[Emc-users] Homming error
Hello... I just wonder how to measure if stepper lost steps and come to idea to use home switch. But I have no idea how to implement this. The idea is simple: 1'st home the machine 2'nd make test moves 3'rd rehome machine In rehoming the EMC just need to read position before setting new home position as is dictated in ini file. and the difference is measure of lost steps... or accuracy problems. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On Wed, 2010-04-07 at 16:59 +0200, Slavko Kocjancic wrote: > Hello... > > I just wonder how to measure if stepper lost steps and come to idea to use > home switch. > But I have no idea how to implement this. The idea is simple: > 1'st home the machine > 2'nd make test moves > 3'rd rehome machine > > In rehoming the EMC just need to read position before setting new home > position as is dictated in ini file. and the difference is measure of lost > steps... or accuracy problems. The accuracy of that system depends on how good your home switch is. I've always used a dial indicator against the spindle for that type of measurement. Same concept but easier since you don't have to home. Dave > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
2010/4/7 dave > > The accuracy of that system depends on how good your home switch is. > > I've always used a dial indicator against the spindle for that type of > measurement. Same concept but easier since you don't have to home. > > Dave > > The accuracy should be good enought. I put the mark on coupler and case. The home is in same step mark near all the time. If not then I know something is wrong. But i have no indication if something is wrong for complete turn. ..and Y coupler is in hidden place hard to see it. so for now I use g0z0 g0x0y0 and check the coupler. Rehome and report after that should be easyer. (and add a few lines into homming procedure) -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 4/7/2010 5:59 PM, Slavko Kocjancic wrote: Hello... I just wonder how to measure if stepper lost steps and come to idea to use home switch. But I have no idea how to implement this. The idea is simple: 1'st home the machine 2'nd make test moves 3'rd rehome machine In rehoming the EMC just need to read position before setting new home position as is dictated in ini file. and the difference is measure of lost steps... or accuracy problems. I would connect the home input to the probe input. Home first, then do some moves, then do a probe towards zero, and look at the point where the probe (home switch ) tripped. Note: you might have to have this connection inactive during homing (maybe it complains that the probe has been hit without a probing move..). You can achieve that in various ways (either an and2 with a digital output set by M5x or simply connect the home switch with the probe after you did the homing, etc). Regards, Alex No virus found in this outgoing message. Checked by AVG - www.avg.com Version: 9.0.801 / Virus Database: 271.1.1/2796 - Release Date: 04/07/10 09:32:00 -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
2010/4/7 Alex Joni > > > On 4/7/2010 5:59 PM, Slavko Kocjancic wrote: > >> Hello... >> >> I just wonder how to measure if stepper lost steps and come to idea to use >> home switch. >> But I have no idea how to implement this. The idea is simple: >> 1'st home the machine >> 2'nd make test moves >> 3'rd rehome machine >> >> In rehoming the EMC just need to read position before setting new home >> position as is dictated in ini file. and the difference is measure of lost >> steps... or accuracy problems. >> >> >> > I would connect the home input to the probe input. > Home first, then do some moves, then do a probe towards zero, and look at > the point where the probe (home switch ) tripped. > > Note: you might have to have this connection inactive during homing (maybe > it complains that the probe has been hit without a probing move..). You can > achieve that in various ways (either an and2 with a digital output set by > M5x or simply connect the home switch with the probe after you did the > homing, etc). > > Regards, > Alex > > Can't do that as I have home as limits. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Statistically you will loose as many steps going forward as you will going backwards, sending the machine to 0,0 will not guarantee that you have not lost and made up steps throughout your job. Greg - Original Message - From: "Slavko Kocjancic" To: "Enhanced Machine Controller (EMC)" Sent: Wednesday, April 07, 2010 5:38 PM Subject: Re: [Emc-users] Homming error > 2010/4/7 dave > >> >> The accuracy of that system depends on how good your home switch is. >> >> I've always used a dial indicator against the spindle for that type of >> measurement. Same concept but easier since you don't have to home. >> >> Dave -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
In my experience with steppers that are running on the edge (mechanically or electrically), they almost never lose steps the same amount in both directions. Best thing I can think of doing is to put a 1" dial indicator on the questionable axis, write a test program or better yet, patch an existing one, so the axis deflects the indicator and pauses there, then goes away from the indicator and runs a lengthy program away from the indicator then have it come back and touch the indicator at the end of the run. If you do that right, you can loop the program and let it run for a while, if anything is out of wack, that will show up as a changing indicator position. You can work the home thing out also, but the dial indicator thing is is a lot more flexible IMO and works on almost all machines.. If your machine is way out of wack, be careful how you mount your indicator so it doesn't get wiped out if the machine is erractic. Dave On 4/7/2010 4:39 PM, Belli Button wrote: > Statistically you will loose as many steps going forward as you will going > backwards, sending the machine to 0,0 will not guarantee that you have not > lost and made up steps throughout your job. > > Greg > > > - Original Message - > From: "Slavko Kocjancic" > To: "Enhanced Machine Controller (EMC)" > Sent: Wednesday, April 07, 2010 5:38 PM > Subject: Re: [Emc-users] Homming error > > > >> 2010/4/7 dave >> >> >>> The accuracy of that system depends on how good your home switch is. >>> >>> I've always used a dial indicator against the spindle for that type of >>> measurement. Same concept but easier since you don't have to home. >>> >>> Dave >>> > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > > -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Oh boy am I truly glad I am not a statistic! So the solution to the statistical dilemma is to put a mark on the shaft and bearing housing. Since we know how many mm the table moves per steppermotorrevolution we travel a multiple of that distance And the stepper shaft will be in exactly the same position on either end of the travel if all is right. Jan de Kruyf. On Wed, Apr 7, 2010 at 10:39 PM, Belli Button wrote: > Statistically you will loose as many steps going forward as you will going > backwards, sending the machine to 0,0 will not guarantee that you have not > lost and made up steps throughout your job. > > Greg > > > - Original Message - > From: "Slavko Kocjancic" > To: "Enhanced Machine Controller (EMC)" > Sent: Wednesday, April 07, 2010 5:38 PM > Subject: Re: [Emc-users] Homming error > > >> 2010/4/7 dave >> >>> >>> The accuracy of that system depends on how good your home switch is. >>> >>> I've always used a dial indicator against the spindle for that type of >>> measurement. Same concept but easier since you don't have to home. >>> >>> Dave > > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Dave pravi: > In my experience with steppers that are running on the edge > (mechanically or electrically), they almost never lose steps the same > amount in both directions. > > Best thing I can think