With auto level probe, prints are skewed?

Laser8302

I put in a 3d touch probe specifically for auto bed leveling. After leveling the range is something like -0.05 to +0.148, about one layer height. I've set my z offset correctly.

Here's my problem, I have performed 3-4 mesh validations with a 0.2mm layer height, on the left side of my bed, it seems to grind the nozzle. On the right, it's too far away and the filament doesn't stick to the bed.

On the bed map, the left side is a bit higher than the right, but shouldn't the mesh take care of that?

I will note that I think my z couplers are stretched so I ordered a new set.

So any clues into why it would print like this are welcome.

Der_Muck

Without your start skript nobody knows how your system works.

Laser8302

Recv: Bed Topography Report:Recv: Recv: (0,5)                                   (5,5)Recv: (40,264)                                (264,264)Recv:   0.038    0.049    0.104    0.156    0.139    0.100Recv: Recv:   0.066    0.101    0.178    0.213    0.213    0.191Recv: Recv:   0.078    0.140    0.243    0.247    0.274    0.243Recv: Recv:   0.076    0.149    0.254    0.289    0.290    0.273Recv: Recv:   0.048    0.141    0.240    0.269    0.266    0.261Recv: Recv: [ 0.005]   0.094    0.158    0.199    0.227    0.201Recv: (40,41)                                    (264,41)Recv: (0,0)                                     (5,0)Recv: ubl.mesh_is_valid = 1Recv: ubl.storage_slot = 1Recv: echo:Bed Leveling OnRecv: echo:Fade Height Off

Here's what happens with a G26 (mesh validation)
As you can see on the left, there's some nozzle grinding and on the right nothing sticks. In the center it's ok. 

Also in my start script I made sure the last command is M420 S1 to make sure the mesh is on.

Also here is my bed level graph:

Der_Muck

Hmm I dont know what G26 is, just try to use UBL, that will work.
What do you use, BLtouch or IR?

Laser8302

I've got a bl touch (3d touch, from china... )
I think the old z endstop is blocking one side of the gantry from going down and causing the uneven printing. The side that is higher/not grinding the nozzle is where the z endstop block is.
I also have very stretched couplers so I'm going to replace those.

cs2000

Will almost certainly be that old Z endstop mate. Standard practice is to remove it totally, or move it down on the adjustment screws so its not making contact any longer.

Laser8302

OK so another item I found out.
The mesh insert value. If you don't have it set to a value LARGER than your largest probe offset, the mesh won't be valid after the first G29 P1 command.

Two solutions:
After G29 P1 finishes and you do G29 S1, you have to send G29 P3 C
and then another G29 S1. This will auto-fill the missing mesh values. If you do G29 P2, you have to use your LCD to move the print head up and down and move between points.

My solution:
Change the mesh insert value to be 2-3mm larger than my largest probe offset. IE my probe is at Y=42 X=12 so my mesh insert is 45mm. Now each G29 P1 produces a valid mesh.

michaellpz

Hello Laser8302,
Currently i hanging over the same problem. I configured the jgaurora for ir probe. All thinks are running fine. But on the right side the nozzle is lower than at the left side. If i raise the z-offset the the print at the right side sticks fine but dont stick at the left side. 
if i manuelly change the bed height all things are same. This shows me, the printer is using the bed leveling. 
Sadly i dont unterstand your solution with mesh insert value. I dont know this value. where i can change this.
It would be nice to get some help.

Best regards
Michael

Laser8302

So first, I've downloaded the latest Marlin 1.1.X bugfix. It's based on Marlin 1.1.9 so if you have that loaded you should be able to just copy your configuration.h and configuration_adv.h to the new code.

Second, I am using this as my start code:

G21 	;set units to millimetres
G90 	;set to absolute positioning
M106 S0 ;set fan speed to zero (turned off)
G28 	;home all axis
G29 L1	;Load Mesh
G29 A 	;turn on mesh bed levelling if enabled in firmware
G29 J 	;Probe 3 points on the bed
G0 Z0.2 X10 Y10 F6000 ;move to front corner above bed
G92 E0 	;Zero extruder
G1 X60 E9.0 F1000.0 ;Start prime line
G1 X100 E21.5 F1000.0 ;continue line
G92 E0 	;zero the extruded length again

The G29 L1, G29 A and the G29 J are key.

G29 L1 = Load mesh in slot 1

G29 A = activate mesh

G29 J = probe 3 points, defined in configuration.h in the UBL section. (Marlin 1.1.9 auto-defines these points.)

This seems to have fixed my mesh leveling issues!!!! I just had a problem with the UBL_Z_RAISE_WHEN_OFF_MESH. I tried commenting it out, but it ended up just raising anyway. So instead I've set it to 0 and flashed my firmware. I'm going to start a print and see how it goes.

