Really weird behavior of dual nozzle

Crazy new problem last week, and I’m at a complete loss. I posted this in the discord as well, and one other person is experiencing the same thing but we’ve yet to get a resolution. I’ve been trying to print with PVA supports on my A250 with dual extruder. After buying new PVA (was thinking the stuff that came with the dual extruder might have been low quality or wet, so bought new roll), I re-ran all calibrations. I did the sensor calibration first, then leveling, then the z off set then the xy offset and finally the confirm xy offset. All calibrations were done with 2 rolls of PLA. I then switched back to the PVA on the right nozzle and printed my first layer square test print to check for level and z-offset. Ended up dropping left nozzle by .25 and right nozzle by .35. Next I started trying to print a small box (normally used for flow rate calibration) upside down with supports underneath and the round wipe tower. Two things started happening. 1. The PVA would stick to the bed fine when printing the brim, but when it started printing the first layer inside the box, it didn’t want to stick to the bed and the nozzle would grab the filament and start pulling it, resulting in a big mess. I raised the nozzle temp and bed temp some and this seemed to help. I think this might just be a speed issue and should be easily resolved. The next thing is the crazy part. 2. The left nozzle prints the PLA brim and then starts printing the tower. As it’s making it’s first layer circles, the head starts lifting off the bed. On one of the prints, I watched the head lift a good 3mm off the bed as it was printing, then lowered back down to the correct height, then raise again, then lower again. On another print, I saw the nozzle doing the same inconsistent raising and lowering of the head during the brim print. My first suspicion was the sensor being dirty, so I cleaned it with a little bit of rubbing alcohol on a paper towel. I finally just shut the machine off and decided I would come back to it later…thoughts?

I am going to install Luban from scratch on a new computer and put the dual extruder on my other 250 and see if the same thing happens, though I am 99% sure this isn’t Luban related since the other report came from someone using Cura.

No real idea what is happening, but one thought: I find it highly unlikely that the nozzle would lift 3mm due to some sensor failure or so. I’d bet on this being in the G-Code and a slicer issue. Still, I’ve no idea which setting in the slicer might cause this… If it does not go away, I’d suggest to try a different slicer.

Thought the problem was gone… Updated firmware, complete re-calibration starting with tramming, and we got a completely successful print. Now this… Caught it right after, so the video doesn’t show it, but the results are pretty obvious what happened… Video

I have no idea… if it is not in the GCode, I guess this is a case for the Snapmaker support…

Same file, tried again… Did it in a different spot. Definitely not in the g code, and here’s a picture of the new video of it doing it in action.

I assume that the steppers of Z are doing the movement? Or is there any unusual sound that indicates that some mechanics are faulty?
One wild guess: The memory of the controller that holds the bed levelling data is damaged, and the bed level compensation algorithm does stupid things as a result.
To check if my shot in the wild is a hit, you might temporarily switch off bed level compensation. According to here: Bed Leveling State | Marlin Firmware you should be able by using
M420 S0
But I never tried it, may be wrong here.