Accelerometer as Z Probe sensor?

oliof

Has anyone tried using an adxl345 or similar mounted to the print head as a Z probe? Either as an analog probe like the mini or probe or as a simple digital probe?

oliof

Looks like this will be in Klipper soon https://github.com/KevinOConnor/klipper/issues/3741#issuecomment-759105616

Apsu created this issue in KevinOConnor/klipper

closed [FR] Use accelerometer as Z probe #3741

arhi

@oliof said in Accelerometer as Z Probe sensor?:

adxl345

I never tried adxl345 but some ~10 years back I used some analog accelerometer ( ADXL330KCPZ + MCP6284 amp, so I guess similar to 345 only analog) attached to my PIC32MX based board to try to

use it as endstop
use it to detect "problems"
use it to detect "too many vibrations" and slow down print

results were:

I never managed to get it to reliably work as endstop
I never managed to train any of the detection systems to detect "problems" (skipped steps, failed prints, blocked axis ...)
I had huge print quality improvement by reducing print speed. the algo was rather simple, originally it was using 3 analog inputs but then I added a quad comparator and or gate so if any of the 3 analog inputs were to go over the threshold set by the pot on the comparator the output of the or gate would go high - if any "high" input was seen on the "Detection" pin the "max speed" for XY was dropped by 10%, so after few layers print would slow down to a crawl... then I modified it to "if there's more than 3 impulses inside 200ms slow down by 10%" that worked lot better...

Anyhow, that firmware was POS (no acceleration, no jerk... ) so after I moved to a firmware written by someone else I gave up on the accelerometer as it was not needed any more as accel/jerk settings solved the problem I was having

oliof

@arhi that's interesting but the progress made in the bug I linked above makes me somewhat optimistic that this could work out (-:

arhi

@oliof well for starters "me not making it work" does not in any stretch of imagination means "no one can make it work" ... but I have my reservations about repeatability and precision as for e.g. in theory piezo disks should be the perfect endstops and should be way more precise and more repeatable than anything else out there but I know ppl that are finally happy with their printers after moving away from piezo... same here, I don't think it's solving "a problem", there's enough great sensors out there today, adding a new way to detect boundaries is not bad thing but hardly something that we need. with all the fancy stuff we did in 12 years of reprap project I don't see a significant increase of precision over simple, cheap, microswitch