Bugs in autobed leveling

I just joined the forum and could not find the problems I'm facing.
Since a few days I have installed a capacitive sensor for the auto-bed leveling and enabled the code in Marlin.
My 3D printer is based on Prusa I3.

It seems to work, except:

1. When I do a 3 point calibration of which all 3 points have the same Y value (so I could figure out if the calculations in one plane where done correctly). The whole 3D printer is not responding any more and the display shows behind the Z height all kind of strange symbols. Probably this has to do with "dividing by 0". So since I know; I do not use identical Y values any more.

2. When I do a 3 point calibration of which all 3 points have a small deviation in Y value (for instance a difference of 1 mm), the Y and Z are switched (!!!). So when I move Y 10 mm in my pronterface it moves the hot-end 10 mm (Z) down. So since I know; I do not use such small difference any more.

3. But my main problem is the corrected Z after auto bed leveling. Without auto bed leveling I have currently a difference of 1.5 mm (measured between (X,Y) = ((50,90) - (160,90)). But after auto bed leveling (G28, G29) it becomes 2.5 mm.

I checked the code with a colleague and found only one difference and that is the position of my sensor. My sensor is placed to the left of the hot-end (50mm) and a bit more in the front (10mm). Those I found on the internet all have one of the two equal to 0mm.

Could anyone help me why I have more offset with leveling?

Sounds, like you have to reduce z-probe offset?
-Olaf

Can you explain a bit more what you mean?
The sensor is 5 mm above my heat-bed (and has a capture range of 10 mm).
So in the Marlin code I put:
#define Z_PROBE_OFFSET_FROM_EXTRUDER -4.8

What should I reduce?


Roelof

Don't measure the distance between z-probe and nozzle, but the trigger-height of the probe.
-Olaf

Sorry Olaf my text was not clear enough.
The mentioned number is my trigger height.
Is there a limitation in the Marlin code on trigger height? (should it be for instance always less than 2 mm?)

So I did a few new tests today, which might give a hint where to look.

First the 3 points defined in the Marlin code:
#else // not AUTO_BED_LEVELING_GRID
// with no grid, just probe 3 arbitrary points. A simple cross-product
// is used to esimate the plane of the print bed

#define ABL_PROBE_PT_1_X 15
#define ABL_PROBE_PT_1_Y 150
#define ABL_PROBE_PT_2_X 15
#define ABL_PROBE_PT_2_Y 50
#define ABL_PROBE_PT_3_X 110
#define ABL_PROBE_PT_3_Y 80

// these are the offsets to the probe relative to the extruder tip (Hotend - Probe)
// X and Y offsets must be integers
#define X_PROBE_OFFSET_FROM_EXTRUDER -50.0
#define Y_PROBE_OFFSET_FROM_EXTRUDER -11.0
#define Z_PROBE_OFFSET_FROM_EXTRUDER -4.8

Then I did a measurement on these 3 spots before auto-bed leveling and found (using a piece of paper) the following Z values just touching the bed:
Z1 = -1.4 mm
Z2 = -1.9 mm
Z3 = -1.2 mm

Next step I used the commands G28 -> G29, which gave me the following answer:
Z1 = 6.88
Z2 = 6.12
Z3 = 4.11

Next I did the same paper distance measurement to verify on the same 3 X,Y values if the auto-bed leveling worked. But I had to correct Z with
Z1 = 0.5
Z2 = -0.3
Z3 = -3.3
So especially Z3 is extreme different.

Can anybody explain why point 3 is not correcting good?

So I just did a grid auto-bed leveling test.
Now I got the following measurements after the leveling:
Z1 = -1.5
Z2 = -1.5
Z3 = -1.5

That sounds perfect, the bed is now fully corrected.

But my question remains: Why is the 3 point leveling not working?