of doing is to put a 1" dial indicator on the > questionable axis, write a test program or better yet, patch an > existing one, so the axis deflects the indicator and pauses there, then > goes away from the indicator and runs a lengthy program away from the > indicator then have it come back and touch the indicator at the end of > the run. If you do that right, you can loop the program and let it run > for a while, if anything is out of wack, that will show up as a changing > indicator position. > > You can work the home thing out also, but the dial indicator thing is is > a lot more flexible IMO and works on almost all machines.. > > If your machine is way out of wack, be careful how you mount your > indicator so it doesn't get wiped out if the machine is erractic. > > Dave > For now my machine work o.k. But in very rare case I got small error and wan't to know when. I'm talking for drilling PCB for example. When you drill PCB and got somewhere error for example 0.5mm or 1mm it's invisible for naked eye. But if after drilling I send machine to gome position and get home difference then board can be assumed junk. As I say for now I just check coupler's. And Y coupler is burried in machine and the Y axis is problematic one. I got random error in rare case.. maybe 1 error per month. But wan't to know when hit's me. If I make check with dial indicator's then all is ok. As I home machine when clamping work that solution can be just fine. I just wonder why every time I post something here got feedback (that's not good, just use this and .., you don't need that.. etc etc...) and near none gave solution? Seem's that I'm not alone. The Bernhard Kubicek write nice filter option and was included in git and working few days after that was cripled and doesn't work any more. The problem is posted but noone wan't to correct. Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Jan de Kruyf pravi: > Oh boy am I truly glad I am not a statistic! > > So the solution to the statistical dilemma is to put a mark on the > shaft and bearing housing. > Since we know how many mm the table moves per steppermotorrevolution > we travel a multiple of that distance > And the stepper shaft will be in exactly the same position on either > end of the travel if all is right. > > Jan de Kruyf. > > Correct. I do that. But to do that in Y axes I need a loot of digging under bellow to se the bearing. I'm shure that few lines of code can make beter solution. ...just can't do as my knowledge is limited in this area and I (unsucesful) bother other people. Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
> I just wonder why every time I post something here got feedback (that's > not good, just use this and .., you don't need that.. etc etc...) and > near none gave solution? You asked for a way to test your axis, and got 3 solutions: * use a dial indicator * use the home switch for probing * mark the coupler and watch the position change If you don't like any of the solutions, then I don't see why anyone is to blame. > Seem's that I'm not alone. The Bernhard Kubicek write nice filter option > and was included in git and working few days after that was cripled and > doesn't work any more. The problem is posted but noone wan't to correct. I am not sure of what issue you are talking about, also not about where it is posted. Bugreports get reported here, where I couldn't find anything: http://sourceforge.net/tracker/?group_id=6744&atid=106744 If you want to help resolving such an issue, it really helps to provide further information (then it gets sorted faster): * when the original patch was added, * when it stopped working, * a patch that fixes the issue (either patch or a pull address from your repo where it's fixed). Obviously the above aren't needed if you can't provide them, but if they are available it will get sorted fast, otherwise you'll have to wait/hope for some devel to have interest to look at the issue. For this git bisect can help a lot: http://book.git-scm.com/5_finding_issues_-_git_bisect.html Most likely it was changed unwillingful by a developer which hasn't used this feature. (git bisect will probably tell you who it was). For developing related things the devel mailing list is probably more appropriate. Best regards, Alex -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Alex Joni pravi: >> I just wonder why every time I post something here got feedback (that's >> not good, just use this and .., you don't need that.. etc etc...) and >> near none gave solution? >> > > You asked for a way to test your axis, and got 3 solutions: > Maybe that is the problem. I didn't ask for solution. (My english is not so good so maybe I ask by mystake.) I propose solution. The software solution to just show message with home difference when homing is done. (just need something like "in basic" Print "X offset ",current_machine_position - machine_home_position just before setting home position before after homing axis. (of course for each axis when homed) > * use a dial indicator > Not good. The error is to rare > * use the home switch for probing > How to do that if home is limit too? > * mark the coupler and watch the position change > That's work. The problem all thing is under bellow and it's not practical. > If you don't like any of the solutions, then I don't see why anyone is to > blame. > > I don't want to blame anyone. >> Seem's that I'm not alone. The Bernhard Kubicek write nice filter option >> and was included in git and working few days after that was cripled and >> doesn't work any more. The problem is posted but noone wan't to correct. >> > > I am not sure of what issue you are talking about, also not about where it > is posted. > Bugreports get reported here, where I couldn't find anything: > http://sourceforge.net/tracker/?group_id=6744&atid=106744 > As I remember it's posted on linuxcnc forum and emc-developers group. And when I do git pull I got that filter menu to work. few day's lather the filter menu dissapear. bkubicek provided zip.file as response all is on the linux cnc forum. *http://tinyurl.com/y9u5yoh *After patch then menu again works but next git pull kill that. > If you want to help resolving such an issue, it really helps to provide > further information (then it gets sorted faster): > * when the original patch was added, > * when it stopped working, > * a patch that fixes the issue (either patch or a pull address from your > repo where it's fixed). > It's all posted at *http://tinyurl.com/y9u5yoh *Seem's bad idea to have two locations for that. I like forum but the response is low. > For developing related things the devel mailing list is probably more > appropriate. > > Best regards, > Alex > I like EMC2 but has filling that 'developers' send's homemade machine's and hobby operator in background with their wishes. I just can't bet it why the targeting big commercial machines with GPL software more than hobby one? em I missing something?!? Slavko -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Slavko, 1. I understand you have a desperate need, but at the same time you cannot hope to solve the issue with software that is still pre-alpha stage, unless the developer of that software is willing to assist you heavily. 