Hello Gentlemen, this is mainly to the EMC developers. Since January I have tried to get my steppers run smoothly and evenly. I wrote several mails to the list on this problem, but received little echo. Now I seem to be near to the solution which might be of interest to the developing team.
My steppers rattled and lost steps regardless of the conditions I gave them. I've changed almost everything in the setup to get the motors run correctly. Result: No cure for the problem of "rattling steppers loosing steps" (name of my former thread). - I tried 4 different motherboards from 350 MHz Intel to 2.4 GHz AMD - I tried 3 newly installed harddisks with two different live CD's downloaded 2008 and 2009 - I tried running EMC from the live disks directly - I tried all this with two different driver boards - I tested all axis independently and in common - I made a LED test plug for the printer cable to visualize the timing with little sockets to access the signals with a scope probe - I changed the order of pulse and direction output at the pins of the parport - I commented all "reset" lines in .hal to switch off doublestep like I was told - I varied the parameters of steplen and stepspace in .hal up and down to extreme values - I had a friend (informatics expert) do all this again to exclude my person as the source of trouble (which is the most likely cause of computer problems) - I looked at the pulses at the parport with two different oscilloscopes and found that the pulses came rather irregularly (the old scope couldn't even trigger it). Finally I noticed: only 3 successive pulses (5 us) came with a constant pause in between, the fourth pause was twice as long which caused the motors to dance a 5/4 waltz no matter what speed I chose. This could be influenced only by varying the microstep rate from half to 1/4 or 1/8 step. The longer pause seemed to appear then as the eighth or the sixteenth one correspondingly (hard to get a steady picture for counting them on the scope). To put it shortly: I found no solution and came to the conclusion that nobody in the world ever tried running steppers with EMC stepconf because I never heard anything about it and nobody answered to my mails about it. All this troubleshooting took me about two months since I sometimes have to earn some money, too, and almost all my nerves. Note that I did all this with the "test axis" function of stepconf because this is a very convenient way to see the result of parameter variations immediately. Now, yesterday evening it occurred to me that stepconf apparently ignores settings in .hal and .ini files that I made with the editor when going beyond its apparent default limits. I made a micky mouse G-Code list and tested this with axis instead of stepconf, ad lo and behold! the problem was solved! (A little time jitter was still present but bearable). I resisted my instant urge to dance on the table only because there was no space on any table. Can you imagine, by the way, what my house looks like meanwhile? My wife told me that if I wouldn't get all the equipment mentioned out of the way (I piled it up in 3 rooms), she would, which could become a serious case - wars originated from lesser causes.;-)) Seriously: After all my efforts I am inclined to exclude all other causes of the misbehaviour than stepconf. Is it really true that nobody ever reported problems with the "test axis" function of stepconf? Maybe it has been fixed in 2.3? I have the impression that it must be a minor bug when counting the pulses because there is a connection with the microstep feature. I am aware that expositions now have caught all the attention of the gang. But when NAMES and EMC expo are over, please someone would have a look at this? Thank you for the learning Peter Blodow ------------------------------------------------------------------------------ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users