Beacon RatOS Calibrate keeps failing

[dawn swift]

I keep getting "Sample spread too large" errors when trying to complete the BEACON_RATOS_CALIBRATE macro. Sometimes it's pretty early in the calibration, a few times it has made it to the measure gantry twist part of the calibration before erroring. I have tried heat soaking with the bed at 85c for 30mins and it doesn't seem to make a difference. I don't feel any wiggle in any of my Z-axis screws or bed mounts. I don't feel any wiggle on the beacon board or fan duct. I tried lowering the movement speed to 80%. The beacon is plugged into one of the USB3 ports on the Pi4.

Vcore 3.1 300 being upgraded. Added the Beacon (cable soldered directly to the board), a new Orbiter, EBB42g2 toolboard, upgraded to a Pi4, and am deploying RatOS 2.1 RC3 on a new mSD card. Hotend is a Mosquito Magnum. Beacon is mounted on this fan duct (in ABS-GF):
https://www.printables.com/model/490799-eva-extruder-tri-horn-duct-with-beacon-mount

Thoughts -
I don't know how to accurately measure the offset between the tip of the nozzle and the beacon. Not sure if that would make a difference for this error?
Something is wrong on my config... although it booted up that way out of the configurator. The printer boots up unable to home. Clicking the home buttons gives an error "must home axis first". I have to issue a SET_KINEMATIC_POSITION before I can home any of the axes and then it homes normally / endstops work correctly. Is this relevant to this problem?

Any info I can provide? Troubleshooting I can try?

The probed distances also seem very inconsistent. Even the BLTouch was more consistent than this. Probe samples exceed sample_tolerance
5:45 PM
Probe samples exceed sample_tolerance
5:45 PM
Probe samples exceed sample_tolerance
5:45 PM
probe at 150.000,150.000 is z=-0.029063
5:45 PM
probe at 150.000,150.000 is z=0.005312
5:45 PM
probe at 150.000,150.000 is z=0.006250
5:44 PM
Probe samples exceed tolerance. Retrying...
5:44 PM
probe at 150.000,150.000 is z=-0.030313
5:44 PM
probe at 150.000,150.000 is z=0.005312
5:44 PM
Probe samples exceed tolerance. Retrying...
5:44 PM
probe at 150.000,150.000 is z=-0.001875

[quasi shuttle]

Your nozzle looks like it may have debris. First step I would clean it and try again, that may take care of it.

[dawn swift]

I got it hot and scrubbed off all of the plastic that would come off with a folded paper towel. It's still giving Probe samples exceed tolerance a bunch of times before erroring out.

[quasi shuttle]

I usually do a little scrub with sandpaper and that has always worked when I got that message.

[dawn swift]

What grit?

[quasi shuttle]

Whatever is handy, usually 320, but only a light scrub.

[compact crystal]

I had to calibrate using a smoth bed plate, using the rugged pey one gave me the same errors

[compact crystal]

You can anlso configure a larger tolerance

[dawn swift]

I'm using smooth pei already. It does have a hot nozzle crash at about 150,150 from a few years ago though. I've tried moving the build plate ~15mm forward to move that flaw. Idk if that is enough of a difference either way. What do you recommend setting the tolerance to?

[dawn swift]

I gave the nozzle a gentle scrub with maroon (320-400) scotch brite. That didn't make any difference. I had a wild idea and I tried the textured pei print surface that I bought with the printer from Ratrig originally and never used. That bed made it all of the way through calibration!

What does that mean though? Is the calibration I did on the textured print surface invalid for the smooth one? What if I stuck a smooth pei (or pex?) sheet on the other side of the textured bed that passed?

Also, I got a "Very High gantry twist: 171.111198μm." message. I read through the Ratrig documentation, but I'm not sure I understand what the problem is.

I ran the dual surface scans with the print surface rotated 90deg as instructed and they both look and measure almost identical

[quasi shuttle]

As per RATOS instructions you were actually supposed to do the calibration on a textured bed as the nozzle can burn into the smooth plate. Someone else will have to chime in for the gantry twist, mine was good out of the box so I never had to research that.