2. These problems (over 30 years of working with CNC, I might say) have always been solved with dial or shaft position monitoring. There just was not a friendly developer that quickly knocked up some code. 3. I am not sure if this has been mentioned yet: Your driveshaft might be floating and the bearing or the shaft in the bearing might be moving a little bit every now and again. Put a dial indicator on the shaft end and jerk the table properly to see if the shaft wants to move. Thats all the light I can give at the moment. Cheers Jan de Kruyf. On Thu, Apr 8, 2010 at 10:36 AM, Slavko Kocjancic wrote: > Alex Joni pravi: >>> I just wonder why every time I post something here got feedback (that's >>> not good, just use this and .., you don't need that.. etc etc...) and >>> near none gave solution? >>> >> >> You asked for a way to test your axis, and got 3 solutions: >> > > Maybe that is the problem. I didn't ask for solution. (My english is not > so good so maybe I ask by mystake.) > I propose solution. The software solution to just show message with home > difference when homing is done. (just need something like "in basic" > Print "X offset ",current_machine_position - machine_home_position just > before setting home position before after homing axis. (of course for > each axis when homed) > >> * use a dial indicator >> > Not good. The error is to rare >> * use the home switch for probing >> > How to do that if home is limit too? >> * mark the coupler and watch the position change >> > That's work. The problem all thing is under bellow and it's not practical. >> If you don't like any of the solutions, then I don't see why anyone is to >> blame. >> >> > I don't want to blame anyone. >>> Seem's that I'm not alone. The Bernhard Kubicek write nice filter option >>> and was included in git and working few days after that was cripled and >>> doesn't work any more. The problem is posted but noone wan't to correct. >>> >> >> I am not sure of what issue you are talking about, also not about where it >> is posted. >> Bugreports get reported here, where I couldn't find anything: >> http://sourceforge.net/tracker/?group_id=6744&atid=106744 >> > As I remember it's posted on linuxcnc forum and emc-developers group. > And when I do git pull I got that filter menu to work. few day's lather > the filter menu dissapear. bkubicek provided zip.file as response all is > on the linux cnc forum. *http://tinyurl.com/y9u5yoh > *After patch then menu again works but next git pull kill that. >> If you want to help resolving such an issue, it really helps to provide >> further information (then it gets sorted faster): >> * when the original patch was added, >> * when it stopped working, >> * a patch that fixes the issue (either patch or a pull address from your >> repo where it's fixed). >> > It's all posted at *http://tinyurl.com/y9u5yoh > *Seem's bad idea to have two locations for that. I like forum but the > response is low. >> For developing related things the devel mailing list is probably more >> appropriate. >> >> Best regards, >> Alex >> > I like EMC2 but has filling that 'developers' send's homemade machine's > and hobby operator in background with their wishes. I just can't bet it > why the targeting big commercial machines with GPL software more than > hobby one? > > em I missing something?!? > Slavko > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 04/07/2010 04:42 PM, Slavko Kocjancic wrote: > 2010/4/7 Alex Joni >> I would connect the home input to the probe input. >> Home first, then do some moves, then do a probe towards zero, and look at >> the point where the probe (home switch ) tripped. >> >> Note: you might have to have this connection inactive during homing (maybe >> it complains that the probe has been hit without a probing move..). You can >> achieve that in various ways (either an and2 with a digital output set by >> M5x or simply connect the home switch with the probe after you did the >> homing, etc). >> >> Regards, >> Alex >> >> Can't do that as I have home as limits. Could you temporarily disable the limits for testing purposes? Mark -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
That's where mag mounts come in real handy. I have a 2" depth gauge that I used on my machine. Mis-calculated the first run. Good thing it was magnetically attached rather than clamped on... Mark On 04/07/2010 08:33 PM, Dave wrote: > In my experience with steppers that are running on the edge > (mechanically or electrically), they almost never lose steps the same > amount in both directions. > > Best thing I can think of doing is to put a 1" dial indicator on the > questionable axis, write a test program or better yet, patch an > existing one, so the axis deflects the indicator and pauses there, then > goes away from the indicator and runs a lengthy program away from the > indicator then have it come back and touch the indicator at the end of > the run. If you do that right, you can loop the program and let it run > for a while, if anything is out of wack, that will show up as a changing > indicator position. > > You can work the home thing out also, but the dial indicator thing is is > a lot more flexible IMO and works on almost all machines.. > > If your machine is way out of wack, be careful how you mount your > indicator so it doesn't get wiped out if the machine is erractic. > > Dave > > On 4/7/2010 4:39 PM, Belli Button wrote: >> Statistically you will loose as many steps going forward as you will going >> backwards, sending the machine to 0,0 will not guarantee that you have not >> lost and made up steps throughout your job. >> >> Greg >> >> >> - Original Message ----- >> From: "Slavko Kocjancic" >> To: "Enhanced Machine Controller (EMC)" >> Sent: Wednesday, April 07, 2010 5:38 PM >> Subject: Re: [Emc-users] Homming error >> >> >> >>> 2010/4/7 dave >>> >>> >>>> The accuracy of that system depends on how good your home switch is. >>>> >>>> I've always used a dial indicator against the spindle for that type of >>>> measurement. Same concept but easier since you don't have to home. >>>> >>>> Dave >>>> >> >> -- >> Download Intel® Parallel Studio Eval >> Try the new software tools for yourself. Speed compiling, find bugs >> proactively, and fine-tune applications for parallel performance. >> See why Intel Parallel Studio got high marks during beta. >> http://p.sf.net/sfu/intel-sw-dev >> ___ >> Emc-users mailing list >> Emc-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/emc-users >> >> > > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
>> You asked for a way to test your axis, and got 3 solutions: > > Maybe that is the problem. I didn't ask for solution. (My english is not > so good so maybe I ask by mystake.) Ok, I didn't understand that like that. > I propose solution. The software solution to just show message with home > difference when homing is done. (just need something like "in basic" > Print "X offset ",current_machine_position - machine_home_position just > before setting home position before after homing axis. (of course for > each axis when homed) well, you are proposing to make a change to the software so it detects a fault of your machine. I am not sure if anyone else wants this feature, and you surely understand we can't incorporate all features possible. http://en.wikipedia.org/wiki/Feature_creep Otoh, the change you want implies about 2-3h worth of digging + coding, and some additional testing. Maybe someone who is interested in this will spend that time on the feature. Then you can either get a patch (which you apply on your local sources), or it gets incorporated into emc2 if enough people want it. >> * use the home switch for probing >> > How to do that if home is limit too? You can have a special M-code called which unhooks the home limit connection, so you can do the probe, then (using another m-code) reconnect it again. > As I remember it's posted on linuxcnc forum and emc-developers group. > And when I do git pull I got that filter menu to work. few day's lather > the filter menu dissapear. bkubicek provided zip.file as response all is > on the linux cnc forum. *http://tinyurl.com/y9u5yoh > *After patch then menu again works but next git pull kill that. >> If you want to help resolving such an issue, it really helps to provide >> further information (then it gets sorted faster): >> * when the original patch was added, >> * when it stopped working, >> * a patch that fixes the issue (either patch or a pull address from your >> repo where it's fixed). >> > It's all posted at *http://tinyurl.com/y9u5yoh > *Seem's bad idea to have two locations for that. I like forum but the > response is low. That's because