Scott, You might check whether pncconf is producing the proper polarity in the INI file for INPUT_SCALE (for the encoder) and OUTPUT_SCALE (for the PWM & servos).
I believe there was a bug in an earlier version of pncconf. I ran into that and had to edit the INI file by hand to make both values negative, which was required for my setup. Karl On 08/17/2011 07:53 PM, Scott Hasse wrote: > OK, so this has turned into a more interesting problem than I initially > thought it would be. I've used halmeter to verify that, for instance, > pin pid.x.enable is indeed getting enabled when the machine gets to the on > state. Additionally, the pid.x.output becomes non-zero after the machine is > turned to the on state, so I believe EMC2 for some reason thinks the motor > needs to move. > > One odd thing is that I have triple verified in the pncconf open loop test > that pressing + moves the table in the X+ direction and the encoders reflect > positive movement. Negative moves the table in the X- direction and the > encoders reflect negative movement. My encoders might not be absolutely > scaled correctly, but they are in the ballpark. This "correct" > configuration requires me to have the motor directions reversed. Under that > configuration the mill wants to run away fairly rapidly. However, when the > directions are aligned such pressing - in the pncconf open loop test > actually moves it in a X+ direction, it reaches a following limit but not > nearly as rapidly. > > With the servo amp input signal grounded, the axes do indeed move a tiny > amount, but my understanding is that is OK and the the EMC2 PID feedback > loop should correct that. > > I've put the full pncconf, hal and ini configuration files under revision > control here: > > http://code.google.com/p/sector67-sandbox/source/browse/trunk/ProjectSheetCake/#ProjectSheetCake%2Femc2-configs%2Fpncconf > > The configs are 100% generated via pncconf, no hand-editing. > > This one for now has me stumped, so I'll have to sleep on it. Again, thanks > all for the great advice. I still suspect I have a simple problem, but I > cannot figure out what. > > Scott > > > > > > On Wed, Aug 17, 2011 at 4:03 PM, James Louis > <james.lo...@gastechnology.org>wrote: > >> Scott, >> >> You got a lot of replies on a topic with a lot of fog surrounding it, but I >> need to chime in here so you don't repeat my mistake: I broke a motor >> coupling about a year ago during servo tuning. The sound almost scared the >> coolant out of me! >> I am no controls expert, but I now know why it happened. It was because I >> didn't model my own system before blindly tinkering with the ini file. >> Before doing anything I would advise knowing your Anilam servo amps. For >> example, my Yaskawa servo amps close a current loop internally, then have a >> velocity loop outside of that with adjustable P and I (via autotuning). >> EMC2 then closes the outermost position loop using only P and FF1. This is >> known as a P/PI Cascade Loop with Velocity Feed Forward. The I and D terms >> in my ini file remain at zero! >> It took a lot of reading outside of EMC2 documentation to cut through the >> fog, but it is now clear to me. There are several possible ways of nesting >> loops and combining control terms beside PID, so don't fall for the "add a >> little of this and then a little of that" approach. >> >> Jim >> -----Original Message----- >> From: Scott Hasse [mailto:scott.ha...@gmail.com] >> Sent: Wednesday, August 17, 2011 2:32 PM >> To: Enhanced Machine Controller (EMC) >> Subject: Re: [Emc-users] Mesa 5i23/7i33TA servo instability question >> >> I'm in the shop working on this right now and am wondering how to verify >> the >> PID loop is providing feedback to each axis. I've been using pncconf to >> generate the configuration, and assumed since it allows you to set PID that >> it would configure it to feedback automatically. However, since I've been >> using pncconf as a crutch, I'm not sure off-hand how to config PID feedback >> is working. I see P, I, D values in the ini file for each axis, but I am >> still seeing small amounts of servo instability. >> >> I see the instability even with the servo signal grounded. >> >> I have followed this page: >> >> http://wiki.linuxcnc.org/emcinfo.pl?Tuning_EMC2/HAL_PID_Loops >> >> and verified that the output voltages are correct, but get stumped at the: >> >> Set up PID loop add the loop and interconnect step. >> >> Thanks in advance for any advice, >> >> Scott >> >> On Wed, Aug 17, 2011 at 7:55 AM, John Thornton<bjt...@gmail.com> wrote: >> >>> The 1100M came both ways with glass scales (earlier model) then with >>> encoder feedback. Mine is the encoder variety of 1100M. >>> >>> John >>> >>> On 8/17/2011 7:28 AM, andy pugh wrote: >>>> On 17 August 2011 04:15, Scott Hasse<scott.ha...@gmail.com> wrote: >>>> >>>>> To the point, though, I am having one not-so-minor problem. When I >>> enable >>>>> the servos, they are not completely stable, and I'll get following >> error >>>>> almost immediately after enabling them. They don't run away, but the >>>>> encoders show them moving ever so slightly, and physically the motors >>>>> actually are moving. >>>> Are you retaining the velocity feedback to the amps, so that they run >>>> an internal velocity loop, and then you use velocity commands to close >>>> your position loop based on the encoders? >>>> >>>> The previous Anilam system I chatted to someone about had velocity >>>> tachs on the motors and linear scales, no rotary encoders. >>>> >>>> So, in Anilam control is the feedback from linear scales, and encoders >>>> in EMC2 mode? ------------------------------------------------------------------------------ Get a FREE DOWNLOAD! and learn more about uberSVN rich system, user administration capabilities and model configuration. Take the hassle out of deploying and managing Subversion and the tools developers use with it. http://p.sf.net/sfu/wandisco-d2d-2 _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
