Jon, I will be able to see the chatter/hunting in the pid.error signal in Halscope, so I will try to reduce it with less d-term and try adjusting FF2 to compensate for f-error. Right now my values are roughly p-term=8500, i-term=375, d-term=45 -
I am using the amplifiers in velocity mode with tach feedback, so if what you say is true, then I have a lot of hope of getting this right. I agree that higher count encoders will reduce the amplitude of the oscillations and make the chatter quieter and higher pitched. I will try it tomorrow and give a report. Thanks, Tom --- On Fri, 12/5/08, Jon Elson <[EMAIL PROTECTED]> wrote: From: Jon Elson <[EMAIL PROTECTED]> Subject: Re: [Emc-users] servo chatter at rest is going to drive me nuts... To: "Enhanced Machine Controller (EMC)" <emc-users@lists.sourceforge.net> Date: Friday, December 5, 2008, 9:41 AM Tom wrote: > I have all 3 axes of the Kasuga converted to EMC2 control using 500 line > encoders (2000ppr)on brush servos and Advanced Motion PWM amps. > The ballscrews are 5mm pitch. I have the ini set for 10160 encoder counts > per inch. > While pid tuning I noticed that in order to get my ferror down to below > .002" during the accel/decel ramps, I had to reduce the deadband. This was > after optimizing p,i,d,ff0,ff1,ff2. This causes the servos to continually > hunt about +/- .0002"(two tenths)while at rest which makes a godawful chatter. > The symphony of clicks, moans, pops and chatter is going to drive me nuts > eventually. Of course, I can always turn the machine off in Axis to stop the > noise - but I want to leave the machine on while fixturing, setting tool & > work offsets, etc. > Strange. I have two servo-controlled mills here, and they both hunt at times. The one with velocity servo amps and DC tachs is usually very quiet, with hunting a a several Hz rate which is not audible. (Once in a while one particular axis breaks into a 500 Hz while which is quite annoying. Touching the tach for a second damps it out.) The other one is a PWM driver controlled only by EMC, and it grumbles and clatters now and then, but usually settles after a couple second. I have found setting the D term is quite tricky, and too much D causes a lot of oscillation. I usually have stationary following error of about .0002" on the velocity servo machine, and .00005" on the PWM one. The velocity servo machine has encoder resolution of 20000 counts/inch, the PWM machine is 128,000 counts/inch. If you see an association between following error and encoder resolution, you are right. I think with proper amounts of FF2 (it takes VERY little) you can get the error down during the accel/decel, and maybe turn down P to reduce the noise. The problem is that the encoder is doubly quantized, introducing a LOT of noise into the PID loop. First, position is quantized by the encoder counts themselves, then it is quantized again by the servo loop sampling it. So, the D term, the difference between following error last cycle and this cycle, has a tremendous amount of noise at 1/2 the sampling rate. So, it is best to use as little D as possible, to turn this noise down as low as you can. Jon ------------------------------------------------------------------------------ SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------------------------------ SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users