bed height issue with auto level

Hi all.

I've a strange issue with my auto bed level.
I'm using Marlin (latest version) with my Geeetech I3X.

I've mounted a capacitive sensor and correctly configured the auto bed level feature (2 points grid).
Apparently all seems fine (G29, Z axis compensation during print and small objects).

I've tried to print my first large object and I noticed that the first layer is not the same across the bed.
On the bottom right side the compensation seems not to work, while in the center (and in the left part) it's much better.

So I printed a bed calibration object, attached the results.
As u can see, the bottom and mid right squares are not uniform, as the bed is too far.

These are my configuration constants:

    #define LEFT_PROBE_BED_POSITION 0
    #define RIGHT_PROBE_BED_POSITION 131
    #define BACK_PROBE_BED_POSITION 150
    #define FRONT_PROBE_BED_POSITION 75

I cannot test it with 3 points grid since my sensor is a bit far from the extruder and during compilation I've an error:

"The X axis probing range is not enough to fit all the points defined in AUTO_BED_LEVELING_GRID_POINTS"

Any idea (apart of manual calibration )?

Thanks!!

I assume your probe is at the right side of the hotend? The space you are leaving without probing is too big and probably the reason for the failure. Can you move your probe or increase a little the coordinates?

7 cm on both front and right is too big.

Thz ggherbaz!

Sensor is on the left side of the noozle (made with this design ) :

 #define X_PROBE_OFFSET_FROM_EXTRUDER -65
 #define Y_PROBE_OFFSET_FROM_EXTRUDER -27

(I assumed that left is with minus sign..)

So should I decrease the front and right values?
How much?

Sorry I'm lefty.

At the front you have it at 75, reduce it to as close to 10 as possible, back is at 150 make it as close to 180 as possible, left is at 0 so it's ok and right is at 131 you need to pass the 150 mark to get a decent matrix 170 will be best.

Thz again ggherbaz!

I tried with these new values:

 #define LEFT_PROBE_BED_POSITION 0
 #define RIGHT_PROBE_BED_POSITION 140
 #define BACK_PROBE_BED_POSITION 195
 #define FRONT_PROBE_BED_POSITION 19

(on the right I cannot move over 140 since my bed is 205 and -65 of the sensors position)

Now the new print is different: there's a problem on the upper right (back/right position) -it's too low- and in general on the left side -it's too high-.
The center is still good and the right is better.

What's the correlation between the PROBE_BED_POSITION constants and the bed height matrix?
Is there a good rule to follow instead of try and error?
I mean.. it's better to have a square instead of a rectangle? a centered one or a widest one?

Two questions:

1.- Do you have EEPROM enabled?
2.- Do you have LCD or print with computer?

Once the probe register the points the matrix is generated be drawing lines between those points in the Z plane, ABL do not compensate if your plate is warped, only if it is tilted / angled on the corners. Using 3 points instead of 2 helps a little more because the added line segments.
The main problem you have is that you are missing still 60mm (30% of the plate) on the side and if the angle gets higher on that spot the printer can't do nothing about it.

Your best bet is to redesign your probe holder to either reduce the offset with your nozzle or to put it on the front aligned with your nozzle, or level first your plate manually and then use ABL if your plate is warped you need to fix that first.

Edited 1 time(s). Last edit at 07/02/2016 05:23AM by ggherbaz.

1) yes, it is.
2) I print with computer.

Understood, it's a normal linear interpolation, so with only 2 points ABL can compensate tilt, not warp..
I'll print a different sensors holder and come back.

Thz!!

Back again.

I printed a new holder, closer to the noozle.
Offsets are now:

#define X_PROBE_OFFSET_FROM_EXTRUDER -35
#define Y_PROBE_OFFSET_FROM_EXTRUDER -44

I'm now able to have a 3 points grid.

and new coords:

#define LEFT_PROBE_BED_POSITION 5
#define RIGHT_PROBE_BED_POSITION 170
#define BACK_PROBE_BED_POSITION 186
#define FRONT_PROBE_BED_POSITION 26

I also set a Z offset in slic3r (G92 Z1.0, tried also with 0.7/0.8).

But still issues persist... the rightmost part and bottom center (mid-front) are still too thin.

I noticed that marlin prints the ABL results, here are mine:

21:59:06.952 : Bed x: 5.00 y: 26.00 z: 0.68
21:59:15.889 : Bed x: 87.00 y: 26.00 z: 0.48
21:59:24.921 : Bed x: 169.00 y: 26.00 z: 0.05
21:59:33.906 : Bed x: 170.00 y: 106.00 z: -0.25
21:59:42.474 : Bed x: 88.00 y: 106.00 z: 0.29
21:59:51.267 : Bed x: 6.00 y: 106.00 z: 0.35
22:00:00.091 : Bed x: 5.00 y: 186.00 z: 0.31
22:00:08.996 : Bed x: 87.00 y: 186.00 z: 0.16
22:00:18.028 : Bed x: 169.00 y: 186.00 z: -0.27
22:00:18.067 : Eqn coefficients: a: -0.00 b: -0.00 d: 0.74
22:00:18.067 : planeNormal x: 0.00 y: 0.00 z: 1.00
22:00:18.067 : echo:endstops hit:  Z:-0.27

Is there something I can do?

Edited 1 time(s). Last edit at 07/03/2016 04:31PM by Ferberto.

Are you sure the bed is flat? Marlin's ABL assumes that it is. Use a metal straight edge and check.

Thz stephenrc!
This evening I'll check..

I'm reading about the mesh leveling: does it help in case of warped bed?
Is there a way to have a non linear compensation?

Thz

My assumption it's that you have a bent in the platform, no all ABL settings are linear and doesn't compensate for curvature.

The best option you have is to add a glass sheet on top of your aluminum bed or try to flatten the current one.

21:59:33.906 : Bed x: 170.00 y: 106.00 z: -0.25
22:00:18.028 : Bed x: 169.00 y: 186.00 z: -0.27
21:59:24.921 : Bed x: 169.00 y: 26.00 z: 0.05

These are the readings you need to pay attention to, specifically the ones with the minuses.

Use a level and try to center the bubble in all directions and in a X, then try again if problem persists then use a glass on top and see if helps.

Just checked, actually the bed seems quite linear, but it's difficult to have a clean measure since the differences should be bunch of microns...
I used a straight rod and a level.

I'll try to put a glass on it, in the meanwhile I'm going to print on half bed

Thz all for ur support!

Just a last question.. for ABL is it better to have the widest area for probing (regardless of the form -square, rectangle-)?
Or is it better a good square even if it's smaller?

Well it all depends on what you are printing, for small pieces you can use the center of the bed where the heat will be most even, but for large size prints you need to have the whole surface mapped.