Two questions regarding expectations for G29

Hello Rappers,

When I repeatedly perform a G29 auto-level (after G28 home), the printer returns inconsistent matrices. So I want to see if I correctly understand what the results should be if everything is operating normally.

Question 1: Should repeated applications of G29 (auto-level) result in the same matrix as long as nothing is loose or has otherwise mechanically changed?

Question 2: Should the center of the G29 matrix normally be very close to zero since the center of the auto-level probing sequence usually occurs in the same X, Y location as the location where the "home" Z reference height was measured?

Thanks very much.

Answer 1: Yes. Assuming nothing has changed mechanically the matrix should be identical within the sensor error.

Answer 2: Yes, assuming your Z probe offset is correct.

The most common cause for meshes changing between runs is the X axis becoming misaligned. When there is no current on the Z motors it's very easy for this to happen. Using dual Z drivers allows you to use G34 to re-level the X axis with respect to the bed, or you can use a single Z motor with a closed loop belt to ensure the two Z lead screws move evenly.

Thank you MMcLure. I do have two steppers for the z lead screws but they are locked together with a closed belt. What I’m seeing is that, with repeated applications of a G28,G29 sequence, the center value of the mesh alternates between 0.00mm and 0.21mm. There are, of course tolerance variations of about +/-0.01mm, but the center value of the mesh will never fail to alternate between those two basic values with each successive application of a G28,G29 sequence. I’m not sure which is more confusing: the fact that it varies at all, or the fact that it alternates between the two values.
Also, the behavior seems unaffected by any z-offset value.

What mode of G29 are you using? I have heard reports of issues with Bilinear leveling - I personally use UBL so I haven't seen any issues similar to yours.
One thing to ensure is that the Z offset is correct before you generate the mesh - you should be able to print something small in the center of the bed without bed leveling enabled so the leveling is only adjusting for variations in the bed surface. For the same reason bed corners should be trammed before generating a mesh.

Thanks. I’m using bilinear leveling. I’ve heard of ubl but have never used it, so I don’t know if it would address this. Issue.

I’ll definitely set the Z offset before I print, but i first need to figure out how to get auto leveling matrices that don’t alternate in height by 0.21mm.

To anyone who uses bilinear aotoleveling: if you perform a G28, G29 sequence, is the center point of the matrix zero (or very close)?
Then, if you repeat the G28, G29 sequence, is the matrix essentially the same?

I wasn’t able to find or fix this problem, but I was able to work around it.
I was able to prove that the matrix where the center point was zero was, in fact, “correct” and the offset one was “incorrect”. So as long as I save (M500) the correct and then recall (M501) and activate (M420 S1) it before a print, everything works ok.
I would rather do an autolevel before each print like most people do, but I’d have a 50% chance of getting the “incorrect” one and either losing the print or gouging the bed.