5i25 + 7i77 Joint 0 Following Error PID Tuning Help needed

More
02 Feb 2018 02:36 #105307 by eighthaxis
Hello all I am getting back to my 5-Axis VMC project and am having some trouble tuning servos.
I have read what I can but I just need a little help to guide me along.

LinuxCNC 2.7.0
Mesa 5i25 + 7i77 bought as a package with matching firmware already installed.
Glentek GA4567PA Servo Amplifiers velocity type +-10VDC control input
Brushed DC Servos with encoders 2000 PPR Differential Quadrature Output
Screw is 5TPI
encoder count set to 40000

I currently have the X axis Servo removed and mounted in a vise on the table with a piece of tape around the shaft.
When I jog continuously slowly +-X at 1.5IPM it works fine. If I try to increase IPM over the 1.5 point it gives the Joint 0 Following Error.
I have played with several PID and F settings to no avail.
I have tried with Mesa's sample 5i25_7i77 Example hal and ini files as well as from scratch using PnCConf Wizard with the same results.
initial PID 50, 0, 0
f0 for all f's

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

More
02 Feb 2018 02:47 #105308 by eighthaxis
Here are the hal and ini files.

File Attachment:

File Name: my_LinuxCN...2-01.ini
File Size:4 KB

File Attachment:

File Name: my_LinuxCN...2-01.hal
File Size:14 KB
Attachments:

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

More
02 Feb 2018 09:47 #105320 by RotarySMP
Have you tried opening up the following error parameters...

FERROR = 0.5
MIN_FERROR = 0.05

and then using the HALscope to see where the error occurs?
Mark

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

More
02 Feb 2018 14:50 #105328 by PCW
1.5 IPM is _very_ slow

I have seen problems like this where people swap the analog outputs to reverse the feedback ( thereby shorting out the 7I77 outputs )

I would verify that you are gettting +-10V out of the 7I77 ( you can check this by setting both min_ferror and ferror to
1 inch limits, disabling the drives (remove main power) and turning the motor shaft when LinuxCNC is running and
in the machine-on state. At a P setting of 50 (50V/inch) you should get +- 10V at +- 1 turn.

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

More
04 Feb 2018 06:13 #105388 by eighthaxis
Thank you both for your help.
RotarySMP I tried your settings of
FERROR = 0.5
MIN_FERROR = 0.05
to no avail. No noticeable difference.
I did fail to mention that when I open pncconf and perform open loop servo testing that the servos move fine at any requested speed 0-10VDC.
PCW in open loop test mode I verified analog voltage output of the 7i77 and it was within a hundredth of a volt of any requested voltage positive or negative from 0-10VDC.
I tried your suggested experiment and when I try to turn the shaft while LinuxCnc is running and the 7i77 sends out a signal to the meter but it is so fast it doesn't even register a digital value before the joint following error estops LinuxCnc.
When I rotate the servo by hand the DRO in Axis does register the accurate location.

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

More
04 Feb 2018 13:06 - 04 Feb 2018 13:13 #105397 by PCW

I tried your suggested experiment and when I try to turn the shaft while LinuxCnc is running and the 7i77 sends out a signal to the meter but it is so fast it doesn't even register a digital value before the joint following error estops LinuxCnc.


This is expected with a .05" min_ferror, set both errors to al least 1 inch when testing

Also you should set the FF1 value to 0 for initial tuning
( or if you know the expected axis velocity at 1V, set FF1 to 1/IPS_at_1V )
Last edit: 04 Feb 2018 13:13 by PCW. Reason: wrong math

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

More
09 Feb 2018 16:23 - 09 Feb 2018 16:26 #105647 by eighthaxis
Thanks again Peter for all of your help.
Using your input and a little bit more attention to detail on my side I have got the machine to successfully move without errors. It accurately moves in .001" increments and in continuous movements.
Now I am trying to get through the Servo tuning aspects and am definitely feeling challenged.
I am trying to follow info gathered from many people including you. There is just a big learning curve.
I turned the P up until it was unstable, then backed it down to where it felt stable then backed it down 30% as per your suggestion on another post.
Now I need to work with the rest of the variables. After another day of trying I will post some Halscope pictures to get more help getting the error down. I think the Mesa Hardware is amazing and I can't wait to start milling!
Last edit: 09 Feb 2018 16:26 by eighthaxis.

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

Moderators: cmorley
Time to create page: 0.158 seconds
Powered by Kunena Forum