On Thursday, February 02, 2012 08:51:35 AM Viesturs Lācis did opine: > 2012/1/31 Peter C. Wallace <p...@mesanet.com>: > > This can be mitigated by putting a common mode choke (Big > > > > ferrite bead) over the three motor leads (DO NOT include the ground > > lead) > > 1) I wrapped each of 3 power wires for motor around a separate ferrite > bead. 2) I connected machine frame to common earth ground. > That isn't the ideal method, although at 50 or 60 hz, not terribly important. We normally use the chokes as common items, where all leads are wound on the same toroid form. The latter doesn't block load surge noises because those will generally result in magnetic cancellation. The general idea is to block the longitudinal noise, keeping that soundly choked off in the motor itself. Brush arcing noises etc.
I have not personally noted a problem in my lashup, but I've tried to maintain a star grounding and shield connection scheme too. But when I hooked up a probe a few days ago, I did find that I had to ground the probes shield to the machine frame, else when the probe circuit was closed to ground, halscope was showing quite a bit of noise that seemed to be related to the powerline frequency. > > Please try the encoder sample rate lowering example I provided. This > > will quickly help to see if there is a noise problem. Noise piked up > > by the encoder wires either inductively or capacitively tends to be > > short pulses. By lowering the input bandwidth a lot of this kind of > > noise can be filtered out. It also delays things, making stable setups harder to achieve because of the lag in the feedback. although if you've limited it to 3mhz with the code below, I can't bring myself to pointing a finger at it. You still have far more effective bandwidth at 3mhz than any modification of the servo drive, the rotational inertia of the motors armature mass likely limits its rate of change to well below 50 kilohertz even in an ironless motor. Ed: Is there some method he could use, using halscope, that could characterize a mechanical resonance from torsion in the drive and/or this questionable encoder & prove there is an instability causing lag someplace. The only thing I can think of ATM would need a function generator to supply a low amplitude, enough to move the machine 20 thousandths back and forth, varying this drive frequency from about 0.1hz up. If its the lag in the encoder, the lag should be both visible in halscope, and varying the frequency of the function generator might make it resonate & go crazy at a relatively low frequency. But that would require a function generator which I doubt Viesters has access to. Is there some other tool such as another pwmgen cobbled up out of hal stuff in linuxcnc's bag of tricks that could do this? Just thinking out loud, YMMV. > 3) Edited INI file, so that firmware load line is this: > CONFIG="firmware=hm2/5i23/SVSTTP6_6_7I39.BIT num_encoders=3 > num_3pwmgens=3 num_stepgens=2 enable_raw" > > > 4) Added to HAL file these lines: > > setp hm2_[HOSTMOT2](BOARD).0.raw.write_address 13312 > # = 0x3400 = encoder Sample Rate Register > > setp hm2_[HOSTMOT2](BOARD).0.raw.write_data 14 > # sample rate is clocklow (48 MHz)/SRR+2 > # so divide by 16 = 3 Mhz > > setp hm2_[HOSTMOT2](BOARD).0.raw.write_strobe 1 > # do write > > 5) In "Show machine config" I additionally ran: > setp hm2_5i23.0.raw.write_strobe 1 > > > Here's one way to check for lost encoder counts: > > > > Mark the shaft and the stator with a pencil/scribe at DRO reading of > > 0. Move around a lot or until you have the stall problem, turn off > > motor power and hand crank back to DRO 0 and check pencil mark > > alignment. This should get you to within a few 10s of counts > > 6) Scratched a mark in pulley and straight line in motor mounting plate. > Aligned so that mark was on the line. > Started up Emc. > In Hal Meter I opened encoder.0.rawcounts pin, which had a value of 0. > Jogged the joint, motor stalled at least 10 times, each time I stopped > it with F2, then pressed F2 again, jogged again until it stopped and > started to oscillate, so I stopped again with F2 and so on. > After a while I turned off motor power (encoder power remained) and > aligned the mark and the line (I could not see the screen while doing > that). > Rawcounts pin had a value -1 > > 7) I tested - jogging bare motor works just fine, which, in case of > acceleration lag in encoder, is obvious. > > The first 5 points were implemented so that all of them were in > effect, when last point was conducted. > > The observed impact of first 5 points - NONE. > > So my conclusion is - machine is not losing encoder counts, there is > no significant noise in encoder signal. > > Could it be anything else, but the encoder problem? > > Viesturs > > ------------------------------------------------------------------------ > ------ Keep Your Developer Skills Current with LearnDevNow! > The most comprehensive online learning library for Microsoft developers > is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, > Metro Style Apps, more. Free future releases when you subscribe now! > http://p.sf.net/sfu/learndevnow-d2d > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) My web page: <http://coyoteden.dyndns-free.com:85/gene> Power, n.: The only narcotic regulated by the SEC instead of the FDA. ------------------------------------------------------------------------------ Keep Your Developer Skills Current with LearnDevNow! The most comprehensive online learning library for Microsoft developers is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, Metro Style Apps, more. Free future releases when you subscribe now! http://p.sf.net/sfu/learndevnow-d2d _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
