Need Bed Leveling Suggetions


I recently assembled the kit from Marty. This kit includes the rack and pinion for bed leveling.

I have adjusted the printer (including z offset) so that when the hot end is in the front left corner (X20 Y16 Z0) I can just barely fit a sheet of paper between the nozzle and the bed. I get a really nice first layer when printing a 20 x 20 calibration cube (I configured the slicer to only print the first layer). As I move the hot end over to the front right corner I can see the gap between the nozzle and bed increase. When the hot end is in the front right corner (X180 Y16 Z0) the gap is so big I can fit 4 sheets of paper between the nozzle and bed. Printing that same first layer of the calibration cube in this location does not turn out well at all.

I ran G28 and G29 before moving the hot end around and measuring the gaps. I’ve done this with the bed code and heated to 60C. I run G28 and G29 within the start script of each print. The results are the same - the gap is greater on the right side of the bed.

I verified the left and right ends of the X carriage are even and that the gap between the MDF board and the bottom of the bed are even, both with a micrometer.

I would like to print some large objects that take up nearly the full width of the bed, so I need to get this leveling issue resolved.

Any ideas on what I’m missing here?



That’s not “level”. You kinda have to change the gap so the hotend is at the same distance from the print bed. The MDF may or may not end up being further from the print bed.

One thing for sure, auto bed leveling isn’t perfect, it work well if your bed is skewed a bit on one side, but it doesn’t do well with “curved” (convex or concave middle) print bed. Knowning this, you still need to level the bed by hand at least once to get it close to help the auto bed leveling.

When printing moving the hotend, you should’nt see more than a quater turn rotation of your Z motor. If they have to compensate more than that, you should definitly level it by hand.

You might want to look into printing one of those :
This one fit 8mm rods, but it shouldn’t be too hard to remix the openscad file for 10mm. It helped me a lot to level my bed even tho I use auto bed leveling.


What can you attribute the extra gap on the right side to (if anything)? Does moving the extruder/xcarriage to the right cause the z-axis coupler on the right side to “spring” up causing a lift on that side? I have seen that before, if there is a little binding happening somewhere with the belt assembly or the threaded rod not being quite centered. Look closely at where the belt goes around the tensioner pulley on the right side as you move that way and see if something is happening. When something like that does happen, often it is not corrected for by the bed leveling algorithm, because it doesn’t go as far to the right, and so it does not “see” the deflection, so it is not reflected in the bed level coefficients. BTW, what are your bed coefficients (they are printed to the screen at the end of G29)?


You’re absolutely right. I’m not sure what I was thinking last night. I’ll print a mounting bracket for my depth gauge and take some measurements.

Thank you.


I did not observe any sudden movements with the z-axis coupler on the right when moving the x carriage to the right. I don’t see anything unusual going on with the belt or tensioner pulley on the right either.

I moved the x carriage back and forth from X20 Y16 to X180 Y16 several times. I see that both z-axis couplers turn ever so slightly. As the x carriage moves from right to left the z couplers turn to right. As the x carriage moves from left to right, the z couplers turn to the left. Again the couplers turn so little during this movement it’s almost imperceptible. I suck the hex wrench in the coupler just to make easier to catch the movement on video. Could this movement in the z couplers be the firmware compensating for the topology of the bed as measured by the G29 command?

I switched from a 2 dimension grid to a 3 dimension grid for bed leveling (G29 now measures at 9 points instead of 4). Afterwards, the difference between the nozzle and bed on the left and right sides is much smaller, but still big enough to make a difference with the first layer.

With a z offset of 11.14 the bed coefficients are as follows:

Send: G29
Recv: Bed x: 15.00 y: 50.00 z: 10.09
Recv: Bed x: 100.00 y: 50.00 z: 10.20
Recv: Bed x: 185.00 y: 50.00 z: 10.14
Recv: Bed x: 185.00 y: 150.00 z: 10.05
Recv: Bed x: 100.00 y: 150.00 z: 10.12
Recv: Bed x: 15.00 y: 150.00 z: 9.94
Recv: Bed x: 15.00 y: 250.00 z: 9.83
Recv: Bed x: 100.00 y: 250.00 z: 10.03
Recv: Bed x: 185.00 y: 250.00 z: 10.00
Recv: Eqn coefficients: a: 0.00 b: -0.00 d: 10.12
Recv: planeNormal x: -0.00 y: 0.00 z: 1.00
Recv: ok
Recv: echo:endstops hit: Z:10.00

Thank you.


This might be a stretch. When I setup the rack and pinion probe I noticed there was a slight y axis angle in the probe, probably caused by first layer squishing. With the metal arm of the probe pointed to the front of the machine the offset angle of the probe decreased the angle of the metal arm and required more force to create a hit. When I switched the probe around so the metal arm pointed to the back, it’s angle was increased, it took less pressure to hit and I got better auto bed level results.