the forum is not a supported way to reach emc2 developers. A lot of people wanted to have the forums, so we set it up, but that doesn't mean you'll get usefull or valid responses on the forum. Reading that page (not much time I can spend on it) I don't see a definitive answer. I was looking for: * currently the filter in emc2 is broken, this is the patch to make it work again Instead there is some discussion between you (I guess) and bkubicek, which ends at: "There is now a partially working implementation for filtering." That doesn't tell me the status.. partially working doesn't sound like it should be integrated in emc2 yet. There is a zip file attached containing some sources, but without a proper description and without copyright and license information (until it has proper licensing we surely can't think about integrating it into emc2). >> For developing related things the devel mailing list is probably more >> appropriate. >> > I like EMC2 but has filling that 'developers' send's homemade machine's > and hobby operator in background with their wishes. I just can't bet it > why the targeting big commercial machines with GPL software more than > hobby one? > > am I missing something?!? The main goal of emc2 is to make a solid software which works on a broad range of machines. You can configure it to run on hobby machines, but also on serious (> 3-4t) machining centers. Regards, Alex -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On Thu, Apr 8, 2010 at 11:57 AM, Alex Joni wrote: > >> You asked for a way to test your axis, and got 3 solutions: > > > > Maybe that is the problem. I didn't ask for solution. (My english is not > > so good so maybe I ask by mystake.) > > Ok, I didn't understand that like that. > > > I propose solution. The software solution to just show message with home > > difference when homing is done. (just need something like "in basic" > > Print "X offset ",current_machine_position - machine_home_position just > > before setting home position before after homing axis. (of course for > > each axis when homed) > > well, you are proposing to make a change to the software so it detects a > fault of your machine. > I am not sure if anyone else wants this feature, and you surely understand > we can't incorporate all features possible. > http://en.wikipedia.org/wiki/Feature_creep > > Otoh, the change you want implies about 2-3h worth of digging + coding, and > some additional testing. > Maybe someone who is interested in this will spend that time on the > feature. > Then you can either get a patch (which you apply on your local sources), or > it gets incorporated into emc2 if enough people want it. > > > > >> * use the home switch for probing > >> > > How to do that if home is limit too? > > You can have a special M-code called which unhooks the home limit > connection, so you can do the probe, then (using another m-code) reconnect > it again. > > > > As I remember it's posted on linuxcnc forum and emc-developers group. > > And when I do git pull I got that filter menu to work. few day's lather > > the filter menu dissapear. bkubicek provided zip.file as response all is > > on the linux cnc forum. *http://tinyurl.com/y9u5yoh > > *After patch then menu again works but next git pull kill that. > >> If you want to help resolving such an issue, it really helps to provide > >> further information (then it gets sorted faster): > >> * when the original patch was added, > >> * when it stopped working, > >> * a patch that fixes the issue (either patch or a pull address from your > >> repo where it's fixed). > >> > > It's all posted at *http://tinyurl.com/y9u5yoh > > *Seem's bad idea to have two locations for that. I like forum but the > > response is low. > > That's because the forum is not a supported way to reach emc2 developers. > A lot of people wanted to have the forums, so we set it up, but that > doesn't > mean you'll get usefull or valid responses on the forum. > Reading that page (not much time I can spend on it) I don't see a > definitive > answer. > I was looking for: > * currently the filter in emc2 is broken, this is the patch to make it work > again > Instead there is some discussion between you (I guess) and bkubicek, which > ends at: > "There is now a partially working implementation for filtering." > That doesn't tell me the status.. partially working doesn't sound like it > should be integrated in emc2 yet. > There is a zip file attached containing some sources, but without a proper > description and without copyright and license information (until it has > proper licensing we surely can't think about integrating it into emc2). > > First of all, I am currently 100% ok with the way emc development is handled. For the filtering, there are two things: )First, the patches in axis. I proposed this filtering menu. The idea is to have external programs, aka filters, that expect a gcode filename as first command line parameter. These filters then can do whatever they want, but should output useable gcode in their output stream. So clicking Menu->File->Filter->someprogramname calls "someprogramname currentlyopendgcode.ngc" and replaces the currently opened gcode by that. One filter e.g. could be: invertxcoordinates.sh or scale, or whatever. I proposed this feature, and with my very limited knowledge of python and tcl made a menu. Eppler did some nice reprogramming of the filtering, and made the integration in axis. However, at least on my Ubuntu LTS system, it did not work right away. I had to change one line of code, so my old python could parse it. However there are two more or less unsolved problems: -reloading, or homeing reload the original gcode file, and the filtering is lost However, for me its is just a convenience thing to have the axis integration of the filtering. They obviously can also work on command line level. )Second, there is the filtering tools that I am creating. It started as bash scripts, then i moved to c++ as gcode interpretation became more complex. This can be called by this pachable axis menu, or by command line. Every now and then, if I have some spare time between my two jobs, and if I find an error, I do some reprogramming. One temporary snapshot I posted in the forum, so if somebody knowing to programm really wants to try it. I intend to create a git-repository, and some more info than you get by calling it without command line options. But I don't have much time. - So, altog
Re: [Emc-users] Homming error
On 4/8/2010 7:06 AM, Bernhard Kubicek wrote: > > First of all, I am currently 100% ok with the way emc development is > handled. > For the filtering, there are two things: > )First, the patches in axis. > I proposed this filtering menu. The idea is to have external programs, aka > filters, that expect a gcode filename as first command line parameter. These > filters then can do whatever they want, but should output useable gcode in > their output stream. So clicking Menu->File->Filter->someprogramname calls > "someprogramname currentlyopendgcode.ngc" and replaces the currently opened > gcode by that. > One filter e.g. could be: invertxcoordinates.sh or scale, or whatever. > I proposed this feature, and with my very limited knowledge of python and > tcl made a menu. > Eppler did some nice reprogramming of the filtering, and made the > integration in axis. However, at least on my Ubuntu LTS system, it did not > work right away. I had to change one line of code, so my old python could > parse it. > > However there are two more or less unsolved problems: > -reloading, or homeing reload the original gcode file, and the filtering is > lost > However, for me its is just a convenience thing to have the axis integration > of the filtering. > They obviously can also work on command line level. > > )Second, there is the filtering tools that I am creating. > It started as bash scripts, then i moved to c++ as gcode interpretation > became more complex. > This can be called by this