PID Tuning

  • suraj9735
  • suraj9735's Avatar Topic Author
  • Offline
  • Senior Member
  • Senior Member
More
24 Jun 2021 07:13 #212788 by suraj9735
PID Tuning was created by suraj9735
Hello Forum,

So far I have good progress in PID Tuning. I tuned my linear actuator using Kp. FF1, and FF2 and reduced the following error within 40 micron in 50 mm travel at speed of (G01 X50 F16).

What I observe is when I change the speed or travel length I get more error than 40 micro-meter and need to adjust/retune the parameter. So I am confused what parameter should I tune?

My requirement is
100 mm travel length, Max Velocity 20 mm/sec, Following Error within 10 micro-meter, 0.1 micron overshoot, Vertical Load 5 Kg, Horizontal Load 5Kg.

So, I guess I should tune on at full-load/worst-condition G01 X100 F20 parameter and with horizontal and vertical load. Is the PID parameter will work and error will be within the limit if I run the command under the limit? Please guide me on the actual tuning practices.


I am not able to use internet in my real time Debian operating system, so unable to attach my INI and HAL file, Kindly excuse me.
Thanks & Regards!

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

More
24 Jun 2021 15:00 #212830 by PCW
Replied by PCW on topic PID Tuning
Its hard to guess what the tuning issue are without a plot of following error vs velocity

I would suggest adding a USB--> Ethernet or USB -->WIFI dongle to your system so it has network connectivity.

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

  • suraj9735
  • suraj9735's Avatar Topic Author
  • Offline
  • Senior Member
  • Senior Member
More
05 Jul 2021 14:07 #213762 by suraj9735
Replied by suraj9735 on topic PID Tuning
Dear PCW,

I have done some experiment and collected the data of tuning (pdf is attached). 

This is the max refine tune (P=10, FF1=1.001, FF2=0.0309) I have done, if I change little I am getting more error. but it still doesn't satisfy my requirement (within 10% velocity feedback error, within 5 µm following error, and within 0.1 µm overshoot ). I am getting huge error at start (due to FF2). 1 mm travel length is too short to follow F1000 feed rate, so getting high error in that case. So I need to also give the range of stage that if you want to move length L then keep your Velocity below V, otherwise more following/overshoot error happen.

Please suggest me how to do tuning from now to improve my result?

Thank You.
Attachments:

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

  • suraj9735
  • suraj9735's Avatar Topic Author
  • Offline
  • Senior Member
  • Senior Member
More
05 Jul 2021 14:21 #213764 by suraj9735
Replied by suraj9735 on topic PID Tuning
Some screen shorts is attached, file name is saved with travel parameters.


 
Attachments:

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

More
05 Jul 2021 14:57 #213766 by PCW
Replied by PCW on topic PID Tuning
My guess is that you are mainly seeing the drive characteristics
What kind of motor drives are you using?

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

  • suraj9735
  • suraj9735's Avatar Topic Author
  • Offline
  • Senior Member
  • Senior Member
More
08 Jul 2021 08:48 #214013 by suraj9735
Replied by suraj9735 on topic PID Tuning
I am developing submicron precision ball screw linear stage. And using advanced AC Servo Drive and 200 W AC Servo Motor of Plus Electronics Brand (manual attached). Does PID tuning method depend on drive type I am using? What's the minimum possible following error and overshoot limit I can achieve with my setup. Do I need to use Ki, Kd to improve the result?
Attachments:

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

More
09 Jul 2021 11:47 #214081 by andypugh
Replied by andypugh on topic PID Tuning
There is no inherent accuracy limit to PID. Any limitations are in the hardware and feedback.

If you want a precision of X um then you need to have several encoder counts per x of movement.

Then you should be able to make it work.

I would certainly expect to need some I-gain to pull the last bit of steady-state error out.
The following user(s) said Thank You: suraj9735

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

  • suraj9735
  • suraj9735's Avatar Topic Author
  • Offline
  • Senior Member
  • Senior Member
More
09 Jul 2021 13:25 #214098 by suraj9735
Replied by suraj9735 on topic PID Tuning
As I have attached in my above post regarding start and end fluctuation of following error. It mainly happen due to jump of velocity at start and end, and I am unable to control it. Looking for suggestion to remove the start and end error. I have tried with FF2 tuning. Is there any way to improve apart from FF2?


I know due to inertia the feedback velocity crosses the target velocity, but I wonder why trajectory planner do not stop the acceleration little earlier when it is about to reach commanded velocity? Do I have control on trajectory planner in LinuxCNC?
Attachments:

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

More
09 Jul 2021 13:41 #214102 by andypugh
Replied by andypugh on topic PID Tuning
Depending on whether you are sending velocity commands of torque (current) commands to the drives you might need to tune FF1 or FF2, possibly both.

Limiting the joint acceleration might also be worth considering.

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

Time to create page: 0.071 seconds
Powered by Kunena Forum