MESA 7I76e configuration - X-Axis travel overshoot

More
6 years 1 month ago #184287 by Xpertmill
Hello everyone,

i have a problem on my X-Axis on an open loop stepper machine. The faster I drive the more inaccurate is the positioning of the axis. At 300 mm/ min the error is 0.1mm at 600 mm/min the error is over 1 mm.

The z axis can be positioned nearly without any error. I already checked the backlash. I checked the scale value. Everything seems to be fine.

My suspicion now is in the HAL config. Since i am using a 7i76e card, the pncconf has created a PID controller for the position.

Could someone please check if there is the problem? I have attached my ini and hal.

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

More
6 years 1 month ago #184289 by tommylight
Try omitting these lines for all axis in the hal file
setp pid.x.maxerror .0005
to
#setp pid.x.maxerror .0005
save and restart LinuxCNC.

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

More
6 years 1 month ago #184300 by Xpertmill
Unfortunately this has not solved the problem. When I remove the X-axis then begins to oscillate and the error still remains.

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

More
6 years 1 month ago #184303 by PCW
Sounds like something global is wrong (servo thread frequency wrong maybe)
That is, oscillation is not possible with P=1000 and servo thread rate = 1KHz

What is you host hardware?

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

More
6 years 1 month ago #184304 by Xpertmill
The host hardware is Elitebook G850 from HP with I3 processor. I know that laptops are not optimum for realtime. But since we want to use this machine for research in weightlessness during a parabolic flight campaign (3D printing in space), we almost had to take the laptop for space reasons.

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

More
6 years 1 month ago #184308 by PCW
You could lower the P term to say 500 to see if that eliminates the oscillations,
but something global is wrong. A P term of 1/servo-period means that a calculated
position error is corrected by the next servo thread, and you need a P term of >
2/servo-period to sustain oscillations.

A halscope plot of the following error vs velocity might shed a bit more light on what's
going on. Also a DPLL pre-sample time of -50 usec is probably too aggressive for a host
with poor latency, you might try -250 or so so see if it makes any difference.

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

Time to create page: 0.068 seconds
Powered by Kunena Forum