Having a problem with auto-levelling

I'm having a problem with auto-levelling - the probing sequence runs but the printer just continues as if it hadn't happened, resulting in the first layer being too thin on one side (almost non-existent) and too thick on the other (especially when the nozzle gets to the point when it's far enough over the bed to release the back-pressure). My bed isn't precisely level (but it's close) - certainly should be close enough for auto-levelling to cope.

My setup:

G&C CoreXY printer
DC42's IR Z-probe (v1.1)
Duet 0.6
Firmware version 1.14
Web interface version 1.11

Here's an extract of the start of some of my g-code:

; generated by Slic3r 1.2.9 on 2016-08-16 at 18:17:20

; external perimeters extrusion width = 0.40mm
; perimeters extrusion width = 0.48mm
; infill extrusion width = 0.48mm
; solid infill extrusion width = 0.48mm
; top infill extrusion width = 0.48mm

M107
M190 S65 ; set bed temperature
M104 S210 ; set temperature
G28 ; home all axes
G32 ; auto-levelling
G1 Z10 F5000 ; lift nozzle

M109 S210 ; wait for temperature to be reached
G21 ; set units to millimeters
G90 ; use absolute coordinates
M82 ; use absolute distances for extrusion
G92 E0
G1 Z0.300 F6000.000
G1 E-3.00000 F3000.00000
G92 E0
G1 Z0.400 F6000.000
G1 X90.917 Y36.308 F6000.000
G1 Z0.300 F6000.000
G1 E3.00000 F3000.00000
G1 X92.415 Y34.261 E3.13254 F1800.000
G1 X94.473 Y32.777 E3.26507
G1 X98.119 Y31.910 E3.46089
...

Here is my bed.g file:

; Auto calibration routine

M561 						; clear any existing bed transform
G30 P0 X20 Y20 Z-99999 		; front left
G30 P1 X20 Y180 Z-99999 	; back left
G30 P2 X190 Y180 Z-99999 	; back right
G30 P3 X190 Y20 Z-99999 	; front right
G30 P4 X100 Y100 Z-99999 S0 ; centre, and calibrate
G1 X10 Y10 Z20				; go to the front left side, raise & await next move

The G32 command runs, and it gives me the following output on the console:

Bed equation fits points [20.0, 20.0, 0.012] [20.0, 180.0, 0.192] [190.0, 180.0, -0.423] [190.0, 20.0, -1.020] [100.0, 100.0, 0.067]

I have just tried running this file with the G32 command commented out, and the print is identical to when it was included - which is why I don't believe that the auto-level results are being used.

Now there's a fair chance that this is happening because I'm misusing commands, or cancelling out the G32 results with another command - but I can't see where the problem might be.

As usual - all help and advice gratefully received...

Your files look OK to me.

After running G32, can you see the Z motor(s) turning a little as the head moves in the XY plane?

Here is a test you can do to see if bed compensation is working as intended:

1. Reset the printer or send M561 to cancel bed compensation. Then home the printer.
2. Command the head to 5mm above first bed probe point. Run G30 S-1 to probe the bed at that point and record the trigger height (the height at which probing stopped) as displayed in the web interface and on PanelDue.
3. Repeat step 2 for the other bed probe points. Observe the differences in the trigger heights.
4. Run G32.
5. Repeat steps 2 and 3. You should find that the trigger heights you recorded are now almost exactly the same at each probe point.

If you have probe X and Y offsets configured in your G31 command in config.g, then you will need to allow for those when positioning the head at each probe point.

btw I suggest you put an F parameter on the final G1 command in bed.g, otherwise that move will happen at whatever speed the last G0 or G1 command used.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Thanks David, I'll give that a try and report back. (and I'll also add that F parameter - I have been caught by that!)

OK, tried that test - first of all, the Z-axis motor is moving slightly during printing.

Here are the results (in the order: position / pre-G32 command / post-G32 command)

P0 / 1.57 /1.57
P1 / 1.76 / 1.53
P2 / 1.09 / 1.61
P3 / 0.61 / 1.64
P4 / 1.60 / 1.60

P2 and P3 are on the side where the nozzle is too low.

And... while I was running this test I noticed that the black component (phototransistor) on the IR sensor board was no longer vertical to the board. I straightened it up and tried again, with the following results:

P0 / 2.41 / 1.60
P1 / 2.63 / 1.58
P2 / 2.05 / 1.59
P3 / 1.51 / 1.63
P4 / 2.47 / 1.61

After resetting the Z trigger height I tried another print - apart from still being a little thin on the P3 corner the results were still much better than before. Maybe this could have caused my problem?

Is there any special process for setting the IR LEDs and the phototransistor, for future reference?

It looks like your P3 corner is a long way out of level compared to the others, and that bed compensation is making quite a good job of correcting for it. Are you using 3-point mechanical bed levelling? If so, then I think that either your bed plate is not flat, ot you have a twist in the Y axis - this is a common problem in Ormerod printers, but I don't know if it affects other designs.

The phototransistor and IR LEDs should be aligned with their lenses facing parallel to the PCB, that's all.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

My bed is a glass plate on top of a 3mm aluminium plate heater. I'll put my dial gauge on it tomorrow and work out just how level it really is (once I've worked out how to mount it on the carriage). Since fiddling with the IR sensor I've managed to make a few nice prints, so I must be quite close to a solution.

David,

Out of interest, just how much 'out of level' should the Duet/IR sensor be able to compensate for? I haven't measured my bed on the CoreXY yet but I'm reasonably confident that it's a lot flatter than the identical bed on my Prusa i3, which also has one of your IR Z probes but which is controlled by a RAMPS board. That one gives me consistent results, to the point where I don't even have any bed adjustments - I just let auto-levelling deal with everything.