distributions that feed
> off of Debian. The more eyes on the code - the more users - the more
> polished it becomes.
>
>
>
>
> --
>
>
> Karl Schmidt EMail k...@lrak.net
> 3209 West 9th Street Ph (785) 841-3089
>
Re: [Emc-users] Drilling holes in the back of a stepper motor
On 7/30/23 02:29, John Dammeyer wrote: Gene, I think you are misreading the data sheet. The 115K baud is for setting up the unit. Ie. Pulses per rev. Once that's done the serial isn't used anymore. That didn't stick out very well in my skim read, thanks. That leaves the 8 bit resolution which I a guessing from the squiggles of the wheel, are repeated at 5 per revolution with the index being geared to the synch point between the two patterns. We commonly use a 1 kliohertz servo loop, and if that A/D internal conversion is done at that rate or higher, the result will be essentially the same as direct optical. I only use PID's in one machine here, the G0704, and have found that generally, the are a solution looking for a problem to solve that in the stepper world does not exist, particularly if the steppers are actually stepper/servo's. They put the PID to good use because their control is directly tied to the src of the problem..There the error signal can and is used to control the magnitude of the motor current on a cycle by cycle basis. If the motor hasn't gotten into position, the error rises but is subject to a time limit as to how hard it can hit the motor, resulting in an alarm output that I tie into the e-stop, this only happens if the motor hits the immovable object. It can run into a stationary chuck jaw on my sheldon, see the error at 5 ipm movement, see the lockup and shut down linuxcnc which shuts down the psu's as all that in tied to F2 state, the carbide chip in the tool is not damaged and no mark is made on the chuck jaw, The driver itself drops power to the motor even faster, so the spring in the z drive bounces back about 5 thou. It takes that powerdown to reset the driver anyway, so it all Just Works. And that has never occurred in actual operation, just in my testing by purposely running it into the stopped chuck. The Sheldon and the 6040 mill both feed the stepgen back to motion, and a missed step would be a disaster, The 6040's Z & B are 3 phase stepper/servo's, XY are plain old oem steppers driven by 2m542's. That B drive has worked so well on the 6040 that I've I've ordered stuff to make one like it for the G0704. The BS-1 clone's motor was burned up by the new atpid and was both too big and too heavy for the G0704 and while accurate, was also slow at about .5 rpms. The B axis I made for the 6040 can do well within a degree, and do it at 400 revs, is a 3nm 3phase driving a 5/1 worm which drives the chuck. That honking heavy spindle is lifted effortlessly by a 1nm 3phase. The OEM Z failed to lift at 10% of the xy speeds of around 200 ipm. I use what works. I had two US Digital opticals on my DC servo's and they were crap. I switched to CUI units and all my problems went away They should have just worked. But the $$ and mounting scared me off. The ultra cheap ($22) for the last one I bought, which has never been installed, the coupling failed and I replaced it with 2 layers of shrink tubing, which has worked, and continues to work, year after year for me. The one I bought to get the coupling is still in the box. One has to assume the data sheet is correct when they say 8000 RPM with 2048 PPR. The 115k Baud has nothing to do with that. Depending on where you put it, 8000 might be too slow. That $22 Omron 1000 ppr is rated to 6k. But it came out of the box balanced and does not have enough swing to drive a 5i25 encoders single ended input, so there's a pair of rs485 translators in the cable making its low voltage balanced output into a 5 volt rail to rail signal. Works at 21k revs. The gear ratio in high gear in the G0704 is about an odd tooth over 7/1 in high gear or 14/1 in low gear. And I can bang that 90 volt 1hp motor with 125 volts and 18 amps. I'm amazed I haven't chewed up the nylon gears rigid tapping with it. In either case, I'll likely set this encoder up for 200 PPR so that even with micro-stepping I really still don't care for higher resolution since I have a 25:1 planetary drive after the motor. Assume I run the motor at 600 RPM max speed. That's 10 RPS or 2000 pulses per rev or 20,000 with 10:1 micro-stepping. But wait... 600 RPM divided by 25 is 24 RPM or 0.4 RPS so 2.5 seconds to turn one revolution. At 200, you may discover quatization noise that limits a PGain. My first optical for the G0704 spindle was a 66 slot brass disk, aka 264 edges per rev. quantization error drove the spindle wall to wall and made it noisy, like it was eating up the bearings in the head gears. The disk I made came loose and ate the opto's, which is when I rigged the Omron. Now its so quiet, with 5x higher PGain, that I have to look at it to verify its turning if set below 500 revs. It was all electronic noise from too low an encoder count. I even cheat on the gear change by running the motor at about 20 rpms if its not fully in gear, so I can be running at full song, reach up and change gears because
