SV07 Printer bed will not level

Sovol Klipperized SV07
,
1

Hi,

I have had my SV07 working well for many months. Recently, the hot end began lightly scraping the print bed on one side (the right when looking from the front). This causes the first layer to fail.

I have reset the Z offset, levelled the bed using the adjustment wheels and paper, adjusted the z-tilt and created and saved a bed mesh. I have done the above multiple times, yet the same pattern occurs: the nozzle contacts the right side of the bed.

I am at a loss as to how to proceed.

Any suggestions?

TIA.

2

Check the preload on the rollers for the X axis rails. The bolt on the lower roller on my SV07 wasn’t very tight. After adjusting the eccentric nut and and tightening the bolt the machine behaves better.

3

Thanks for the hint, @cardoc, unfortunately, that did not make any difference.

I have lost track of the number of times I have levelled the bed, etc. It consistently scrapes along the surface on the right hand side of the build plate no matter what.

The consistency seems important, but I can’t say what the issue is.

Even if the bed was slightly out of level, the adjusting of the z-tilt and the creation of a bed mesh I would have thought should have adapted for any slight irregularities.

Any other thoughts?

TIA

4

Long shot… Check the screws at the right Z motor coupler.

Other than that it seems that the right Z motor is loosing counts BUT if that is happening the printer should constantly fail the Z tilt macro.

Have you reached out to Sovol support? Maybe they have seen this before.

5

Thanks, @cardoc

I have tightened the screws at the right motor and at the top and bottom of the gantry.

Interestingly, I was trying PROBE_ACCURACY and had some unusual results:

PROBE_ACCURACY at X:83.000 Y:210.000 Z:5.000 (samples=10 retract=2.000 speed=5.0 lift_speed=5.0)
probe accuracy results: maximum 1.387500, minimum 1.377500, range 0.010000, average 1.384750, median 1.385000, standard deviation 0.002610

PROBE_ACCURACY at X:0.000 Y:210.000 Z:3.385 (samples=10 retract=2.000 speed=5.0 lift_speed=5.0)
probe accuracy results: maximum 1.372500, minimum 1.362500, range 0.010000, average 1.369500, median 1.370000, standard deviation 0.002693

PROBE_ACCURACY at X:185.000 Y:210.000 Z:21.000 (samples=10 retract=2.000 speed=5.0 lift_speed=5.0)
probe accuracy results: maximum -0.010000, minimum -0.460000, range 0.450000, average -0.062500, median -0.018750, standard deviation 0.132646
probe accuracy results: maximum -0.002500, minimum -0.007500, range 0.005000, average -0.005000, median -0.005000, standard deviation 0.001936
probe accuracy results: maximum 0.005000, minimum 0.002500, range 0.002500, average 0.003250, median 0.002500, standard deviation 0.001146
probe accuracy results: maximum 0.005000, minimum 0.005000, range 0.000000, average 0.005000, median 0.005000, standard deviation 0.000000
probe accuracy results: maximum 0.007500, minimum 0.005000, range 0.002500, average 0.006000, median 0.005000, standard deviation 0.001225

Because of the high value at (185, 210), I repeated the test in that one location. The results are inconsistent.

Any thoughts.

I might see what happens if I level the bed manually, without using the probe…

6

Interestingly, after levelling manually, I tried to reset the Z Offset. It failed:

Probe samples exceed samples_tolerance
probe at 83.000,130.000 is z=1.512500
probe at 83.000,130.000 is z=1.237500

I’m beginning to suspect the probe. However, how does that fit with having a repeatable pattern whereby the right-hand side of the bed did not print properly? If the probe was the problem, wouldn’t the effect be more random?

7

There are 2 things that come to mind.

First (which I solved by increasing roller preload on the X axis):
The motion system is causing the toolhead to end up “high” after a fast move. My SV07 would fail the Z-tilt routine about 50% of the time. Playing with the PROBE_ACCURACY I noticed the first sample following a move to the right (positive X) was always the outlier and if I did multiple tests without moving the result was consistently excellent.

Second:
The probe is an is based on an eddy current device. Rather than send the raw eddy output it has internal circuitry that creates a switch signal. SO temperature, magnets, and the edge of the plate can effect the probe.

Run PROBE_ACCURACY several times without moving the tool head at various locations.
Does the “spread” consistently get better after the first result? If so tighten up your motion systems.
Is there a specific location that gives bad results? Klipper allows you to “forbid” probing at known trouble spots (you’re on your own, I’ve never had occasion to use that feature).
Probe throws random bad samples? Replace the probe.