Anyone ever have some crazy issues with Auto bed leveling?

Anyone ever have some crazy issues with Auto bed leveling? I got some weird behavior going on and am really loosing my mind. First off, nozzle to bed offset after a probe is 2.1. Probe has not moved, bed height is not different. I run a probe center of the bed, measure offset, set it with m851 z-2.1, run a print and it starts printing about 10mm above the bed. Adjust offset with m851 1 of three things will then happen. 1 No change 2 I get lucky and it prints but over compensated on the bed height so first layers are smashed. 3 Prints at even higher location then previous attempt. I have gone through so many offset changes, resets, reboots. Probing always results in same offset +/- 20 micron so I am 99% positive its not the inductive probe. Anyone have any thoughts?
Oh latests marlin, downloaded 4/26
Printer was working great until yesterday, was hit/miss until this afternoon where its been mostly miss all day.
Probe is LJ 18A3-8-Z/BY

A common way to do the start g-code for “auto tramming” (lol not gonna start the debate here again) is to insert a G29 line right after you call a G28 home command. This will probe the points you have set in firmware and should compensate appropriately. Check out @Thomas_Sanladerer ​’s video about this for a bit clearer explanation. Happy Printing

Hmm, check your belt tensions and/or stepper voltages? You might be skipping steps or having belt slippage

Thanks guys, @Ishaan_Gov ​ g29 is in start code after g28, probes happen as expected problem happens as print begins. @Ngarewyrd_Shurasae ​ belts are golden, no missed steps, z couplers are tight.

@Jerry_Barr ​, do you have ENDSTOPS_HOMING_ONLY enabled in Marlin (don’t know the exact variable name)?

Fwiw, this is a new behavior on a previously fully calibrated and working quite well machine, I am suspecting either marlin, ramps, or the mega to be at fualt in some way at this point. Just not sure how to diagnose.

What’s the M500 readout for the probe offset? Just confirming, sorry if this is a redundant question

Oh, @Jerry_Barr ​, instead of setting z-offset with M851, try setting it with M212

M501 reads z-probe ofset at z-3.2 which is what LCD reads as well, have tried ± multiple settings with little or no change at this point, randomly it will start to print higher or lower depending on what it wants to do, z position is always reported as the layer height. Endstops are only enabled for homing. I just uploaded marlin again to be safe with same results.

One possibility is that the wire from the level probe to the board may have developed an intermittent fault.

If this is the case, then the motion of the extruder head can cause random behavior in the probe’s electronics. The worst part is that this type of fault is very hard to diagnose.

If the problem persists, it may be worthwhile to rewire the probe, just to eliminate that as the possible source of your problem.

I also have similar issues with a new version of marlin. From what I have gathered from github two weeks back or so, there are some math errors in the code somewhere and auto bed levelling is simply broken at the moment.

@Peter_Hertel ​ I heard that to, but was working fine no coding changes. @Paul_Gross ​ I’ll double check but probes are performing as expected. @Ishaan_Gov ​ m212 is ignored, I can set the offset as I should be able it just seems as if its being ignored or overridden somewhere

If anyone is interested, I am 99% positive its a Marlin issue, made 0 changes, uploaded Repitier firmware, calibrated offset and I am printing again. I am going to try reloading a fresh marlin later to confirm. Still not sure why it surfaced after a week of working fine.

@Jerry_Barr , After the M212, you will need to save to the Printer’s EEPROM using M500; M212 is just temporarily storing the probe offset

M212 is not a valid command in the version of marlin I am using, its mostly found in printerbot forks, after some digging this is a known bug.