Servo tuning problems

More
14 Aug 2012 21:04 - 14 Aug 2012 21:25 #23211 by Todd Zuercher
This is my frist try tuning with LinuxCNC

I am running a 5i25-7i77 combo. With some Teknics SST-3100-UCX drives in analog torque mode. servo period is 400000 (as low as I could get it without getting rtai error messages)

I have been trying to get a basic tune on the bench with one drive and motor before I started putting it on the machine, but I absolutely can not seem to get this thing stable. The motor is very "springy" and no amount of D seems to dampen it for me. Finally I hooked it up to the machine to see if more mass and friction would help and it did. I was able to get a decent movement up (against gravity) but it still rattled like crazy going down.

The attache halscope picture was a movement with my hand on the shaft to dampen it abit. It sure did not seem to be vibrating like the picture showed, it felt like a relatively smooth movement. Without my hand on the shaft it would have began to ossilate wildly durring the movement, but it does remain still when undisterbed

Attachments:
Last edit: 14 Aug 2012 21:25 by Todd Zuercher.

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

More
14 Aug 2012 21:45 #23213 by PCW
Replied by PCW on topic Re:Servo tuning problems
In torque mode you need to add P until it oscillates then D until its stable and the repeat (at some point D will simply make it worse)

If the motors are small enough to safely turn the shaft by hand, you can get a feel for what D does by setting all the
PID parameters to 0 except D, Now you should be able to turn the motor shaft by hand and "feel" the D term (viscous damping)

Motors in torque mode will be hard to tune without a load and sometimes need as much as 4 KHz update rates

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

More
15 Aug 2012 04:45 #23222 by Todd Zuercher
That is what I've been trying, but as I add D, I can't seem to find any improvement. I can find the point that it starts to get worse though. With a P of about 100 and D at 12 I start to see a new wobble at about 1/2 the frequency of the P induced one that quickly goes out of control with further increases, but it still doesn't stop my first problem frequency oscillation.

Are there any tricks to getting a faster update rate. What kind of latency numbers do I need to see in order to bump my servo thread up from the 2.5Khz I have now. I am afraid I am not going to get 4Khz with this MB combination I am using now.

If it matters I used PNCconfig to make my basic setup.

When you say increase P till it oscillates. Do you mean oscillates when still or when disturbed or moved? I had assumed the latter. If you meant the former then maybe I just need to throw a whole lot more P at it.

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

More
15 Aug 2012 04:55 #23224 by Todd Zuercher
Can any one give detailed instructions on how to use at_PID?

I know most others haven't had much success with it, but what's the harm in trying?

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

More
15 Aug 2012 07:16 #23228 by andypugh
Replied by andypugh on topic Re:Servo tuning problems
Todd Zuercher wrote:

That is what I've been trying, but as I add D, I can't seem to find any improvement.

I have very little experience with the LinuxCNC PID, but it is not unusual for the DGain to need to be negative. (as it acts to resist the motion)

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

More
15 Aug 2012 11:07 #23231 by BigJohnT
Replied by BigJohnT on topic Re:Servo tuning problems
The only thing I've heard about at_pid was that it didn't work properly iirc. In any case the only documentation I know of is the man page.

linuxcnc.org/docs/html/man/man9/at_pid.9.html

John

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

More
15 Aug 2012 12:12 #23233 by VNR
Replied by VNR on topic Re:Servo tuning problems

Teknics SST-3100-UCX drives in analog torque mode

In this page www.teknic.com/products/servo_drive.php Teknics SST-3100-UCX is not a torque drive.
May you give me a link to the PDF user manual of this servo drive to see what can i do?... I do not promise you anything ;)

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

More
15 Aug 2012 15:40 #23246 by Todd Zuercher
If you follow the link for the drive on the page you referenced to the drive spec sheet pdf file. You will see that the UCX can do both digital step/dir or analog torque.

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

More
15 Aug 2012 18:38 #23251 by Todd Zuercher
PCW wrote:

In torque mode you need to add P until it oscillates then D until its stable and the repeat (at some point D will simply make it worse)

If the motors are small enough to safely turn the shaft by hand, you can get a feel for what D does by setting all the
PID parameters to 0 except D, Now you should be able to turn the motor shaft by hand and "feel" the D term (viscous damping)

Motors in torque mode will be hard to tune without a load and sometimes need as much as 4 KHz update rates


I tried setting every thing to 0 except D and I could not feel any damping, some small jerks and growls but no damping. If I made D to large with every thing 0 it would take off spinning.

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

More
15 Aug 2012 18:47 #23252 by PCW
Replied by PCW on topic Re:Servo tuning problems
Then something is really broken. There is no way D by itself can do this

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

Time to create page: 0.167 seconds
Powered by Kunena Forum