pachable axis menu, or by command line. > Every now and then, if I have some spare time between my two jobs, and if I > find an error, I do some reprogramming. > One temporary snapshot I posted in the forum, so if somebody knowing to > programm really wants to try it. > I intend to create a git-repository, and some more info than you get by > calling it without command line options. > But I don't have much time. > > > - > So, altogether, I find it awesome that somebody really picked up my idea to > integrate the filtering, within two days after my proposal. Second, I am not > angry that it is not integrated currently, because there are some drawbacks > in the design. I think everything is handled well, if every nobody like me > could have some code integrated without proper testing and quality > assuarance, emc would get bloated and complex and unexpandable. > Also, I don't expect anyone else but me to work for the proper integration. > > Please everybody, don't escalate things and be friendly with each other. Emc > is great, everybody should be thankfull for others work, and obviously work > together and not against each other. > One problem might be that it is very easy to be accidentially insulting in a > secondary language. > > greetings, > Bernhard > -- > > >>Please everybody, don't escalate things and be friendly with each other. Emc is great, everybody should be thankfull for others work, and obviously work together and not against each other. I agree.. Bernhard, I really appreciate your efforts at the filtering.. and the interaction with the other developers, Epler, Alex, Cradek, SWPadnos.and others ..etc. I am constantly impressed by the speed at which communication and collaboration take place regarding EMC2. I need to do some filtering work and I didn't notice that a change had slipped out of the master.. I'm sure we can get it slipped back in if that happened. ;-)I need to go back and look at that patch again. I can appreciate Slavko's frustration if he has an outstanding issue, but compared to dealing with a commercial control company or other closed source CNC software, the help available from the EMC2 community is fantastic!! I simply can't express my gratitude sufficiently to the many people on the IRC and this list who have helped me out. THANKS! just doesn't seem to be sufficient. Dave (Dave911 on the IRC) -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Dave pravi: > is great, everybody should be thankfull for others work, and obviously work > together and not against each other. > > I agree.. > > Bernhard, I really appreciate your efforts at the filtering.. and the > interaction with the other developers, Epler, Alex, Cradek, SWPadnos.and > others ..etc. I am constantly impressed by the speed at which > communication and collaboration take place regarding EMC2. > > I need to do some filtering work and I didn't notice that a change had > slipped out of the master.. I'm sure we can get it slipped back in if > that happened. ;-)I need to go back and look at that patch again. > > I can appreciate Slavko's frustration if he has an outstanding issue, > but compared to dealing with a commercial control company or other > closed source CNC software, the help available from the EMC2 community > is fantastic!! > > I simply can't express my gratitude sufficiently to the many people on > the IRC and this list who have helped me out. > > THANKS! just doesn't seem to be sufficient. > > > Dave > (Dave911 on the IRC) > Hello... I don't want to blame EMC2 community. If I do that then it's not intentionaly. As my english is no so good there is big chance to some post from me can be missreaded. I'm still new in emc2. I'm new to linux too. To be worse I can call myself as newbye for CNC too. I have small CNC for drilling PCB circuit board over 20 years. It's made entirely from aluminum but is to light to make some other thing. so few years ago I start 1m x 0.5m wood router and 30cm x 30cm router.(Before that I made some MDF routers just for fun and learn lot) As smaller is done I make some parts and hit some problems too. I'm not machinist. For now I made just few jigs and few PCB on machine running EMC. the 1'st problem I have is my unconventional motor driver. I need to add custom phase drive. As I only change the last phase sequences (to be able to run machine) the patch 'arrive' just two day's latter. I very like that. The other's two problem's (webcam tab and homming report) sems to be not accepted in comunity. I like webcam as I find it easyest way to center drill holles for PCB. I have that on old machine I see that on MACH3 forum and is pretty hard to understand why not in EMC. I was trying to incorporate it but fail as my knowledge to python in nill. Now I bother with different window's flying around desktop. I do have more wishes for EMC functionality but a lot of them public doesn't need it. But thing's as webcam or home repport I think the public want just doesn't aware the benifit's of it. and again. I don't want to blame people making EMC! If i'm unhappy when my idea is rejected then that's my problem. I'm programmer too (just other languages) and know how many free hour's take project like EMC2. And findin'g people to make that effort for free can't be blamed! Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 9 April 2010 08:28, Slavko Kocjancic wrote: > I do have more wishes for EMC functionality but a lot of them public > doesn't need it. But thing's as webcam or home repport I think the > public want just doesn't aware the benifit's of it. I can see the value of the "homing report". Perhaps if a machine that is already homed is re-homed and the new home position and old position differ by more than a certain amount then a following error should be triggered? (Or perhaps just an information note) It seems to me that if the home position has moved then there is an error, and the user should be told. Most of the time they probably know, and are re-homing because of it, but for those of us running open-loop with steppers it would be valuable information. -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Andy Pugh pravi: > On 9 April 2010 08:28, Slavko Kocjancic wrote: > > >> I do have more wishes for EMC functionality but a lot of them public >> doesn't need it. But thing's as webcam or home repport I think the >> public want just doesn't aware the benifit's of it. >> > > I can see the value of the "homing report". Perhaps if a machine that > is already homed is re-homed and the new home position and old > position differ by more than a certain amount then a following error > should be triggered? (Or perhaps just an information note) > > It seems to me that if the home position has moved then there is an > error, and the user should be told. Most of the time they probably > know, and are re-homing because of it, but for those of us running > open-loop with steppers it would be valuable information. > > ... What to say... Agree. While I'm not C programmer I throw view in source to check how to manage that. But I'm lost. The home routines are in separate file but are little complicated. I have simple aproach in mind. When home move is ended just calculate difference from current position and home position. But found that isn't so easy. The machine coordinate are preseted in midle of sequence too. As machine need to reset coordinate in correct place to get screw compensation ready. Other option seems to be something like: save all joint's positions call doHome calculate differences and print message ...but can't find that doHome point. Seems that homing is just initiated and finish is in motion routine. ?!? just to much for me... Have nice day Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 4/9/2010 6:46 AM, Andy Pugh wrote: > On 9 April 2010 08:28, Slavko Kocjancic wrote: > > >> I do have more wishes for EMC functionality but a lot of them public >> doesn't need it. But thing's as webcam or