This post is more for letting the dev team know of a possible issue and simple fix.
Here is a basic example, When using a BTT Eddy probe and creating a 420mm x 420mm mesh using rapid_scan with essentially a 1mm probe distance, the resulting 176,400 point mesh will cause everything to lockup and cause disconnects and timer issues. Not sure if adding a pps will make this even larger and worse but it basically becomes unecessary when probing at this resolution.
The issue is essentially that stdout is being overwhelmed with prints from bed mesh calculations. Issue was easily resolved by just commenting out these prints/logs and also stops a resulting 1-2GiB klippy.log file from accumulating after a few prints for the day.
first is a print_func @ BedMeshCalibrate::print_gernerated_points
This one causes a large stream of prints which if the mesh is large enough will cause either the heater_verify error to trigger or an MCU counts error. if small enough to not cause either conditions before, you still have to wait 2 minutes while this calculated points are printing. Removing the print_func essentially removes this “blocking” and allow prints to proceed directly after completing rapid_scan. or just remove the call to print_gernerated_points but havent really dug into what all is happening within this method. It also prints on mcu connect if the bed_mesh section includes a large amount of points which causes all controllers to go offline.
second is a logging.info @ BedMeshCalibrate::_dump_points
This one just locks everything up on a probe count mismatch and we have to either wait till done printing or power cycle. Had this one occur when developing a laser line scanner for building a mesh so this case may not actually be an issue.
I know this isn’t relevant to the root cause of the problem you’re describing, but I do want to mention that eddy current scanners cannot give a “point” reading.
Due to their construction and principle of operation, they rather sample the average distance in a kind of fuzzy radius. My Beacon for example seems to have a radius of ~10mm. This can be seen if you scan near the edge of the bed, the mesh will drop off even though the probe is still almost entirely over the bed.
I don’t know what your typical scanning resolution is, but less than ~5-10mm stepover doesn’t really gain any additional precision.
Sorry i wrote the rest of this and i just want to let you know that i am not trying to be a dick. I do see a lot of hearsay and usually expect someone else to correct them but i guess “Cunningham’s Law” is no longer a thing. This also fixes the same issue in my new bed meshing that is unfathomably more accurate and faster. I was just testing the btt eddy because i have designed test equipment for an automation line using this concept but for so much more analysis and was curious about the capabilities of the ldc1612 and of such a cheap device.
I do understand that it is not a discreet point and more a distribution with the center weighted heavily and that the scan radius is essentially the whole inner portion of the coil (fill factor) but specifically strongest in the center. Depending on the coil impedance, frequency, power, and conductance of the material will determine the eddy current distribution and penetration depth. Because we are using a coil, any effects not directly in the center essentially cancel out in the coil (or result in noise). The distribution falls off exponentially from the center ~30% using an air core coil. Using a ferrite core results in ~90% of the center being sensitive.
But say worse case, ignoring basic inductive coupling and how eddy currents are distributed in a conductive medium, i am getting the average over the total 8mm, it is still an average around that point and the value for the next 1mm stepover would be the average around this point which technically you could simply derive a very close approx height for the center within 16 steps.
The edge of a build plate is a special case called the eddy current edge effect as you start getting reflections and other effects. You should take a look at what a crack in steel looks like as it effects the reading of the whole area around it. It also makes it hard to find cracks around holes. Even the lift-off effects are visible and show the variance in the PEI coating.