Terrible bed leveling results but Klipper says everything's OK

[printdebugger]

Basic Information:

Printer Model: Wanhao Duplicator i3 (Monoprice Maker Select V2)
MCU / Printerboard: BigTreeTech SKR 1.4 Turbo
Stepper Drivers: TMC2208
Probe: BLTouch
klippy.log (4.7 MB)

Describe your issue:

I am struggling endlessly to get reasonably results from bed leveling on my printer with Klipper. Several hours of calibration later, this is what I have to show for it:

I have a BLTouch probe and my leveling procedure has gone like this:

1. Ensure the gantry is parallel to the table.

2. Ensure the extended probe tip is mounted below the nozzle.

3. Test the probe accuracy with PROBE_ACCURACY:

   probe accuracy results: maximum 1.995000, minimum 1.987500,
   range 0.007500, average 1.991250, median 1.991250,
   standard deviation 0.002016
   

4. Determine the X, Y offset between the probe and nozzle.

5. Use SCREWS_TILT_CALCULATE to adjust the bed screws until approximately level:

   rear left screw : x=50.0, y=200.0, z=1.89750 : adjust CCW 00:02
   rear right screw : x=200.0, y=200.0, z=1.89750 : adjust CCW 00:02
   front right screw : x=200.0, y=50.0, z=1.89500 : adjust CCW 00:02
   front left screw (base) : x=50.0, y=50.0, z=1.87750
   

6. Use PROBE_CALIBRATE and determine the offset between the nozzle and probe using the paper test. (z_offset = 2.365)

7. Use BED_MESH_CALIBRATE to probe at 25 points and create a mesh for any small deviations in the bed surface. (And save it with SAVE_CONFIG.) Range is -0.198mm to +0.010 mm (Δ = 0.208 mm).

8. Add BED_MESH_PROFILE LOAD=default to GCODE, after homing at the start of the print.

Clearly the bed is too low on the left and too high on the right, but I don’t know why this wouldn’t come across in any of the earlier leveling procedures. Shouldn’t SCREWS_TILT_CALCULATE detect that situation? And shouldn’t BED_MESH_CALIBRATE compensate for it?

(Confusingly, the mesh shows that the bed is higher on the left than on the right, which is the opposite of what appears to be the case with the print. Discourse won’t let me attach a screenshot.)

[jakep_82]

You likely have a small twist in your X axis. Take a look at this to see how to check and compensate.

[printdebugger]

That was exactly it. Thank you.