1. Configuring LinuxCNC
  3. Advanced Configuration
  5. Another servo tuning

Another servo tuning

More
11 Apr 2022 17:15 #239937 by myval
Replied by myval on topic Another servo tuning
At full bore the exe is missing steps. Test setup was dial indicator on one end of full travel open loop and if I run across full lenght few times I get still 0 on dial indicator but fair difference on encoder about 0.02mm on 10 runs. 
Just did the test at 800mm/s well bellow the max for exe and it was spot on.

so yeah the result for future generations do not run EXE box over its max input frequency.

but I would be guessing that the closed loop should be able to compansate for the wrong value. Unusable for machining but I guess I still should be able to set up servo not to overshoot as it is over and under shoots on accel decel.
 

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

More
11 Apr 2022 17:23 #239938 by myval
Replied by myval on topic Another servo tuning
So Andy if I understand correctly change from:
net x-pos-fb <= hm2_7i95.0.encoder.00.position
to this:
net x-pos-fb <= hm2_7i95.0.stepgen.00.position-fb
leaves the f-error calculation from the encoder position?
well this is strange since f-error remains the same dial indicator shows 0 but the encoder count increases slightly every test run.

I am going to have head explosion :-)

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

More
11 Apr 2022 17:27 #239942 by tommylight
Replied by tommylight on topic Another servo tuning


I am going to have head explosion :-)

We need proof ... :)

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

More
11 Apr 2022 17:37 #239943 by myval
Replied by myval on topic Another servo tuning
Will send you a pic soon. I am just falling in a bottomless hole. I think this tuning is bit over my head.

I found replacement for EXE box which runs at 500Khz at least one plus point.
The following user(s) said Thank You: tommylight

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

More
11 Apr 2022 18:18 #239952 by myval
Replied by myval on topic Another servo tuning
now I did 5 runs 0 to -390 and the results are attached.
 
Attachments:

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

More
11 Apr 2022 18:24 #239954 by myval
Replied by myval on topic Another servo tuning
Tommy it is about to happen. (the explosion)
I did about 20 runs 0 -390 and the dial indicator moved about 0.01mm
how is this possible steps in one dir - steps in other dir should equal to the 0 on dial indicator shouldn't it? in open loop setup. if there was play backlash or so it would not increase steady in one direction.

so what the? What am I missing guys? what am I doing wrong?

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

More
11 Apr 2022 19:18 #239960 by myval
Replied by myval on topic Another servo tuning
well I left it in velocity control didn't I so I guess this slight overshoot could be expected. I am guessing

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

More
11 Apr 2022 22:49 #239991 by tommylight
Replied by tommylight on topic Another servo tuning
I am just waiting for something to happen since i can not help as i never used EXE anything, well except stuff i have come to utterly hate = the exe files in windows! :)

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

More
11 Apr 2022 23:44 #239999 by andypugh
Replied by andypugh on topic Another servo tuning

well I left it in velocity control didn't I so I guess this slight overshoot could be expected. I am guessing

In that case it is working incredibly well!

There are a number of ways to configure this, and what you accidentally have might be a good one. 

ie, servo drive in position mode, stepgen in velocity mode, PID active but with P=1000 and FF1 = 1. Then the PID takes stepgen-pos-fb as the feedback. (This is the usual setup for a Mesa stepgen config, as the PID allows for better performance as the theead period varies) 

Alternatively you can put the stepgen in position mode and completely ignore the PID. motor-pos-cmd => stepgen.position-cmd 

In either case you can continue to feed the _joint_ pos-fb from the encoders, and that will give you an accurate f-error reading. 
 

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

More
12 Apr 2022 14:50 #240049 by myval
Replied by myval on topic Another servo tuning
I set it in position control without pid , motor-pos-cmd => stepgen.position-cmd 
just get this massive ferror spike but it runs ok. Hits dial indicator within about 0.01 which is fairly decent I guess would only like to bump it up when I run highspeed grinder spindle on it but that is not current issue.
This acceleration and speed is as good as the machine can go, well can go slightly faster but I don't think there is a need for more speed.

So where do I take it from here?
Why is the ferror so spiky?
How to get rid of that massive spike at the start?
 
Attachments:

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

  1. Configuring LinuxCNC
  3. Advanced Configuration
  5. Another servo tuning
Time to create page: 0.093 seconds
Powered by Kunena Forum