I have been trying to set up mesh bed leveling on my Voron 0.1 with a Klicky Probe, I have been having an issue where the first layer is good on the front of the bed, but much too high on the rear. This is consistent and I can even skew the bed and it will have the same first layer pattern.
This is what it looks like, I can get more of the build plate at a good height, but then the front will become too low.
Hi! I have similar issue on a Voron 2.4 R2/350mm. Maybe we can fix it together.
My list of possible reasons is:
Wrong measurement: Does your results change if you change to Quad Bed Leveling? Should be a CoreXY very similar to my 2.4?. Place the probe points 10-20mm inside the table edges (let’s say: close to the border of the bed/table (sorry: I use “table”). If you have no change, go further in in 10-20mm steps until you reached 1/3 of your tables side length. The approximated flat will change, if your table is distorted and “by fortune” you could get a better solution for your table. Maybe you can tilt the Mesh Bed-Flat “by hand” somehow but I would not recommend this.
Table flatness: Id your table really flat? If the table is distorted too much, nothing can help you (this is my problem, I believe with this huge bed of 350mm). Can you post a mesh bed result picture or the values itself?
Wrong 1st Layer: On you “good” side: how thick is the layer and how thick it should be? Attention: Measuring a layer thickness is very difficult! If surface is not 100% flat (slight underflow helps), it will drive you to distraction!
Klicky Probe Problem: The hardware of the klicky probe is sub standard for measuring (end switch, magnetic connection, design of the probe body), so mine klicky probe shows a strange behaviour after some months of work (rising distance value), and some of my prints went wrong while they are great at the second shot (changing nothing).
So far no systematic error in the compensation logic has been identified
There are user reports who have successfully solved their issue. In all cases it has been mechanical issues (twisted / bent extrusion profiles) that lead to a non linear deviation between probed point and nozzle coordinates
based on the thread linked, I should probably up the bicubic tension to 0.5, but my understanding is that would give me better interpolation, and the probed points would still line up, which they do not.
3: measuring the first layer thickness would be especially tricky since I have a textured bed, but If other things don’t work I can give it a go.
4: Yes, the klicky is not ideal and it definitely has consistency issues, but this issue is consistent so I would be hesitant to blame it on the probe since I get the same results every time. I also tested it with the klicky probe from my friend’s working printer and I had the same results after recalibrating the z offset.
I like the idea, but my main reason to get bed mesh working is to not have to worry if everything is perfect and just let software account for variations. I am impressed that that method works as well as it does.
As a general update, I have been doing a lot of testing, and I believe that I have some sort of mechanical issue with my gantry system. If I home the bed at the back of the bed, and adjust my z offset, I get a z offset that is over 0.2 lower than if I home at the front of the bed and and adjust the z offset there.
This implies to me that either:
The distance from the nozzle to the probe activation point changes based on where the probe/nozzle is in on the xy plane.
or
The front and back of the bed raise different amounts going from the probe activation point to the nozzle.
Either way in this case, probing closer to the nozzle would help, so I am trying to make a version of the klicky probe with less z offset.
Thanks for sharing this information.
Such mechanical issues are exactly the source of such bed mesh issues:
Probing the bed:
You record the mesh with your probe.
During this recording your gantry is at position X1 and Y1 for each point of the mesh
Printing with mesh compensation
Due to the probe offsets, your gantry is now at the position (X1 +/- Probe X-Offset) and (Y1 +/- Probe Y-Offset) as you now need the nozzle and not the probe
Any mechanical deviation between the position X1 and the position (X1 +/- Probe X-Offset) will spoil the compensation effort (same for Y)
I have this same issue with my v0 with kirigami and I think, I have found the issue (but not solved it).
my mesh is nearly straight (probably won the bed lottery), but still, the mesh tilts upwads towards the back. I think, the cantilever bed is to blame. If I take my probe and push it on the bed, the bed tilts downwards just a tad. Currently, I am searching for either a probe that doesn’t require as much force, or a way to get the bed more rigid.
I would be happy if you replied if your bed tilts too?
Colin
this matches almost perfectly with my issue. Leveling the bed with the nozzle means that I have a bed mesh that tilts up in the back. It also matches with a few other things I have done:
1: Trying probes with different x,y,z offsets didn’t make a significant difference
2: Switching the microswitch on the probe from a generic switch to an Omron switch seemed to help, but didn’t remove the issue (less force on the Omron)
Interesting that you still have this issue with the Kirigami, I would have thought that it would have increased rigidity and solved this kind of issue, how much of a tilt do you see? I have been seeing anywhere from 0.17mm to 0.22mm depending on the microswitch
As I leveled my bed using SCREWS_TILT_CALCULATE, I don’t see any software-tilt. If I level it using the paper method, there is a tilt of around 0.4mm. You can try it by leveling the bed by hand and then pressing a probe on the bed while printing. If the layer thickness changes when the switch triggers, there is your problem. Switching to unclicky didn’t fix the problem as well. Probing the bed with 1mm/s speed helped a lot, but didn’t solve the problem completely. I have ordered the “Fysetc full metal frame” from china yesterday. Let’s see, if there is more rigidity in this bed. Additionally, I’ll try a levered button, which feels a lot easier to press.
//edit: the levered (noname) switch does the trick. The D2F-F I used before seems to have too much force (eventhough it is rated at only 0.74N). I’ve ordered some D2F-FL (0.25N) and D2F-FL3 (0.39N).
That’s a lot of tilt, I have been trying a decent amount to fix this, but my testing has solved nothing. I did try using a switch with the lever on it to reduce actuation force, but it didn’t seem to make any difference. I also printed a modified version of the parts that attach the linear rails to the extrusions for the bed, and my bed mounting is now very rigid, and I have seen no improvements, I am not sure if the actuation force could provide this much flexing.
you could try leveling the bed manually and then press the nozzle on in while it prints a first layer. If it doesn’t change the squish, you know at least where not to seek
//edit: For anyone finding this post in the future: My fysetc full metal bed frame is WAY more rigid than the formbot Kirigami bed. There is no measurable tilt anymore. I am happy so far.
Hi, I have a brand new Ender 3 S1 Pro. I installed klipper on it and I am having bed leveling issues with it too. Similar issues to what @Sam_J reported. While I was checking all configurations based in Klipper documentation I notice that when I am running BED_MESH_CALIBRATE in the console the points that the probe is reporting does not match the actual position of the probe, for instance I have this in the last points of the calibration.
type or 11:41 AM
Mesh Bed Leveling Complete
11:41 AM
probe at 251.000,230.000 is z=0.717500
11:41 AM
probe at 198.500,230.000 is z=0.670000
11:41 AM
probe at 146.000,230.000 is z=0.637500
11:41 AM
probe at 93.500,230.000 is z=0.607500
11:41 AM
probe at 41.000,230.000 is z=0.485000
11:41 AM
probe at 41.000,185.000 is z=0.477500
The bed size of the printer is 240X240.
The probe is reporting that it is located at 251.000,230.000 in the last calibration when in fact it is at:
For X - (251 - x_offset) which is actually 251 - 31 = 220
For Y - (230 - y_offset) which is actually 230 - 40 = 190
I am not sure if this is just an issue with how klipper present the output of the BED_MESH_CALIBRATE in the console or actually it is recording those probes exactly in the same way it is presenting them. Is there any way to get the output of the mesh along with the (x,y) coordinates just to bed sure that the bed mesh is being stored correctly?
