What leveling process are you using?
I haven’t heard much from SM. I’m guessing Covid19 is hampering their work.
Currently my best proces is:
Do an Auto level, apply an adjustment matrix, tweak using my “Check Level Hot.gcode” file.
I’m finding that the more I’m tweaking the leveling the more it heats up and consequently the more it changes. I have a definite dip in the bed on the front left.
About my “adjustment matrix”: I did an Auto level, then I did a Manual level, tweaked that to ±.02, then took the difference between that and the Auto level. Using the “M421 InJn Qz.zz” you can increment the values in the matrix. I have a file that increments all the matrix values with one pass. You can turn on Auto leveling and then run that file, it does the Auto level, then applies the adjustment matrix. That gets pretty close, then a few individual matrix point tweaks.
I suspect the best bet is to have a “cold adjustment” file and a “hot adjustment” file. Run the cold one when you first start up and then the hot one once you are warm.
This is very tedious I know, hopefully SM can learn how to improve Auto leveling and especially allow hot leveling.
EDIT: Currently the Manual leveling is broken, when I do it I have to move some of those values in the matrix to their correct location. The matrix effectively needs to be mirrored about the X=Y line. So values in (I,J) 0,0 1,1 & 2,2 are correct. 0,2 swaps with 2,0. 1,2 swaps with 2,1. And finally 0,1 swaps with 1,0.