home repport I think the >> public want just doesn't aware the benifit's of it. >> > I can see the value of the "homing report". Perhaps if a machine that > is already homed is re-homed and the new home position and old > position differ by more than a certain amount then a following error > should be triggered? (Or perhaps just an information note) > > It seems to me that if the home position has moved then there is an > error, and the user should be told. Most of the time they probably > know, and are re-homing because of it, but for those of us running > open-loop with steppers it would be valuable information. > > I agree that the "homing report" could be a valuable thing.I think you are assuming there is no support for a "homing report", but I didn't notice anyone asking if anyone thought it was a good idea or not. Sometimes noise will get on Servo encoder cables also and cause positioning errors also, so it could be valuable to a servo system. I'm not sure that what is wanted couldn't be done in Hal right now.. "on the edge of home, switch grab the current axis position count, display that on a PyVcp screen" ?? If the result isn't zero or whatever it should be (it should always be the same or close) then you have a problem. Regarding the video inside of EMC2; Slavko, are you using another app in a different window right now? Why doesn't that work for you?When they put the video window inside of Mach3, that caused all kinds of issues and it seems like no one is happy with the size of the window within Mach3, the way the cross hairs display etc, etc... Then you can get into camera differences also. So many desires, and so little time. ;-) Dave (Dave911 on IRC) -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On Fri, 09 Apr 2010 15:40 +0200, "Slavko Kocjancic" wrote: > > While I'm not C programmer I throw view in source to check how to manage > that. But I'm lost. The home routines are in separate file but are > little complicated. I have simple aproach in mind. When home move is > ended just calculate difference from current position and home position. > But found that isn't so easy. The machine coordinate are preseted in > midle of sequence too. As machine need to reset coordinate in correct > place to get screw compensation ready. > > Other option seems to be something like: > > save all joint's positions > call doHome > calculate differences and print message > > ...but can't find that doHome point. Seems that homing is just initiated > and finish is in motion routine. ?!? > > just to much for me... Real-time control software tends to look a little different. The code runs from beginning to end at a specific interval - usually every 0.001 seconds for mot of EMC's motion control code, including the homing code. The homing code contains a state machine. If the state is IDLE, then it does nothing. Otherwise, it takes actions depending on the current state. That action may include changing to another state. I'm somewhat reluctant to do something like this inside the homing code. It is a niche function, and would probably not work at all (and even give false indications) if used on a servo machine with index pulses. However, it might be possible to do the detection in an external (custom) HAL component. The homing state is brought out on a HAL pin, so the component can know what is going on. Regards, John Kasunich -- John Kasunich jmkasun...@fastmail.fm -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 9 April 2010 15:38, John Kasunich wrote: > it might be possible to do the detection in an external (custom) HAL > component. The homing state is brought out on a HAL pin, so the > component > can know what is going on. Thinking about it, I am not sure it needs even that. I suspect that you could do it with not a lot more than a mux2 wired to the axis position and the home switch state. I will have a go when I get home (in an hour or so) -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Dave pravi: > I agree that the "homing report" could be a valuable thing.I think > you are assuming there is no support for a "homing report", but I didn't > notice anyone asking if anyone thought it was a good idea or not. > Sometimes noise will get on Servo encoder cables also and cause > positioning errors also, so it could be valuable to a servo system. > > I'm not sure that what is wanted couldn't be done in Hal right now.. > "on the edge of home, switch grab the current axis position count, > display that on a PyVcp screen" ?? If the result isn't zero or whatever > it should be (it should always be the same or close) then you have a > problem. > I was try some thing but for now I got false reading all the time. > Regarding the video inside of EMC2; Slavko, are you using another app > in a different window right now? Why doesn't that work for you?When > Yes. I was able to make thing ussable. It's ussable but not practic. Let's see what and why. If I arrange all windows to not overlap then all are too small so in first place I maximize AXIS interface. (why can't be started maximized at all?!? Or at least remember last position?!?) After that I start Camera window. I right click camera window and select always on top and move that window over AXIS preview place. (just fit inside) After that I start Learn.py and make that window always on top too and move over code window of AXIS. So now I have all windows visible doesn't matter witch is selected. Now I must click somewhere on AXIS interface and jog machine to reference point (with help of webcam) After that I must select and click Learn/Position. Then again I must click somewhere on AXIS interface to select it and repeat jog move to other reference points. And repeat that for all reference points. It's very easy to mess the selected window and keystrokes. I think to easy to make mess. After that I close the cam window and learn window. The coordinate of reference points are stored in file. All that just to save current coordinate in file. (under jogging) The old Turbocnc Can do that with single keypress. And modern CNC machines too! Few times I must scan hole position for PCB board. And in more machine procedure is same. Jog over the hole (with help of camera or optical finder) and pres single key to save coordinate. I don't understand why on EMC2 is too hard to have same thing?! (well keyboard shortcuts are pretty bussy). > they put the video window inside of Mach3, that caused all kinds of > issues and it seems like no one is happy with the size of the window > within Mach3, the way the cross hairs display etc, etc... Then you can > get into camera differences also. So many desires, and so little time. > ;-) > > Dave > (Dave911 on IRC) After digging into camera stuff I realized that linux just doesn't have (yet) good camera support. They need a bunch of drivers and tweaking. So the code to grab picture inside AXIS seem's not good way. Simply won't run on all computers. But I think if the "another tab for webcam" just reflect image from some pipe od even webcast (streaming video) then thing is a lot easyer. I find few videostreaming packages and work's with different cameras. I try to run some on these and result is ok. I can have live video in firefox browser on EMC2 machine and on other (XP) machine in same time.So it's works. It can be done. Someone want that? Well at least one right now here :-) Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
