thanks Sineos, i have try all what you have write.
I know that a lot of things probably got lost between discord chat, and many issues closed on github, but I tried to verify everything you wrote, and I’m genuinely tired of rewriting everything for the umpteenth time
Kevin O’Connor has closed my umpteenth issue, and I am tired after a month of seeing closed posts and disinterest (if not some response from @Arksine and @DaStivi that I thank).
About you point (3.). In marlin the mesh looking good, you can see the screenshots on my previous posts.
Klipper must certainly grow, a great project, a great firmware, but there is also a pinch of arrogance, and saving from making further polemic.
If there will be any developments on @Arksine theory, I will certainly be here to try if it is useful for testing, I am fully available
Thanks again to all those who wanted to try to give me a hand. Thank you.
What does a opening a GH issue and starting a conversation about the same issue in a third location accomplish? Nobody currently knows why it works for many people and not for you, and unfortunately “it doesn’t work for me” is not an actionable bug report. The developer of the mesh module has responded here and on Discord, both of which are intended to be used for support. Hopefully a bug is found and it can be fixed. Until then it’s still just troubleshooting an unknown issue.
FYI, I am interested in seeing the results of the investigation. Arksine wrote the bed_mesh module and is the expert.
I was aware of both the conversation on Discord and here.
The unfortunate reality is that no one knows why the issue is occurring. It does not appear to be widespread. I suspect it will take more work from those experiencing the problem to track down the source of the issue.
I managed to get some TMC2225’s running the Z-Axis on my MK3. My initial thought is that I don’t like them, even at 200ma they resonate like crazy. That said, they seem to work fine on my printer. They don’t have any problem compensating across the entire mesh area. With and without stealthchop enabled.
I’m not sure that this entirely rules out the TMC2208s, DaStivi has a 2209 he plans to test on his Z-Axis. If it isn’t the drivers, I’m at a loss as to what could be the root cause of this issue.
Thanks for your tests Arksine. I also put a 5mm tempered glass plate, the problem persists. The mesh on fluidd always seems to be warped to almost the same odd shape as the previous aluminum plate.
I will continue to investigate.
On the two ender 3 v2 that arrived 2 days ago, everything seems to work perfectly.
Thanks for now
Hello,
I think I have the same issue. The first layer isn’t consistent. In two corners the first layer is okay, but in the other two it is way to far away.
Setup:
Ender 3
SKR Mini E3 (TMC2209)
BLTouch clone
Firmware version: v0.9.1-468
Mesh:
Sample count: 5
probe count: 9*9
algorithm: bicubic
What I have tried:
Repeatability check (range 0,01)
Location bias check → no problems
If I touch the Z spindle during the first layer I can feel that it’s moving
@Arksine I build ah voron 2.4… And one thing I constantly read there (in discord) is that many have first layer issues… Mostly on the larger bed sizes (350x350)
They dont use bltouch there… And the bed should be machined… So deviation is there really low…
I have a Voron 2.4 350. The primary issue is heat. The larger the printer the more heat changes its geometry. A Voron 2.4 350 needs at least 1 hour of preheat time.
A twisted X axis is also a common problem on the larger builds.
I think that there is something particular about the CR10S Pro setup that is preventing Klipper from applying the adjustment to Z. Since Marlin works on these printers it suggests that the firmware is doing something differently.
I have another idea that I want to try on my MK3 when I get some time. When I first tested the TMC2225s each Z stepper had a dedicated driver. It looks to me like you the CR10s Pro is running two motors in parallel on a single driver, so I want to give that a test. Then I should probably test standalone mode.
@DaStivi, did you ever try a TMC2209 on your printer?
FWIW, I don’t believe that @Felix is experiencing the same issue. It sounds like two of your corners are outside of the mesh and there is a continued drop in bed there.
I believe the issue has something to do with larger beds. I’m having the same issue where the mesh matrix matches probed results and isn’t applying the correction. The printer config below was applying bed level corrections perfectly on Marlin before I moved to klipper. I’ve tested 5x5 10x10 and 20x20 mesh grids. I’ve also verified the bed mesh leveling is being saved in the printer.cfg file.
I will upload my printer.cfg file and klippy.log file later today when I get off of work.
Printer info:
Bastardized Ender 3 V2
SKR 2 Version B
Micro Swiss hotend
Enderextender XY upgrade (400 x 400)
Enderextender 600W bed heater (400 x 400)
5mm glass bed
DIRECT drive
EZABL Pro
Hero 5 dual part fan cooler
Linear rails
Hi, just to add that I’m experiencing the same issue of successful and accurate probing and mesh generation but it seemingly not being applied during printing. HOWEVER I have a small bed (220x220).
