Mesh bed leveling incorrect outside of measurement area
I built my MK3.5 two weeks ago and it's been performing great so far!
Today I tried to make a part that takes almost the full width of the build-plate and got poor first layer performance on the left side of the build plate, beyond where the mesh sensor measured (at approximately < ~55mm from the left edge of the plate). Observing the z-axis screws during the print, I could see that as the head crossed that x < 55mm mark the screws slightly raised the nozzle, so this seems likely to be a bed-leveling compensation issue.
What I suspect is that there's a mesh bed leveling extrapolation issue in the MK3.5 firmware. Because the inductive sensor is to the right of the nozzle, it can't get all the way to the left edge of the build plate to take a proper reading. Seems like something is going wrong which is causing the first layer's height to be incorrect outside of the leveling measurement area.
I'll post some pics of the bottom layer once this print is done. The first layer was otherwise flawless except for points x < ~55mm. I tried re-leveling the bed, and restarted the print several times before just deciding to let it run, so it seems repeatable for me. Anything else I should try to diagnose this issue?
RE: Mesh bed leveling incorrect outside of measurement area
https://github.com/prusa3d/Prusa-Firmware-Buddy/issues/3772
There's a bug here where others have reported seeing the same behaviour. Might be worth bringing it to the attention of support too
RE: Mesh bed leveling incorrect outside of measurement area
That looks like the same issue. Thanks for pointing me that way
RE: Mesh bed leveling incorrect outside of measurement area
It appears that it’s skipping the entire left side of the bed. It’s only probing 6 spots left to tight and 7 front to back.
RE: Mesh bed leveling incorrect outside of measurement area
Glad I found this post. I upgraded FOUR of my MK3s machines to 3.5 and I can't use any of them now. The mesh bed scan is definitely off and I'm not even sure if it's actually compensating when trying to print the first layer. Turns out the left side of the print bed was so much higher that it gouged my build plate and ruined it.
Everything on my machines worked properly before the upgrade. Nothing physically was altered on the machines at all so it's got to be the firmware right? Especially since we've seen that the bed leveling scan is only probing 42 points instead of 49. Nearest I can tell, the theories I've read on GitHub are probably correct, that there's no offset programmed in to accommodate for the PINDA since the MK4 use the nozzle to probe, there is no offset. Clearly, this code was not altered for the 3.5.
Also is it just me or does it seem like there's less options to make manual changes compared to the MK3s? For example, I could not complete a proper first layer calibration because I couldn't lower the nozzle enough to get down to the build plate. The maximum you can lower it in the first layer calibration is -2mm. There doesn't seem to be any way to manually set where home should be for the Z axis. My PINDA is at the perfect height, still the same when it was working flawlessly on the MK3s. This does provide some more insight as to the wrong code being used for the bed level probing too.
The reason the Z offset is so high on the 3.5 is because the MK4 code has it set higher to accommodate for the nozzle being the probe, which would sit lower than the PINDA. I worked around this by raising my PINDA about 1.5mm which forces it to bring the nozzle closer to the bed.
In the end my question is, how long do we now have to wait for a firmware fix? I have four machines I need to use and until the bed leveling is addressed I am at a standstill.
RE: Mesh bed leveling incorrect outside of measurement area
My best guess here, but everything points to the lack of proper offset info being programmed in for the PINDA. The MK4 uses the nozzle and does not require an offset. If you look at the way the MK3.5 probes the bed, you can tell that the probing is off on the X axis by the same distance between the nozzle and PINDA. This causes it to drop and entire row of probe points. Also with what I've experienced the lack of a proper offset for the z-axis also causes the nozzle to end up a bit too high.
RE: Mesh bed leveling incorrect outside of measurement area
It looks like they put the offset into the firmware for the probe offset, but they might have left out extending the bed size for probing. A friend of mine was helping me look through the firmware and this seemed like a likely culprit. I don’t have the skills to compile a firmware so I can’t test it. Look at the attached screenshot comparing the mini to the 3.5. 
RE: Mesh bed leveling incorrect outside of measurement area
I have the exact same problem with a MK3S+ upgraded to MK3.5. The printer is pretending to do a 7x7 mesh bed calibration, but only actually doing a 6x7 calibration. The left 10% of the 3x3_cali_rl print is way too high, while the remaining 90% is way too squished.
My hypothesis is that the 7x7 mesh bed leveling is properly calibrating the right 6x7 grid and using a default -2.000mm offset for the left most column. I tried an experiment by lowering the SPINDA probe to the thickness of two sheets of paper instead of using a zip tie. The first layer calibration is consistent all the way across, although it is printing too high. This confirms that the suspicion that the uncalibrated left row of calibration points defaults to -2.000.
