G29 J doubling my mesh values with every run

Hey gang, I'm running Marlin 1.8.3 (and bugfix) on a RAMPS 1.4 board and have run into what I hope is just a mistake on my part.

So let's say I do the usual UBL setup on a new machine
M502
M500
M501
G28
G29 P1

After the probe finishes, I look at my mesh with
G29 T

and see sane values. UBL works great, I'm a happy dude. Now lets say I run

G29 J

to compensate for a tilted bed when I moved the printer or something.
The values in my mesh have now all been doubled and my printer tries to live its dreamlife of being an auger and drills the hotend into the bed.

So I run G29 J again, again the values double. What was originally a 2 lets say became a 4 and now is an 8 .

Is this expected behavior? Am I misunderstanding the purpose of G29 J?

Edited 1 time(s). Last edit at 01/27/2018 07:55PM by scratch.

I'll have to look into that... It is my belief that multiple G29 J's should converge on the same values. I know that does not happen currently. It is on my To Do list.

The G29 J stuff has been changed and is different in the bugfix-v2.0.0 branch. It might be worth loading that firmware on your board and see if the G29 J behaves differenty.

One last question: Were you doing a 3 point or a grid based tilt?

Edited 2 time(s). Last edit at 01/28/2018 09:57AM by Roxy.

just 3 point, no options. I'll give the 2.0 branch a shot and report back.

OK! I'll try to help as much as I can, but I'm out of town on a ski trip. One work around would be to always do a G29 L1 prior to doing a G29 J so the J's are not cumulative.

It is my belief that should not be necessary. I want multiple J's to cause a convergence of values. But that is not happening right now.

PS. Incidentally, the G29 J (for 3-Point leveling) was changed in the bugfix-2.0.0 branch to use the same algorithm that the grid based mesh tilting was using. The reason was two fold. First, it reduced code size and complexity. But the big reason to do that was the 3-Point provided a test case with an exact solution for the grid based. 3 points results can be tested exactly for correctness and helps prove the grid base solution is doing the right thing. (And there is debug code that can be turned on to confirm that if you are interested.)

Edited 2 time(s). Last edit at 01/28/2018 09:02PM by Roxy.

sorry for the delay, got sick., it looks like even on 2.0 bugfix G29 J is doubling all my values.

G29 T, prior to J

READ: . . . . . . . . . .
READ:
READ: . -0.934 -0.778 -0.748 -0.480 -0.505 -0.181 -0.178 -0.198 -0.173
READ:
READ: . -0.828 -0.648 -0.660 -0.372 -0.522 -0.273 -0.566 -0.381 -0.378
READ:
READ: . -0.513 -0.414 -0.119 0.044 -0.233 -0.125 -0.366 -0.286 -0.327
READ:
READ: . -0.413 -0.112 0.284 0.636 0.053 -0.069 -0.282 -0.250 -0.396
READ:
READ: . -0.063 0.034 0.350 [ 0.455] 0.166 0.132 -0.064 -0.009 -0.158
READ:
READ: . 0.148 0.210 0.246 0.313 0.305 0.446 0.054 0.057 -0.157
READ:
READ: . 0.294 0.335 0.415 0.491 0.440 0.459 0.182 0.112 0.020
READ:
READ: . 0.490 0.598 0.527 0.667 0.453 0.633 0.336 0.173 -0.015
READ:
READ: . 0.223 0.210 0.280 0.376 0.279 0.384 0.139 0.176 -0.041
READ: (1,1) (273,1)
READ: (0,0) (9,0)

After G29 J

busy: processing
READ: ok
SENT: M105
READ: ok T:22.47 /0.00 B:23.53 /0.00 @:0 B@:0
SENT: M105
READ: ok T:22.47 /0.00 B:23.53 /0.00 @:0 B@:0
SENT: G29 T
READ:
READ: Bed Topography Report:
READ:
READ: (0,9) (9,9)
READ: (1,299) (273,299)
READ: . . . . . . . . . .
READ:
READ: . -2.126 -1.794 -1.717 -1.161 -1.191 -0.525 -0.500 -0.520 -0.450
READ:
READ: . -1.903 -1.525 -1.529 -0.933 -1.214 -0.698 -1.264 -0.874 -0.849
READ:
READ: . -1.261 -1.046 -0.436 -0.090 -0.624 -0.390 -0.853 -0.673 -0.737
READ:
READ: . -1.050 -0.430 0.382 1.105 -0.041 -0.267 -0.673 -0.590 -0.864
READ:
READ: . -0.338 -0.127 0.525 [ 0.755] 0.197 0.146 -0.226 -0.096 -0.376
READ:
READ: . 0.095 0.237 0.329 0.482 0.484 0.786 0.022 0.047 -0.363
READ:
READ: . 0.399 0.498 0.678 0.850 0.766 0.823 0.289 0.169 0.003
READ:
READ: . 0.802 1.036 0.913 1.213 0.803 1.183 0.609 0.302 -0.056
READ:
READ: . 0.277 0.271 0.431 0.643 0.466 0.696 0.226 0.320 -0.097
READ: (1,1) (273,1)