Menu

Topic-icon SDC2130 : Glitches in position regulation

More
10 months 3 weeks ago - 10 months 3 weeks ago #29534110 by sihubert
Hi there,

I'm using the SDC2130 in position tracking mode to control the position of two linear actuators, with analog position feedback on the actuator's potentiometers. The position of the two actuators are determined by a very simple script, based on the input of a third analog potentiometer acting as a joystick.

The system is generally working, except one important problem.

Random glitches appear in the PI position regulator, resulting in the actuator suddently moving without reason. Checking the graphs in Roborun+ (attached to this message), it appears the loop error jumps without reason. Glitches appear on both motor channels, sometimes together, sometimes on only one channel, and can happen at time interval from a few seconds to sometimes tens of minutes.

For the record, I disabled the RS232 watchdog timer as well as every unused input/output. I tested with several different power supplies : a regulated lab power supply, and two different fully charged 12V batteries. I tried with and without computer connected to the controller.

This issue is critical for my application, and I could not find any answer on the forums. I've posted my configuration and script if needed for troubleshooting.

Any help would be greatly appreciated :)

Thanks in advance.
Attachments:

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

More
10 months 3 weeks ago #29534111 by sihubert
FYI, just upgraded the firmware to latest version, problem still there.

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

More
10 months 3 weeks ago - 10 months 3 weeks ago #29534117 by sihubert
Hello again,

Done a few more tests, and I narrowed it down to Position Tracking mode when Integral Gain is non-zero, and when the position command is rather constant.
The same kind of glitches happen directly after powering on the controller, or when recover from emergency stop in Roborun (ESC key).
Any idea why?

No glitches in Position Relative mode, even on power on.

After careful reading of the PID section of the manual, I also have trouble understanding this paragraph on pg. 137:

"Note that the controller uses one set of Proportional, Integral and Differential Gains for both motors, and therefore assumes that similar motor, mechanical assemblies and loads are present at each channel."

But different set of PID parameters are defined in Roborun for each motors. How is that coherent?

Thanks in advance.

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

More
10 months 3 weeks ago #29534120 by Gabriel_Isko
There is a chance that the error is being caused by electrical noise. Position tracking mode runs a very raw control loop that only runs on the difference between the set point and the measurement, while position relative mode plans a proper trajectory. From your graphs, it seems that position tracking mode was operating with correct behavior. Using position relative mode is our recommended course of action.

As far as what it says in the user manual regarding the same PID values applying to both channels, this is a mistake and I will have it changed.

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

More
10 months 3 weeks ago #29534123 by sihubert
Thank you for your reply.

Indeed the problem solved itself by changing the mode to Position Relative.

However, out of curiosity, could you elaborate more on this "correct behavior"? From the graphs, I see that Motor Command and Feedback are very close to each other (yet indeed subject to electrical noise) when the Loop Error suddently jumps to a very signaificant value. It also does not explain why these jumps disappear when Integral Gain is set to zero.

Thanks a lot for your time.

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

More
10 months 2 hours ago #29534193 by Gabriel_Isko
Closed loop track position performs a PID on the position error directly, and doesn't take any motion planning considerations into account. It has no way of correcting in any system that has any non linear noise. Things like friction and any non static inertia do not play well in this mode. While it is useful in very specific applications, it is generally not recommended in favor of position relative mode. For more detailed information about these modes, you should refer to our User Manual .

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

More
10 months 2 hours ago #29534194 by Gabriel_Isko
Closed loop track position performs a PID on the position error directly, and doesn't take any motion planning considerations into account. It has no way of correcting in any system that has any non linear noise. Things like friction and any non static inertia do not play well in this mode. While it is useful in very specific applications, it is generally not recommended in favor of position relative mode. For more detailed information about these modes, you should refer to our User Manual .

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

Moderators: tonysantoni
Time to create page: 0.103 seconds
Go to top