For us servo folks out there, here are my two cents for what their worth: Tuning is WAY more complicated than I thought several months ago (you were absolutely right Jon). It is not a chili recipe where a pinch of D in your PI makes it the most tasty! After crashing my mill twice and breaking a motor coupling I decided it was time to really understand what was going on.
The first thing I did was to hit the books and model my system with a loop diagram using documentation from my amplifier manufacturer. My EMC2/Yaskawa system ended up being something called a P/PI Cascade Loop with Acceleration/Velocity Feed Forward. That's a current loop nested inside a velocity loop nested inside a position loop. EMC2 only controls the position loop using P, FF1, and FF2 in my particular set up. The I and D values in my *.INI file are now zero. I've also learned that "Integral windup" broke my coupling because I added a "pinch of I" to my chili. Too spicy! I am not there yet, but now when I change a value I want to predict its effect with an accurate model. The fact is that closed loop control involves a lot of math even with "dumb" drives. The loop equations are being solved somewhere! Since I am not an EE the Laplace operator and the frequency domain will remain a mystery to me. As machine builders, however, we are indebted to the EMC2 developers to have the time domain HAL Scope to view our tuning results. If you make a loop diagram and understand what to change and why, then you can avoid the chili tweaking. EMC2 continues to amaze me! Cheers. -----Original Message----- From: Stuart Stevenson [mailto:[email protected]] Sent: Tuesday, October 12, 2010 7:21 AM To: Enhanced Machine Controller (EMC) Subject: Re: [Emc-users] tuning question On Tue, Oct 12, 2010 at 4:31 AM, Richard Arthur <[email protected]>wrote: > My unqualified thoughts :-) > I've seen these spikes a lot whilst tuning the servos on my mill. I > think they are a characteristic of FF1 tuning. It is very tempting to > tweek FF1 and watch the f-error generally tumble, only to be left with > these spikes. Note from Screenshot4 that the spikes occur at the > transition from acceleration to constant velocity or vice versa. When > there is a reversal of the contribution made by FF1, it is almost > instantaneous and it simply isn't possible for the the servo to respond > fast enough. I found by improving the tuning at the amp and then EMC2's > P (I don't find I and D help much if at all), resulted in using much > less FF1 (the spike is always present, though much reduced). Having said > all that, I suspect these spikes have little or no effect on the > quality/accuracy of the machined part as the machine simply can't > respond fast enough (a point I've been unable to verify as sadly my > machine is down at the moment). > > i also find I and D almost no help. I want to find a way to tune the servo drives or to bypass the tuning component of the servo drives. I like the idea of bypassing the tuning component as that then gives me a consistent tuning interface between the machine and the man (me) no matter what flavor drive is on the machine. Hopefully this also will allow the tuning to exploit the best performance of all the components. The best scenario would be a simple (dumb) drive that only amplified the velocity signal. > Regards, > Richard > thanks Stuart -- dos centavos ------------------------------------------------------------------------------ Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today. http://p.sf.net/sfu/beautyoftheweb _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users This communication is for the use of the intended recipient only. It may contain information that is privileged and confidential. If you are not the intended recipient of this communication, the disclosure, copying, distribution or use hereof is prohibited. If you have received this communication in error, please advise me by return e-mail or by telephone and then delete it immediately. ------------------------------------------------------------------------------ Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today. http://p.sf.net/sfu/beautyoftheweb _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
