Auto bed leveling oddities

[Chadd]

I notice that when I use complete auto bed leveling I often have issues with the first layer being stuck well in some areas and not enough in others.

If I take the exact same Gcode and remove the G29 so it doesn't map the bed the first layer prints fine without making any changes anywhere. I have leveled the bed manually with feeler gauges and it is pretty level.

Just wondering if anyone else notices this and has any idea what could be causing it?

My first thought would be inconsistent z-probe, I have seen where people have had actual position data from the bed mapping process. What is the easiest way to get the individual position data from the bed mapping?

Thanks,
Chadd

[Roxy]

Where did you get your Marlin code? If you have the Enhanced G29 code you just issue a G29 V4 to get all the position data. And you might want to check if you have the M48 command. That will test the accuracy (repeatability) of your probe and axis hardware.

Usually, when you have issues with the first layer going down nicely with Auto Bed Leveling, it is because you are not sampling enough points. Some people swear by the 3-Point leveling, but my experience is when you get too far from a sample point, the nozzle is the wrong distance from the glass. Doing a 4 x 4 Grid usually cures that problem.

[Chadd]

This FW was from not too long ago when the thread was going on about ABL with the RUMBA board. I am sampling 9 points on the bed. The odd thing is that it is always toward the center of the bed that the first layer is not pressed down far enough when I use G29.

Thanks for the tips, I have a few things printing right now but will try your suggestions as soon as they are finished up .

Where did you get your Marlin code? If you have the Enhanced G29 code you just issue a G29 V4 to get all the position data. And you might want to check if you have the M48 command. That will test the accuracy (repeatability) of your probe and axis hardware.

Usually, when you have issues with the first layer going down nicely with Auto Bed Leveling, it is because you are not sampling enough points. Some people swear by the 3-Point leveling, but my experience is when you get too far from a sample point, the nozzle is the wrong distance from the glass. Doing a 4 x 4 Grid usually cures that problem.

[Roxy]

This FW was from not too long ago when the thread was going on about ABL with the RUMBA board. I am sampling 9 points on the bed. The odd thing is that it is always toward the center of the bed that the first layer is not pressed down far enough when I use G29.

It would be useful to do a G29 n 5 T and post the results here. (Assuming you have the Enhanced G29...) Almost for sure, you will be able to see what the bed looks like and why some areas have issues.

[Chadd]

It would be useful to do a G29 n 5 T and post the results here. (Assuming you have the Enhanced G29...) Almost for sure, you will be able to see what the bed looks like and why some areas have issues.

Here are the results of the G29 n 5 T

SENT: G29 n 5 T
RECEIVED: Eqn coefficients: a: 0.00 b: 0.00 d: 0.62
RECEIVED:
RECEIVED: Bed Height Topography:
RECEIVED: --0.00246 +0.01879 +0.02154 +0.01879 --0.01421
RECEIVED: --0.07346 +0.02904 +0.05879 +0.04304 --0.03171
RECEIVED: --0.11346 --0.01746 +0.02329 +0.03054 --0.02171
RECEIVED: --0.11571 --0.01196 +0.05629 +0.06604 +0.02254
RECEIVED: --0.08946 --0.03571 +0.00804 +0.05879 +0.07179
RECEIVED:
RECEIVED: planeNormal x: -0.00 y: -0.00 z: 1.00
RECEIVED:
RECEIVED:
RECEIVED: Bed Level Correction Matrix:
RECEIVED: 1.000000 0.000000 0.000353
RECEIVED: -0.000000 1.000000 0.000021
RECEIVED: -0.000353 -0.000021 1.000000
RECEIVED: ok
RECEIVED: echo:endstops hit: Z:0.66
endstops hit: Z:0.66

[Chadd]

Here are the results of the M48

SENT: M48
RECEIVED: M48 Z-Probe Repeatability test. Version 2.00
RECEIVED: Full support at: http://3dprintboard.com/forum.php
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED:
RECEIVED: Mean: 0.719800
RECEIVED: Standard Deviation: 0.001030
RECEIVED:
RECEIVED: ok
RECEIVED: echo:endstops hit: Z:0.72
endstops hit: Z:0.72

[Roxy]

RECEIVED: Mean: 0.719800
RECEIVED: Standard Deviation: 0.001030

Very Very good! And the other numbers say you have a high spot at the center of your print bed.

[Chadd]

Very Very good! And the other numbers say you have a high spot at the center of your print bed.

LOL it is probably a build up of glue stick in the middle of the bed causing that.

What do the bed level correction matrix numbers mean?

[Roxy]

What do the bed level correction matrix numbers mean?

If your bed was perfectly level, the correction matrix will be a diagonal matrix of 1's (with 0's every where else). But as it has to correct for issues you will see the numbers get bigger and bigger (from 0) and the diagonal will shrink from 1's to .9 What is really happening is the matrix [X,Y,Z] is being multiplied by the correction matrix to produce the new coordinates [X',Y',Z']. If everything is perfect the X number gets multiplied by 1 and no Y or Z component (because it is 0) gets added to the new X' number. But as the bed gets tilted, X' is equal to X*.99 + Y*.001 + Z*.001 (or how ever your bed is tilted).

[Carrot_or_Stick]

Never mind, i switched to Repetier firmware