Limitations of bed mesh correction.
This is a continuation of a problem that many experienced on the MK3 that apparently is persistent with the MK4. The problem or behavior is about how the bed mesh algorithms adjusts for irregulates around the edges of the bed. I started looking at this on my MK4 when I notice the nozzle deflecting the build plate downwards when printing on right side of the bed. I was able to reliably replicate this behavior and printed out some full size sheets to scope my problem. See the pic
For me the problem was so bad on the MK3 that I went full tilt with the nylock mod with great success. I will admit the MK4 does a fairly decent job, but the right side of my bed is unusable. Before I go making mechanical changes I would like know why the bed mesh does not correct this problem. I tried to dig into the firmware, but couldn't locate the bed mesh or mesh application routines. So my question to they that know are:
1. Are the bed mesh adjustments clipped in some way to limit the maximum correction along the bed edge, or is the algorithm designed to smooth the bed towards the center, expecting most print surfaces to be closer to the center, at the expense of the edge z correction.
I need to know why the mesh can't correct this before I can move towards and effective mitigation. I know the issue is not unique to my printer, because I've seen pics on other posts that are very similar to my own. It's possible that I have some correctable mechanical issue, but would like to consider all the possible limitations before proceeding.
RE: Limitations of bed mesh correction.
As far as I know there is a maximum correction.
I would first determine whether the bed is really that uneven or whether the Z axis alignment is not correct.
To do this, I created a 5X5 pattern with circles with a diameter of 8mm in the Prusa slicer and sliced them with .2mm.
Then I showed the Z height on the display and noted the values at the respective positions when printing. This creates an exact image of the surface.
If e.g. For example, if all points from left to right become higher, the X axis must be adjusted.
RE: Limitations of bed mesh correction.
I'll give that a shot tonight. At least this method should give me the actual z-heights across the bed. That is if the reported z-values have sufficient resolution, and are real values, not bed mesh adjusted values, I don't know why that would be true, but I've never had anyone tell either way. I seem to remember the reported z values on the display header being rounded to 10ths of a mm, if that is true then it might not have enough resolution, but we'll see tonight.
It would be a lot easier if I had Octoprint connected so I could use the console. But I thought I would give Prusalink/connect a try. My other printers use kKipper and your can see the bed mesh adjustments being applied during the print, with values reported down to 0.001 mm. None of that information is being reported to the user in the Prusa world.
Speaking of the lack of a console, do you know of a console application that works with the new buddy board?
RE: Limitations of bed mesh correction.
The values given are the actual values to 2 decimal places. The number of digits displayed has nothing to do with actual accuracy. Therefore, in my opinion, there is currently nothing on private printers that comes close to the accuracy of a load cell.
RE: Limitations of bed mesh correction.
great info. I assume you mean that the indicated z position (to 2 digits) is the actual dynamic distance above the 0 reference Z height. Which I assume is calculated from the center of the bed on the these printers.
I'll post for reference what I get by using your method in a couple hours.
RE: Limitations of bed mesh correction.
Well the method worked as advertised, but I had to slow down the printer to I record the values.
After several iterations, it turns out I'm getting some excessive z-axis drift, drag, or sag. haven't figured out which side yet, but I was able to run the z-axis to the stops and get a pretty good first layer on one of the attempts.
So now I have something to go after and until I get it corrected, I'll have to run the carriage against the upper stops before each print and bed mesh.
thanks for your recommendation.