Autolevelling produces inconsistant first layer on large print

Hey

Im doing a print which is nearly as large as the bed but Im having some problems getting the first layer to stick.

Two of the corners (diagonally opposite) produce a really nice wide line with good contact but for the other two the nozzle is too far from the bed, the line is narrower and invariably comes unstuck.

Ive had a look through Configuration.h and there dont appear to be many settings associated with autolevelling, is there a way of making a manual calibration? If i hold a spirit level to the bed there is a slight bulge around the centre though its not consistant with the issue Im seeing

 

Thanks for the replies. I don't think the centre bulge is causing the problem. The corners where it is lifting are right on top of two of the auto-leveling sample spots so I'd have thought the nozzle height at these points would be particularly well dialed-in, I guess it's still in its infancy as a tech though.

I'll whack some tape on there and see what happens. In the long run I might try to build a custom bed as I want to extend the heated area out a bit anyway, though there is oh-so-much else to tinker with before I get to that point

 

Can you take a picture of what you mean? It's probably a calibration issue if you're using matter control.

 

Hey

this is about as good a photo as I could get. As you can see it's already come unstuck at this stage, the left hand corner has good adhesion with a nice wide track, but the right hand corner is too far from the bed. Both these corners are right on top of auto-levelling bed probe points so I'd have thought would be accurate.

I've tried pausing the print, nudging the z axis down a few tenths before resuming but this would result in no filament at the previously good corner and still poor adhesion at the bad corner.

 

Code:

; Generated with MatterSlice 1.0
; filamentDiameter = 1.75
; extrusionWidth = 0.4
; firstLayerExtrusionWidth = 1
; layerThickness = 0.3
; firstLayerThickness = 0.3
; automatic settings before start_gcode
G21
M107
M190 S50
M104 S205
; settings from start_gcode
G28
M114
G1 X0 Y0 Z5 F5000
M109 S205
G28 Z0
M114
M565 Z0
G29
G1 X0 Y0 Z0.1
; automatic settings after start_gcode
G90
G92 E0
M82
; Layer count: 16
; LAYER:0
M107
G1 X228 Y228.5 Z1.3 F9000
G1 X228 Y228.5 Z1.3
; TYPE:SKIRT
G1 X228 Y228.5 Z1.3
G1  F1800
G1 X0 Y228.5 Z1.3 E28.437 F1080
G1 X0 Y25.5 Z1.3 E53.757
G1 X228 Y25.5 Z1.3 E82.194

Interesting to see that the first z-coords are 1.3 (mm?) that would seem too high. I'd post my MatterControl settings but I'm not currently at home.

I've just discovered what the Print > Advanced > First Layer setting is for; will see if that helps tonight.

 

Can you enable terminal then send G28 and G29? Report back with the values, you might need to use the snipping tools as I don't think you can select text in their terminal

 

@Mike Kelly Here's the G28/G29:

Code:

Send: G28
Recv: ok
Send: G29
Recv: Bed x: 15.00 y: 20.00 z: -0.68
Recv: Bed x: 107.00 y: 20.00 z: -0.07
Recv: Bed x: 199.00 y: 20.00 z: 0.27
Recv: Bed x: 200.00 y: 130.00 z: 0.23
Recv: Bed x: 108.00 y: 130.00 z: -0.04
Recv: Bed x: 16.00 y: 130.00 z: -0.65
Recv: Bed x: 15.00 y: 240.00 z: -0.57
Recv: Bed x: 107.00 y: 240.00 z: 0.08
Recv: Bed x: 199.00 y: 240.00 z: 0.50
Recv: Eqn coefficients: a: 0.01 b: 0.00 d: -0.76
Recv: planeNormal x: -0.01 y: -0.00 z: 1.00
Recv: ok
Recv: echo:endstops hit:  Z:0.50

@Ziggy I'm only using MatterControl to slice the model; after that it's handed over to Octoprint so unless MC is injecting a second auto-level command into the gcode it should only be executing the G29

So the z offset is set to 1 and the first layer height is 0.3mm so that's where the Z1.3 initial coordinates come from. I've experimented with this and can can only get z offset down to about 0.85 before I'm scraping the print bed in certain places (and still failing to stick elsewhere)

 

I'm not entirely sure how mattercontrol handles autolevel. I do know it does some funny things with G-code, like adding a height value to each G1 command.

 

Hey, thanks for all the suggestions. I've switched auto-level off in MatterControl and retried a medium sized print which went a lot better than it had previously.

@Ziggy I'm setting the z offset in Printer > General > Z Offset. It seems that while the M565 always sets it to zero MC is actually offsetting every Z coordinate in the gcode, i.e. in a different print the first layer height is 0.2mm, z offset 0.8mm and the gcode reads

note I'm not stating this as fact; it requires some further investigation.

So I think my problem is an uneven bed. Auto-levelling seems to be adjusting out any tilt on the bed; not undulation, that would explain why a G29 operation produces a normal.

In theory it would be possible to write a script which compensates for undulation by probing the bed (many more than the standard nine times) and selectively putting down filament in the valleys so to even out the surface; using linear interpolation to fill in the blanks between sample points. It wouldn't need to do this across the whole surface of course, just the area needed for the print. I might have a pop at writing such a script but I need to get much more familiar with the printer first

 

Yeah I'd much prefer being able to tweak a single value rather than have that value intrinsically embedded throughout the entire code; not sure if there's a good reason why MC takes the latter approach.

I also prefer the idea of fixing the hardware (regarding the undulations) rather than trying to compensate for it in the software. How does having a sheet of glass on top effect the heated bed? Does the heat transmit okay by itself or do you use a thermal paste or similar to help?