The problem with the homing report is the destruction of data. When the home position is found, the machine position count is set to zero (or whatever the HOME_OFFSET value is). So, the old home position is destroyed by that operation. To stay with the canonical encoder definition as it is, you would have to store the position JUST before the home operation completed. I think you could make a custom HAL component that recorded the machine position every servo cycle. Whenever it saw ENCODER_INDEX transition from true to false, it would save the value recorded from the previous cycle. This would be the position one millisecond (at the default SERVO_THREAD rate) before the home position was found. I think this would give you a pretty consistent way of checking for drift of the home position. On the first home whe EMC is started, it would show the distance traveled from where the machine was to the home position. On later homing operations, the value should be very close to zero, assuming the approach to the home switch is slow. (Maybe watching ENCODER_INDEX only works where the index pulse is being used, how does the position count get set for stepgen, for instance? I'm only familiar with servo-like interfaces.) I think this will work if there's a signal that tells you the home operation is happening NOW, and doesn't require any change to EMC2 itself. You would monitor this with HalMeter or the show hal signals menu. Jon -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 9 April 2010 16:41, Slavko Kocjancic wrote: > After that I start Camera window. I right click camera window and select > always on top and move that window over AXIS preview place. (just fit > inside) > After that I start Learn.py and make that window always on top too and > move over code window of AXIS. > So now I have all windows visible doesn't matter witch is selected. > Now I must click somewhere on AXIS interface and jog machine to > reference point (with help of webcam) Two possibilities occur to me: 1) You could add "jog" and "store" buttons to the learn.py interface. 2) You could hook up a USB joypad for jogging (well worth doing anyway) and use one or more of the buttons on that to store the learned position. (You can link physical buttons to hal pins that will run G-code fragments, see the [HALUI] section on this page:http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Adding_More_Controls_To_Simple_Remote_Pendant -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Andy Pugh pravi: > On 9 April 2010 16:41, Slavko Kocjancic wrote: > > >> After that I start Camera window. I right click camera window and select >> always on top and move that window over AXIS preview place. (just fit >> inside) >> After that I start Learn.py and make that window always on top too and >> move over code window of AXIS. >> So now I have all windows visible doesn't matter witch is selected. >> Now I must click somewhere on AXIS interface and jog machine to >> reference point (with help of webcam) >> > > Two possibilities occur to me: > 1) You could add "jog" and "store" buttons to the learn.py interface. > I'm hanging on other side. To add command "save machine coordinate to file" and of course binded with key shortcut. > 2) You could hook up a USB joypad for jogging (well worth doing > anyway) and use one or more of the buttons on that to store the > learned position. (You can link physical buttons to hal pins that will > run G-code fragments, see the [HALUI] section on this > page:http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Adding_More_Controls_To_Simple_Remote_Pendant > > Nice. I have plan to setup yoistick but for now I'm stuck to keyboard. But If I have yoistick I have no clue how to write coordinate to file with button press. Yes I know I can link buton to some HAL pin or Gcode. But how from Gcode Open file, add coordinate to the end and close file? The way can be custom program M1xx. That can open and close file and even write betwen. But how to retrive machine coordinate ?!? I know M1xx can be uset easily for output. But how to get input from HAL? I looking that for near 1'st day I start with EMC. Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Jon Elson pravi: > The problem with the homing report is the destruction of data. When the > home position is found, the machine position count is set to zero (or > whatever the HOME_OFFSET value is). So, the old home position is > destroyed by that operation. > > Just that. Before writing that HOME_OFFSET just read previous value and show difference. But doesn't work! The destruction of data is done earlyer too! That's is done to screw_comp file come at right position! So isn't simple as I think. Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 4/9/2010 12:04 PM, Jon Elson wrote: > The problem with the homing report is the destruction of data. When the > home position is found, the machine position count is set to zero (or > whatever the HOME_OFFSET value is). So, the old home position is > destroyed by that operation. To stay with the canonical encoder > definition as it is, you would have to store the position JUST before > the home operation completed. I think you could make a custom HAL > component that recorded the machine position every servo cycle. > Whenever it saw ENCODER_INDEX transition from true to false, it would > save the value recorded from the previous cycle. This would be the > position one millisecond (at the default SERVO_THREAD rate) before the > home position was found. I think this would give you a pretty > consistent way of checking for drift of the home position. On the first > home whe EMC is started, it would show the distance traveled from where > the machine was to the home position. On later homing operations, the > value should be very close to zero, assuming the approach to the home > switch is slow. (Maybe watching ENCODER_INDEX only works where the > index pulse is being used, how does the position count get set for > stepgen, for instance? I'm only familiar with servo-like interfaces.) > > I think this will work if there's a signal that tells you the home > operation is happening NOW, and doesn't require any change to EMC2 > itself. You would monitor this with HalMeter or the show hal signals menu. > > Jon > > > Hi Jon, Slavko is running steppers so I believe he has a base thread running also which should be plenty fast enough to catch the step count just before hitting the home switch. :-) And if not, then I would think that a base thread could be added to increase the accuracy. Just has to be worked out. Dave -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 4/9/2010 12:13 PM, Andy Pugh wrote: > On 9 April 2010 16:41, Slavko Kocjancic wrote: > > >> After that I start Camera window. I right click camera window and select >> always on top and move that window over AXIS preview place. (just fit >> inside) >> After that I start Learn.py and make that window always on top too and >> move over code window of AXIS. >> So now I have all windows visible doesn't matter witch is selected. >> Now I must click somewhere on AXIS interface and jog machine to >> reference point (with help of webcam) >> > Two possibilities occur to me: > 1) You could add "jog" and "store" buttons to the learn.py interface. > 2) You could hook up a USB joypad for jogging (well worth doing > anyway) and use one or more of the buttons on that to store the > learned position. (You can link physical buttons to hal pins that will > run G-code fragments, see the [HALUI] section on this > page:http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Adding_More_Controls_To_Simple_Remote_Pendant > > > >>You can link physical buttons to hal pins that will run G-code fragments, see the [HALUI] section on this. Yes, very cool and as Andy pointed out a while back that fragment can call a G code subroutine which can have even more code in it.. :-) I might be trying that out very soon. Dave -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