The printer would likely work fine if I had a taller nozzle and the SPINDA probe could be set for a calibration setting of -2.000. 🙂
It also means that the firmware should be easy to update. All they need to do is make the printer run a full 7x7 calibration.
For now, I will set the SPINDA probe back to the old height and only run prints on the right 200mm of the bed. The first layer calibration will focus on the lines on the right, ignoring the back and forth square on the left.
Steve
RE:
Is the x-axis parallel to the bed? This thought is independent from potential issues with the MK3.5 firmware.
I recently observed a bad first layer on one of our MK3S+. Our beds are spring modded so that meshbed levelling has to correct just the rest. I attached a laptop to USB and fired up an old fashioned terminal program (like pronterface). Entering G80 starts meshbed-leveling. After it has finished enter G81 so that the grid matrix with the measured heights is written to the terminal. The numbers indicated that the x-axis was slightly tilted. Maybe somebody has accidentally rotated one of the z-spindles. After a z-calibration where the axis bumps into the upper stops everything was pristine again. Please note, that I had added a vertical M5-screw into each upper end stop in order to fine adjust the horizontal alignment of the x-axis just a fraction of a mm deeper if necessary.
If you want to really observe the z-axis doing meshbed corrections, you can let a dial gauge touch the x-motor-support from below. Quite eye opening.
RE: Mesh bed leveling incorrect outside of measurement area
The Z-height was calibrated to the top end-stops. I didn't measure the exact height of the X axis rails relative to the bed under the assumption is that the mesh bed leveling would correct for any slight variation.
Here is the latest calibration square. Ignore the lower row, I was doing a live Z adjust on the squares marked 1600 (too low) and 1500 (too high). Everything else printed at -1.550mm. It is consistent all the way across until the left most column where the printer rapidly adjusts because it thinks it measured a SPINDA reading of -2.000.
Steve
RE: Mesh bed leveling incorrect outside of measurement area
Please have a look into the link @blue posted in this thread. It seems that there still is indeed an error or a "not yet implemented feature" in the firmware.
RE: Mesh bed leveling incorrect outside of measurement area
I also seem to have the same problem. I want my nylock mod back 🙁
RE: Mesh bed leveling incorrect outside of measurement area
Here is a quick follow-up image showing the symptoms on my Mk3S+ upgraded to MK3.5 with the 5.2.1+13903 firmware. I printed a single layer 240x200x0.2mm rectangle using input shaping. The 200mm on the right looks great. The 40mm on the left where the calibration was skipped was printed way too high. The failing region is very drastic occurring on left most 1/7th of the print bed.
I will treat my MK3.5 printers as 210x210mm printers until a firmware fix is released. My guess is that this is a relatively easy fix. It certainly is easy to see if the printer does a complete 7x7 calibration. Print quality on the good portion of the bed is excellent.
Steve
RE: Mesh bed leveling incorrect outside of measurement area
Someone on the Prusa GitHub already opened the firmware, changed the offset parameters and recompiled it into a custom firmware that addresses this issue. It seems to be an easy fix all things considered. Makes me wonder why it's taking so long to get a patch or new FW release from Prusa to fix this.
RE: Mesh bed leveling incorrect outside of measurement area
Yeah every body got the issue.
For the moment ensure printing on the right side
RE: Mesh bed leveling incorrect outside of measurement area
FYI the 5.2.2 firmware is out which fixes this issue, https://github.com/prusa3d/Prusa-Firmware-Buddy/releases/tag/v5.2.2
I upgraded both of our mk3.5 printers and they are now printing the first layer calibration as usual and doing a 7x7 mesh level check.
RE: Mesh bed leveling incorrect outside of measurement area
Great news! I just updated
RE: Mesh bed leveling incorrect outside of measurement area
I upgraded my MK3S+ to 3.5 today and also upgraded the firmware to 6.0.0 right away. I recognized that the bed levelling only measures a 6x7 mesh and found this thread.
But different than described here, my prusa misses the right column.
Any suggestions how to solve this?
RE: Mesh bed leveling incorrect outside of measurement area
The 6.0.0 firmware does selective mesh bed leveling around the target print area. Were you printing an object that only filled the left side of the screen?
RE:
Thanks for that hint, but not quite sure if that's the whole story.
The print was a quickly set up first layer test: 162mm x 162mm centered on the print build. So the x-range of the print was from 44mm to 206mm. Laying a 7x7 mesh over the print area the last column covers 208 to 250mm. Indeed outside of the used area. But the first column ranges from 0 to 42 mm and would be also outside the used area. So I would expect, the printer should have used only a 5x7 mesh, leaving the first and last column out.
But the first column was measured.
I 'll test this tomorrow and place a print to the right edge.