CXW
Problems with bed Leveling, Z-probes, Servo twitching
13. February 2020 10:54
I have an Ender 3 which I upgraded with a BIGTREETECH SKR Mini E3 V1.2 running their fork of Marlin.
Bed leveling is causing problems and I would appreciate any help you might give.

My leveling sensor is a servo probe - a DIY 3D printed strut with a push switch at the end,
attached to the hot end housing with a servo which swings the strut down for readings
then swings it up when printing. This configuration is enabled in the firmware and it all works - just not very well.

The printer has been working hard for about a year, so I tightened some bolts and took up slack in the Z axis with the offset cams but that didn't help.

Bed mesh readings show differences in corrresponding points of up to 1.3mm in successive bed probes.
The contours - changes between adjacent points - remain very consistent, but their absolute height is what changes.
The obvious conclusion is trhat the probe is not placed consistentl;y between probes.

But I enabled the M48 probe repeatability test, and it shows stardard deviation of 0.06 and a range in the readings (max - min) of 0.2mm.
This is not good enough - less than 0.01 for standard deviation is a maximum acceptable.
So I have two problems - the probe is inconsistent and its placement changes bewteen deployments.

I noticed that the servo twitches and more so during the test. This moves the probe during measuring and might account for one problem.
By connecitng an external servo directly to the board and showing it still twitched, I eliminated inductive coupling and the servo itself from the causes.
Using my DSO I traced the cause to a 50 microsecond jitter froming from the board itself.
So right there, the board is producing a problem. Anyone know how to resolve it or even report it?

There is a Marlin option to power down ("detach" - actually stop the pulsing signal to) the servo when not being moved.
This removes the jitter but it causes the servo to coast past its set point unpredictably.

I'm currently working on a fudge. I will re-enable constant servo power, and modify the probe strut so that it deploys a stop against the fan housing.
I can set set the deploy angle a little to high. The hope is that the servo will contantly try to push the strut past the position where the stop stops it.
This tensions the pribe against the stop and may remove the movements I see with power on, and the coasting I see with power off, but it is an ugly fudge and puts more strain on the servo and thr power supply.

I will not be able to test this today, but even if it works, I don't like it, so any insights or better isdeas would be welcome.

Thanks
Chris
In diesem Forum dürfen leider nur registrierte Teilnehmer schreiben.

Klicke hier, um Dich einzuloggen