Prusa mk3s DIY
SKR 1.4 Turbo
TMC 2209 dual Z
Genuine BLTouch / 8mm inductive probe /4mm inductive probe all tested with same outcome
I feel it would be useful for as many of experiencing the same issue to make themselves known and add their hardware setup and print.cfg klippy log etc so that we can start looking for any common patterns between them as it seems to be effecting an in increasingly larger group of users but still a minority.
Will add mine later also when back.
Cheers
J
I’ve printed a first layer calibration stl and from the pictures it appears the X coordinates are flipped based on the physical results. I do believe z corrections are being applied because I can see the Z values changing during the first layer print.
I’ve attached pictures of first layer print, the origin is in the bottom left corner of the main picture. The dots on the left are squished to much, the middle dots are spot on, and the dots on the right are all to high.
Does anyone know how to easily reverse X axis mesh settings to test this theory?
I just joined here and I thought this might be helpful to those experiencing issues with CR-10S Pro (v1/v2). I have an original CR-10S Pro v1 that I heavily modified, including installation of original BL Touch using Creality OEM aluminum bracket. I am using the Duet3D Mini 5+ to control the printer instead of the OEM board.
My experience has been opposite to what is being reported here. I’ve had lots of bed mesh issues with Marlin and all these issues disappeared when I switched to Klipper (about a year ago). The referenced configuration file has the correct parameters to create a perfectly centered, symmetrical mesh. I currently use a 10x10 mesh and I get pretty much perfect first layers irrespective of where I print on the bed.
Mind you, I am currently using the Creality glass bed that is reasonably flat, but I will be switching to Energetic PEI soon. One thing worth noting is that on the CR-10S Pro the bed is quite heavy and some of the measured mesh errors are actually created by the bed trolley rollers and the Y-extrusions.
I may have found a solution to my bed level issue. From my prior post on the 27th, one thing you will notice from the main picture is the location of the bed screws and other thing is that I’m using a magnetic base as well. One test I ran last night was to increase my mesh points to 80 x 80, essentially giving me around 6400 points on my mesh. While this did crash my MCU every time I tried to home an axis, I was able to visualize the mesh in the tune window of Fluidd. What I was able to see was the impact the bed screws were having on my bed mesh which was generating false dips in the bed.
The first step I took was to identify the screw regions as faulty regions and input them into my printer.cfg file.
Next, as part of the 80 x 80 mesh, I noticed ripples in the mesh which is noted as a possible issue when using a bicubic algorithm. I was using the default bicubic_tension =0.2. I never increased the value. Rather I continued to decrease the matrix mesh, eventually getting to 6x6 but never entirely removing the ripples. Once I reached 6x6 I switched from bicubic to Lagrange and the ripples disappeared.
Finally, I realized the left to right issue on the bed was due to the x-axis not moving symmetrically on the left and right. This is most likey due to the large x-axis of 400. I’ve tried tightening the v roller but there is still too much play on the right side of the gantry. I’ve ordered a couple of T8 Anti Backlash Spring Loaded Nuts. I already had a dual z setup but when I moved to the enderextender XY upgrade I didn’t have a long enough belt to reach the right side of the gantry. I thought I could get by without it but now I see I do need to add the dual z back into the system. I’ve ordered a longer belt as well and will report back if this resolved the issue shown in the images from left to right.
If you are curious what flat spotted X-gantry wheels result in, have a look at the following two images. The first one shows my bed mesh after I had a somewhat controlled nozzle crash into the bed while I was experiencing issues with BL Touch installation. This occurred while I was also experimenting with inexpensive but allegedly “good quality” polycarbonate wheels from AliExpress. The second one shows the exact same installation just after I replaced the awful polycarbonate wheels with the reasonably high quality Openbuilds wheels. Note that there is still a wave in the Y axis that I believe is caused by the design of the Y rails on the CR-10S Pro.
When you install the anti-backlash nuts make sure that you install them the correct way or they could make things worse. The springs have to be ABOVE the gantry, pushing the gantry down with gravity assist. Unfortunately many guides, including some created by allegedly reputable places show the anti-backlash nuts installed INCORRECTLY with the springs below the gantry, resulting in the springs trying to lift the gantry against gravity. Also get quality anti-backlash nuts. The first ones I got off Amazon were simply awful, they looked like something I would file in my basement with a hand file after a New Years’ Eve party. Ones I currently use came from ‘REPRAP Official Store’ on AliExpress and they are of very high quality.
