Vertical Machining Centre Retrofit with Chinese Servo Drives Build Thread (NZ)

If you are using step/dir commands, open loop (no position feedback into Linuxcnc) then you do not need to do any tuning in Linuxcnc, the default settings should be correct. The servo loop is closed in your servo drive. The only tuning would be within the drive probably using it's configuration software.

hey Todd

I have been tracking down your old posts on this for the past few days haha

I am running in open loop like the second method you suggested in this old thread here

old thread on step and dir with encoders

I am trying to use the encoder output from the drives to graph the following error and velocity error so I can tune the drives as the chinese servo tuning software to hard to get working.

I have halscope working fine now

currently I am measuring the joint.0.f-error

also I just connected up another input from the encoder velocity

net x-vel-feedback =>hm2_5i25.0.encoder.00.velocity-rpm

that way I can graph the difference between the commanded velocity and the velocity feedback

(just realised that my scale is not accurate though. the commanded velocity is in m/min and the encoder velocity is in rpm. (better fix that)

I am getting a bit stuck I think working out what I actually want to tune. my servo drives have both a velocity Pid loop and a position pid loop. I am getting a bit stuck working out which bit to tune first. etc

currently I am just changing values in the servo drive and seeing what they change and trying to graph them.

actually just to confirm this is how I think my setup is working

the commanded position is given to the stepgens and they run a pid loop with the stepgen feedback. (this is a stock pncconf setup i think) I shouldn't have altered the tuning within linuxcnc

and then the actual encoder position goes the to DRO( joint feedback etc)

so I can see the actual position of the motor in the DRO but that doesn't change the PID loop.

if anyone has any ideas I would appreciate it

Some pointers:
Set the stepgen in position mode or default, leave all settings for that axis at default, forget the encoder and check if the machine works properly (or that axis since you have 1 encoder for now). Tune the drive if needed or there are errors in positioning.
Only after you are sure that position mode is working correctly you can continue to close the loop in lLinuxcnc by setting the stepgen mode to velocity for that axis and activating the encoder in the hal fie for the feedback. It should not need much tuning.

Hi Tommy

I will check that I have the stepgen in position mode. I think it is.

currently I have the encoder unconnected to the PID Loop but still updating the DRO so I can check the actual encoder position.

I have halscope all working ok I think and I can graph the various signals fine.

what I am actually struggling with is not a linuxcnc thing at all.

In the servo manual the tuning parameters are called different names to what linuxcnc calls things. EG I don't know for sure what is P what is I and what is D

by the look of things there is a velocity loop in the drive that needs to be tuned as well as the position loop inside the drive. As you advised I am only using the halscope to graph the error. the halscope PID is set up as default values

I will take some photos tomorrow showing how I have halscope setup

regards

Andrew

P = proportional
I = integral
D = derivative
Also search for granite devices, argon drive tuning, has plenty of info about tuning those types of drives that might help you.