FERROR depends on Accelleration??

More
28 Sep 2009 15:45 #880 by gtom
Can somebody tell me how axis*.ferror is calculated???
i have a setup with linear encoders and stepgen in velocity-mode, Pid is setup correctly and the system ist running fine & smooth...

Now i wanted to reduce the FERROR and MIN_FERROR to the lowest possible values but somethin strange happened:

With a LOW acceleration the initial FERROR (when the motors start to move) is really high, running the same Axis with an HIGH accelleration FERROR gets extremely small...

Can sb explain me WHY this happens???

regards, Thomas

Please Log in or Create an account to join the conversation.

More
29 Sep 2009 11:33 #888 by BigJohnT
With feedback the calucation is a simple calculation.

from source: joint->ferror = joint->pos_cmd - joint->pos_fb;

Some suggestions from the IRC are:

[06:11] <Valen> probably backlash would be my guess
[06:11] <Valen> with high accel the motor powers through it quickly
[06:12] <archivist> or stiction
[06:15] <Valen> ferror shouldn't be affected by accel

John

Please Log in or Create an account to join the conversation.

More
29 Sep 2009 15:35 #893 by gtom
Well, first i thought about backlash also, but: Backlash on this machine is alway the same... it does not care about acceleration or speed...

folowing some Halscopes:
Attachments:

Please Log in or Create an account to join the conversation.

More
29 Sep 2009 15:36 #894 by gtom
Same Machine, nothing changed except max_accel

Please Log in or Create an account to join the conversation.

More
29 Sep 2009 23:16 #900 by gtom
Well, i made 2 halscopes with the hostmot2 stepgen velocity-cmd & velocity-fb.

The frst scope was the scope with the maxaccel at a high value.
vel-cmd & vel-fb are nearly the same and ferror-lim is ramping similar...

The second scope was with the maxaccel at a low value. As you can see the velocity-fb is different from the velocity-cmd, thats ok cause the stepgen is accelerating..

I would expect the ferror-lim to be the same as the velocity-fb but it is NOT!

This ONLY happens with hostmot2 stepgen in velocity- mode.

regards, Thomas

Following the 2 halscopes...
Attachments:

Please Log in or Create an account to join the conversation.

More
29 Sep 2009 23:16 #901 by gtom
This is the second one...

Please Log in or Create an account to join the conversation.

More
30 Sep 2009 02:16 #902 by PCW
Isn't this exactly what you would expect to happen if you lower stepgen_maxaccel?
If the step generator is not allowed to accelerate as fast as the trajectories max_accel
you will get just what you have plotted here...

Please Log in or Create an account to join the conversation.

More
30 Sep 2009 04:53 #903 by gtom
I did NOT lower the stepgen_maxaccel, i've lowered the AXIS max_accel
will try it tomorrow with another setup for TRAJ,STEGGEN Max_Accels and see what happens

What is interesting: Thought that TRAJ MAX_ACCEL has to be higher than the Joints MAX_ACCEL...
As i remember there has to be a overhead for STEPGEN & TRAJ MaxAccels... !?
So i set TRAJ.MAX_ACCEL somewhat higher than AXIS_MAXACCEL -> no change
Went back to TRAJ MAX_ACCEL < AXIS MAX_ACCEL -> does not really lower the acceleration!
Set TRAJ MAXACCEL much lower than AXIS MAX_ACCEL -> Acceleration remains as defined in AXIS MAX_ACCEL

Thanks, Thomas

Please Log in or Create an account to join the conversation.

More
30 Sep 2009 05:32 #904 by gtom
Another strange behaivour:

the STEPGEN_MAXACCEL does NOT have any effect... if i set this Value in the ini-file it is ignored... If i start EMC then -> View Hal -> hm2.stepgen.X.Maxaccel is ALWAYS the same as the axis.maxaccel !!!

If this Value does not really do anything WHY is it there???

Please Log in or Create an account to join the conversation.

More
30 Sep 2009 06:15 #905 by gtom
Starting EMC

-> View HAL -> hm2.stepgen.X.maxaccel = 100

-> Servo- Kalibration -> MAX_ACCEL -> change Value to i.e 200 -> hm2.stepgen.X.maxaccel also changes to 200!

-> seems that there is no way to change hm2.stepgens MAXACCEL!

Please Log in or Create an account to join the conversation.

Time to create page: 0.163 seconds
Powered by Kunena Forum