Laser8302

OK so that didn't work as well as I expected... I forgot after flashing I need to do M502, M500 then also a M501...

So I also had some problems leveling the bed (who doesn't?)
I forgot to do M420 S0 (turn off bed leveling) when leveling the bed.
Now that I have a good (level) bed I'm going to try yet another print...
not sure why this printer has been soooo tough on me.

Laser8302

I halved the Z-axis acceleration, and changed the threshold to 0.5 from 0.3
Now my bed topology report shows a max deflection of 0.15mm and my probe st dev is 0.002mm vs before the change it was closer to 0.03mm!

Trying another print now... Fingers crossed yet again.

Laser8302

More hopes dashed. Over the long weekend I had, I was unable to repeat my success of my first few prints...
I am almost 100% certian it's due to the leveling of my X-axis. I checked with bed leveling ON with a 0.1mm feeler gauge and the nozzle is further away on the left side of the bed than the right. I could slide the gauge under the nozzle with no resistance on the left, and couldn't get it under the nozzle on the right.

So here's my plan...
Get the bed as level as possible using a bubble level in several locations
Do the same with the x-axis
home the printer and turn OFF bed leveling (M420 S0)
Move the x-axis left and right with a 0.1mm feeler gauge, set the nozzle gap at each side close to the front of the bed by adjusting the lead screws first, then the bed level.
turn on bed leveling with G29 P1, M420 S1
move the X-axis across the bed, back and forth with a 0.1mm feeler gauge and level each side using the lead screws then the bed level screws
(The M420 S1 will make the X-axis compensate for some of the unevenness of the bed. the rest will be my adjustment.)
Try another G29 P1 and check the gap.

I hope this will help. I know of at least one other person on the forum that is struggling with this same issue.

Der_Muck

You have those problems with the IR probe right?

Laser8302

3D touch (cloned BL touch)
Last night I had a discovery.
I tightened one of the screws on the probe, and suddenly the "slope" of the bed shifted from the right side to the left side... Now the slope is reversed???

I'm going to venture a guess that this touch probe is to blame at this point. Even though my M48 command shows 0.002mm std dev. I'm going to investigate a new style. Maybe a servo that moves a bar with a microswitch attached.

Laser8302

Der_Muck,
I'm using a 3d touch (cloned BL touch...) probe. I mouted it on the front of the print head so that it's X-offset is 0, Y offset is ~41mm.

Hello Laser8302,
Currently i hanging over the same problem. I configured the jgaurora for ir probe. All thinks are running fine. But on the right side the nozzle is lower than at the left side. If i raise the z-offset the the print at the right side sticks fine but dont stick at the left side. 
if i manuelly change the bed height all things are same. This shows me, the printer is using the bed leveling. 
Sadly i dont unterstand your solution with mesh insert value. I dont know this value. where i can change this.
It would be nice to get some help.
Best regards
Michael

Hi Michael, Sorry I missed your post and didn't respond.

So in the firmware where you set up your Z-probe, there's this value (search for it in the configuration.h file)

# define MESH_INSET

If you're using bilinear, you can define the bed probe size there.

Make sure you have your probe mounted firmly and level. If the wires are tight, there may be some induced change in height due to the wire tension when on one side of the bed vs the other.

Samuel Pinches

Welcome @michaellpz - hopefully Laser8302's post above is helpful.

Laser8302

So I found my main problem and a possible solution.
I never checked the X-axis vertical alignment. With a short bubble level I measured it and found the right side to have the bottom bar kicked out enough that it was throwing my probe off. The left side was properly level somehow, it's just the right hand side. Maybe the rods are bent?

The two mounts I've used so far put the probe 40mm to the front and to the back. I noticed when switching probe locations that the "tilt" switched sides of the bed. With the probe at the back, the left side of the bed was closer to the nozzle. Wit the probe at the front, it caused the right side of the bed to be reported "lower."

To fix this seems somewhat straightforward. Since my A5 is an older model, it has the metal Z-axis rod guides at the top. I loosened the screws and found a good position that levels the x-axis better. I'm going to print out a spacer in PLA or ABS to put at the top to hold it there. If it twisted the other side too, I might try a spacer near the stepper on that side to fix the twist that is introduced by my fix.

I haven't had time to fix it just yet, too much family summer fun and work going on. I'll let you know when I do. For now, I'll leave you with an illustration of what I'm talking about.

Laser8302

Last night, I printed my top spacer. It needed 4.3mm at the top right. I'll take pictures tonight and post.
The bottom left is now slanted inward, which doesn't seem to affect the probe/print as bad as the upper portion did.
UBL seems to have the bed dialed in really well now.