You could implement a similar function using a contrived method, that is not part of the G28 command or other homing function. Basically you would have a subroutine to deliberately move the machine to the home switches and write the x,y,z values to file as the switches are crossed. Assuming they should be 0,0,0, the new, 'wrong' values might be 0,3,0. Then a program outside EMC could interrogate the file data. It might also be time stamped. The routine would then have to move back to resume machining. So depending on how critical your project is, you could use a text or other editor to insert this gcode as a call function every 500 lines, or 1000 lines, or maybe after every tool change, into your gcode before you run the file. I don't know how difficult it would be to implement this using EMC. When I worked with an Emco mill(stepper), some programs ran for more than 24hrs. I wanted to insert a 'home' call at the start of every new pass. If a few steps were somehow lost, at least one would be starting with a 'clean slate', and not have an incremental error. But it could not be done. Roland PS (now I'll wait to get shot to pieces by the gurus...) On 9 April 2010 18:42, Slavko Kocjancic wrote: > Jon Elson pravi: > > The problem with the homing report is the destruction of data. When the > > home position is found, the machine position count is set to zero (or > > whatever the HOME_OFFSET value is). So, the old home position is > > destroyed by that operation. > > > > > > Just that. Before writing that HOME_OFFSET just read previous value and > show difference. But doesn't work! The destruction of data is done > earlyer too! That's is done to screw_comp file come at right position! > So isn't simple as I think. > > Slavko. > > > -- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Dave pravi: > On 4/9/2010 12:04 PM, Jon Elson wrote: > >> The problem with the homing report is the destruction of data. When the >> home position is found, the machine position count is set to zero (or >> whatever the HOME_OFFSET value is). So, the old home position is >> destroyed by that operation. To stay with the canonical encoder >> definition as it is, you would have to store the position JUST before >> the home operation completed. I think you could make a custom HAL >> component that recorded the machine position every servo cycle. >> Whenever it saw ENCODER_INDEX transition from true to false, it would >> save the value recorded from the previous cycle. This would be the >> position one millisecond (at the default SERVO_THREAD rate) before the >> home position was found. I think this would give you a pretty >> consistent way of checking for drift of the home position. On the first >> home whe EMC is started, it would show the distance traveled from where >> the machine was to the home position. On later homing operations, the >> value should be very close to zero, assuming the approach to the home >> switch is slow. (Maybe watching ENCODER_INDEX only works where the >> index pulse is being used, how does the position count get set for >> stepgen, for instance? I'm only familiar with servo-like interfaces.) >> >> I think this will work if there's a signal that tells you the home >> operation is happening NOW, and doesn't require any change to EMC2 >> itself. You would monitor this with HalMeter or the show hal signals menu. >> >> Jon >> >> >> >> > Hi Jon, > > Slavko is running steppers so I believe he has a base thread running > also which should be plenty fast enough to catch the step count just > before hitting the home switch. :-) > And if not, then I would think that a base thread could be added to > increase the accuracy. > > Just has to be worked out. > > Dave > Dave you are right. It's stepper based. And base thread is set very conservative at 50uSec. Stepper's start buzzing if I set base period at 60uS and computer's fell lazy is I set BT at 20uSec. Latency of system is aprox 16uSec. I have no problem with dir/step timming as I use phase drive. Slavko. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Roland Jollivet pravi: > You could implement a similar function using a contrived method, that is not > part of the G28 command or other homing function. > > Basically you would have a subroutine to deliberately move the machine to > the home switches and write the x,y,z values to file as the switches are > crossed. Assuming they should be 0,0,0, the new, 'wrong' values might be > 0,3,0. Then a program outside EMC could interrogate the file data. It might > also be time stamped. The routine would then have to move back to resume > machining. > > So depending on how critical your project is, you could use a text or other > editor to insert this gcode as a call function every 500 lines, or 1000 > lines, or maybe after every tool change, into your gcode before you run the > file. > > I don't know how difficult it would be to implement this using EMC. > When I worked with an Emco mill(stepper), some programs ran for more than > 24hrs. I wanted to insert a 'home' call at the start of every new pass. If a > few steps were somehow lost, at least one would be starting with a 'clean > slate', and not have an incremental error. But it could not be done. > > Roland > PS (now I'll wait to get shot to pieces by the gurus...) > > > I'm not shure what you mean. The problem to handle this is that switch is limit switch too. So any code moving axis can't reach that switch as soft limit is trigered before. If I override soft limit then switch act as limit and triger hard limit.. So the only rehular way is while homming. -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
> But how from Gcode Open file, add coordinate to the end and close file? > The way can be custom program M1xx. That can open and close file and > even write betwen. But how to retrive machine coordinate ?!? I know M1xx > can be uset easily for output. But how to get input from HAL? I looking > that for near 1'st day I start with EMC. Halsampler and Sampler are one way (and probably the best) In your hal file add the lines: loadrt sampler depth=3 cfg="FFF" addf sampler.0 servo-thread net xlog halui.axis.0.pos-feedback => sampler.0.pin.0 net ylog halui.axis.1.pos-feedback => sampler.0.pin.1 net zlog halui.axis.2.pos-feedback => sampler.0.pin.2 (You will need the line HALUI = halui in your .ini file [HAL] section Then invoke " halsampler -n 1 >> logfile " in any way that suits you. -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
On 9 April 2010 18:00, Dave wrote: > Just has to be worked out. This seems to pretty-much work. Note that it doesn't work unless you put the mux and the oneshot into a thread before the motion functions. The axis position on hitting the limit is pretty consistent (enough to show a position loss) and can be seen my looking at xcheck in Hal Meter or linking it to a PyVCP control. loadrt mux2 count=1 loadrt oneshot count=1 addf oneshot.0 servo-thread addf mux2.0 servo-thread net xcheck axis.0.joint-pos-fb => mux2.0.in1 net xcheckfb mux2.0.in0 <= mux2.0.out net xchecktrig oneshot.0.out => mux2.0.sel setp oneshot.0.width 0.0001 setp oneshot.0.rising 1 setp oneshot.0.falling 0 net both-home-x => axis.0.home-sw-in => oneshot.0.in -- atp -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Homming error
Slavko Kocjancic wrote: > Jon Elson pravi: > >> The problem with the homing report is the destruction of data. When the >> home position is found, the machine position count is set to zero (or >> whatever the HOME_OFFSET value is). So, the old home position is >> destroyed by that operation. >> >> >> > > Just that. Before writing that HOME_OFFSET just read previous value and > show difference. But doesn't work! The destruction of data is done > earlyer too! That's is done to screw_comp file come at right position! > Yes, you have to have a HAL component that records the axis position every servo cycle, and can then report the last value it saw before the home operation completed and altered the value. I know how to do this on a servo system that homes to the encoder index pulse, as there is a hal signal that goes true when it starts the search for index and goes false when it finds the index and resets the encoder count to zero. The last value recorded when the ENCODER_INDEX signal was true is what you want. I don't know how this function works with the stepgen module. Jon -- Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users