On Sunday 29 November 2020 20:24:27 Jon Elson wrote: > On 11/29/2020 07:02 PM, Gene Heskett wrote: > > I should be at about the servo tuning step by this time > > next week, something I have only done with steppers, and > > likely at less than optimum settings. > > > > Is there a tut on adjusting such a servo? > > You could look at my page on servo tuning at : > http://wiki.linuxcnc.org/cgi-bin/wiki.pl?PWM_Servo_Amplifiers > > It is aimed at my PWM controller/PWM servo amps, but > generally goes through the steps. > My procedure has evolved a bit. What I do is set I and D to > zero and P low (5 - 50, depends on the > system) and then adjust FF1 while observing pid.nn.error > with Halscope, and also showing encoder velocity to have > something for the scope to trigger on. When you get the > error down to small value > (.001" or less) then you may want to add a TINY amount of > FF2 to help with the accel/decel transients. > You have to find a compromise here, as friction helps with > decel but hurts with accel. > Then, turn up P until some oscillatory behavior appears, > then add a little bit of D. If FF1 is good, then > you don't need P to be real high to get good following. > > The general scheme changes a bit depending on whether it is > a velocity servo, torque servo or > a voltage servo that behaves kind of in between.
This makes sense based on what I have the spindles set for, and have extremely stiff rpm control. But what characteristic is it that actually determines what sort of a servo it is? As in torque vs velocity? I intend for this to work under cutting loads. That is why the double reduction of two worm drives in series that this will be. > With a true velocity servo, FF1 does almost all the work, > and P can be quite low. That sounds like FF1=1.0 should be the starter value once I have a valid SCALE where SCALE counts=exactly 1 full revolution of the indexer so I can ask it to do 37.55 degrees and get it. And go from there to the limits of my ability to measure it. Might have to modify my hal code to measure 1000 revs, although a hundred will take quite a while and 2 digits s/b enough. Eventually the remainder will be just noise. Thanks a bunch Jon. The picture is a bit clearer now. > > Jon > > > _______________________________________________ > Emc-developers mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-developers Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/gene> _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
