Back
[00:09:49] <skunkworks_> https://www.youtube.com/watch?v=X3sKTwfrtOE
[11:14:31] <Roguish> skunkworks_: watched the video you just linked to. are the motors simple dc brushed? encoders like HP (agilent) ? Any gear reduction besides the ball screw?
[12:21:47] <skunkworks_> Roguish, simple brushed servo with heds encoders - the where bought surplus years ago..
[12:22:01] <skunkworks_> 2.5mm pitch with a 2.5 reduction
[12:22:10] <skunkworks_> (aprox 10tpi)
[12:27:52] <skunkworks_> archivist, those cheap breakout boards seem to pass 20khz pwm just fine
[12:28:21] <archivist> er wot?
[12:28:56] <archivist> the ones I use for steppers?
[13:34:19] <KGB-linuxcnc> 03Dewey Garrett 05master 06b7772 06linuxcnc 10src/emc/usr_intf/axis/scripts/axis.py axis.py: bugfix for missing slider items detection * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=06b7772
[13:50:09] <KGB-linuxcnc> 03Dewey Garrett 05master dfcf1ba 06linuxcnc 10src/emc/usr_intf/axis/scripts/axis.py axis.py fix typo err * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=dfcf1ba
[14:29:00] <skunkworks_> archivist, yes
[15:10:44] <skunkworks_> pcw_home, do you see this?
http://electronicsam.com/images/KandT/testing/emco/emcofollowng.png
[15:10:50] <skunkworks_> what are your thoughts?
[15:11:32] <skunkworks_> certainly usable as is...
[15:32:35] <skunkworks_> http://electronicsam.com/images/KandT/testing/emco/emcofollowng2.png
[15:59:11] <pcw_home> do you crank the integral term up till its unstable and back off?
[15:59:35] <pcw_home> did you I mean
[16:07:10] <pcw_home> More I might reduce the width of the error during accel/deccel
[16:07:11] <pcw_home> though the initial peak is hard to get rid of without a limited jerk trajectory
[16:07:13] <pcw_home> Also I suspect the error peak at the beginning of motion is larger that the error peak at the end of motion because of stiction
[16:36:17] <skunkworks_> pcw_home, thanks - I will play with it. it is working better than I thought it would have... :)
[16:36:31] <skunkworks_> that is 30in/s^2
[16:38:50] <pcw_home> You may have some non-linearity in the PWM also (The BOB OPTOs could easily cause a 10% or so PWM dead zone near 0)
[16:41:42] <andypugh> I think I need to do something clever with gain-scheduling on my lathe. For some reason it takes at least 1A to move the cross-slide, and it takes the pid a long time to get there with small errors. It tends to “stick” about 0.04mm either side of the setpoint.
[16:42:21] <andypugh> Leave it long enough (minutes) and it gets there.
[16:43:39] <andypugh> Not quite sure where the problem is, and the mechanics are hard to disassemble to find out. It was designed by an amateur ;-)
[17:00:59] <pcw_home> what servo thread period are you using?
[17:41:24] <andypugh> pcw_home: Me? 1000uS. But I doubt that is the issue.
[17:42:09] <pcw_home> you can get high D and P terms at a higher servo thread rate
[17:42:21] <pcw_home> higher
[17:43:23] <skunkworks_> adding a bit of ff2 seemed to balance the peaks at the beginning and the end.
[17:43:35] <skunkworks_> pcw_home, no optos in the bob - only in the drive
[17:43:41] <skunkworks_> just some buffer chips
[17:43:47] <pcw_home> OK thats good
[17:44:12] <skunkworks_> that is at 1khz
[17:44:46] <pcw_home> faster may improve things somewhat
[17:45:36] <skunkworks_> still - .0002 following error peak.. I think that is probably pretty good
[17:47:22] <skunkworks_> that is 10 encoder counts i think
[17:55:38] <pcw_home> probably swamped by mechanical errors
[17:56:07] <pcw_home> so not much to gain by getting too fussy
[18:10:03] <skunkworks_> right
[18:10:18] <skunkworks_> there is atleast a < .001 backlash I think
[18:18:08] <skunkworks_> that was running archer (radioshack) 1 amp variable power supply ;)
[18:18:13] <skunkworks_> for testing
[18:18:43] <skunkworks_> 25v
[18:19:41] <skunkworks_> I think we will use the supply out of the emco - it is 40v - pretty stout
[18:52:36] <pcw_home> you will have to retune FF1 if you change the voltage (if its a bare HBridge)
[19:06:15] <skunkworks_> sure
[19:06:44] <skunkworks_> I think the drive just passes the pwm straight through - other than current limit..