Re: [Emc-users] Drilling holes in the back of a stepper motor
Gene, I think you are misreading the data sheet. The 115K baud is for setting up the unit. Ie. Pulses per rev. Once that's done the serial isn't used anymore. I had two US Digital opticals on my DC servo's and they were crap. I switched to CUI units and all my problems went away. One has to assume the data sheet is correct when they say 8000 RPM with 2048 PPR. The 115k Baud has nothing to do with that. In either case, I'll likely set this encoder up for 200 PPR so that even with micro-stepping I really still don't care for higher resolution since I have a 25:1 planetary drive after the motor. Assume I run the motor at 600 RPM max speed. That's 10 RPS or 2000 pulses per rev or 20,000 with 10:1 micro-stepping. But wait... 600 RPM divided by 25 is 24 RPM or 0.4 RPS so 2.5 seconds to turn one revolution. I'm thinking that if it takes 1 turn to release a TTS holder that's still 2.5 seconds. Faster than me with a couple of wrenches. And 25 seconds for 10 turns to release an R8 holder. Also faster than me mucking with the nut. Unless I would use an electric drill to spin it loose. But then think of the process. 1. Pick up two wrenches and loosen draw bar. 2. Put down wrenches. 3. Pick up portable drill and spin draw bar. Now maybe impact drill could do both. 4. Once 10 turns are complete catch R8 tool. Twenty five seconds for it to be done automatically is actually not so bad. John > -Original Message- > From: gene heskett [mailto:ghesk...@shentel.net] > Sent: July 29, 2023 10:37 PM > To: emc-users@lists.sourceforge.net > Subject: Re: [Emc-users] Drilling holes in the back of a stepper motor > > On 7/30/23 00:12, John Dammeyer wrote: > > Hi Gene, > > Best answered with the data sheet. > > https://www.cuidevices.com/product/resource/amt11.pdf > > > A Capacitative device. Not a hall effect. The capacitative may be faster > than a hall. The reason I'm anti hall is the time lag in the hall world > due to the time it takes to do the A/D conversions, which in turn means > the hall encoded setup moves in jerks depending on the speed of that > conversion. The 42C series of small motors, seen as ideal for a 3d > printer, are a disaster because they don't move step by step but jerk by > jerk, frequency of jerk determined to the speed of the A/D. Optical can > determine where the motor is and which direction its turning from any > edge of either A or B signals. At sufficient microstep divisors they can > move quite pretty smoothly. I'd assume this time lag in less for the > capacitative, but at the same time there is the Nyquist effect but the > capacitative conversion is simpler, but the serial output still enforces > a lag in the data stream. Only optical, which at higher and fixed > resolutions is instant. The question remains then "is capacitative fast > enough". And that IDK. The 115 kilobaud output says no to me. That > alone would make me go shopping for a ABI encoder. > > But I'm known to be picky. That's an optical $22 Omron 1000 ppr on the > back shaft of the 1 hp in my g0704. With the gear ratio being > switchable, and that encoder rated for 6k revs, but I watch spindle revs > in the tach display. Top spindle is 3k revs. So times a hair over 7100, > its seeing motor revs of nearly 21k revs at full song and not missing a > beat for about 5 years now. Scale for high gear is a hair over 7100 per > spindle rev. I don't use the index, that is generated by a screw glued > to the spindle going by an ATS-667 hall effect. That has its own > direction problem I'm not smart enough to fix. Someday... > > > Not cheap in Canadian $. > > https://www.digikey.ca/en/products/detail/cui-devices/AMT113Q- > V/4835229 > > > > John > > > > > >> -Original Message- > >> From: gene heskett [mailto:ghesk...@shentel.net] > >> Sent: July 29, 2023 9:02 PM > >> To: emc-users@lists.sourceforge.net > >> Subject: Re: [Emc-users] Drilling holes in the back of a stepper motor > >> > >> On 7/29/23 22:36, John Dammeyer wrote: > >>> Here's my solution. Drilled and milled 0.065" aluminum. Two counter > > sunk > >>> holes in the back. The modified flat head screws epoxied in serving as > >>> threaded studs. Then followed standard CUI installation and alignment > >>> instructions. > >>> > >>> Now to interface to it and write software to capture the motion and > >> position > >>> John > >>> > >> Just one question John, is that encoder optical, or hall effect? > >> Optical is real time, hall effect is not. > >>> > -Original Message- > From: Curtis Dutton [mailto:curtd...@gmail.com] > Sent: June 26, 2023 6:28 PM > To: Enhanced Machine Controller (EMC) > Subject: Re: [Emc-users] Drilling holes in the back of a stepper motor > > On a past life before I started using servos I had attached encoders to > > my > steppers to stop the machine when steps were missed. I used us digital > encoders that came with adhesive backed mounts from the factory. > They >
