PID tuning for LS Mecapion setup with MESA 7i77

More
25 Mar 2020 09:11 #161407 by alex.statie
Hello everyone!

Trying to PID tune the following setup:

L7SA004A Driver and APM-SB04 motor controlled by Analog Speed input Mesa 7i77.

The setup works overall, but the motor encounters periodic shocks.

Attached the hal scope:



Any idea what causes this shocks?

Thanks!
Attachments:

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

More
25 Mar 2020 09:56 #161409 by PCW
Hard to tell if this is a noise issue or a tuning issue
Can you post your hal/ini files?

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

More
25 Mar 2020 10:18 #161411 by alex.statie
Attached.

The "problem" is on Axis Y - joint 2

File Attachment:

File Name: plasma.hal
File Size:11 KB

File Attachment:

File Name: plasma.ini
File Size:9 KB


Thank you!
Attachments:

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

More
25 Mar 2020 10:37 #161414 by PCW
Just to test whether this is a noise issue, can you set the encoder sample frequency to 1 MHZ (1000,000)

Note that if the drives are in velocity mode, FF1 does 99% of the tuning

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

More
25 Mar 2020 12:26 #161421 by alex.statie
Could you please give me just a bit more info on how to do that?

Should i create a new thread with higher frequency and assign the halscope to that thread?

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

More
25 Mar 2020 14:12 #161431 by PCW
setp hm2_5i25.0.encoder.muxed-sample-frequency 1000000

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

More
25 Mar 2020 14:28 #161436 by alex.statie
Thank you!!

Added the config in the hal file and rerun the test.



By the way, also triggered a servo auto gain tuning in the meantime - nothing has changed.
Attachments:

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

More
25 Mar 2020 14:34 #161439 by PCW
Can you lower the P term to say 0.5 to see if the noise is feedback related?
also including the complete halscope screen would help (so the time/div is visible)

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

More
25 Mar 2020 14:44 - 25 Mar 2020 14:45 #161443 by alex.statie
Lowered P to 0.5

Had to run at a much lower speed - from F3000 to F300 - otherwise i would trigger following error with this P value.

Here is the result:
Attachments:
Last edit: 25 Mar 2020 14:45 by alex.statie.

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

More
25 Mar 2020 14:58 #161447 by PCW
Note: you need to tune FF1 to lower the following error, this should be done at
a low P value

Use the calibrate menu and adjust FF1 so the following error is minimized
If error is in the same direction as the velocity (like the plot above) you have
too little FF1, If the error is in the opposite direction of the velocity, you have too
much FF1

This should allow you to get to the previous speed

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

Moderators: snowgoer540
Time to create page: 0.257 seconds
Powered by Kunena Forum