Tried everything, can't get bed level.  

Re: Tried everything, can't get bed level.

Check this post, http://shop.prusa3d.com/forum/original-prusa-i3-mk2-f23/p-i-n-d-a-probe-misaligned-t1191.html#p8981

I think your bed is a bit twisted and needs to be corrected by aligning.

Re: Tried everything, can't get bed level.

My probe is inside all of the circles. Even though my y axis is flat (all 4 feet touching, doesn't rock when tapping of opposite corners) outside of the frame, can it not sitting flat when installed in the frame be an indicator that it's still twisted? I checked my y axis for flattness before installing it back into the frame multiple times.

Re: Tried everything, can't get bed level.

I had to shim up the corners of my machine to get the bed level. It sounds like your rods are twisted a bit and that will throw everything off. give that a look.

Re: Tried everything, can't get bed level.

Sounds very much as though the 10mm threaded rods are not fully home in the Z frame.

Undo each of the 2 nuts and make sure it is fully home; you may have either a paint overspray or Y frame width issue.

Peter

Re: Tried everything, can't get bed level.

I have to ask, especially with so many people reporting bed leveling problems...

Isn't that *exactly* what PINDA and "full mesh leveling" was supposed to solve? I am having a difficult time understanding why this can't be adjusted for with a combination of PINDA and firmware.

I have measured, calibrated, and re-tuned for the past two days with extremely expensive digital calipers, and I am STILL having more squish on the left side than the right side. The slope is very even, and I've tried to adjust and re-seat the M10 rods int the Z-frame several times now. I am 100% sure they are fully seated. The PINDA probe now hits every circle dead-center even without the new XY calibration feature, and even the calibration says "bed is perpendicular".

It is extremely frustrating to not have a clear way of addressing what strikes me as a very simple issue.

What am I, and so many others, missing? So far PINDA has done a great job at "detecting" XY offset problems, but I have not observed it doing much else.

Are there any other methods to debug this situation without just "re-checking everything"?

Re: Tried everything, can't get bed level.

Are you sure you pushed both bearings under the bed towards eachother (towards the middle?) as stated in the manual ?
If you don't it could result in the bed being a bit higher on 1 side.
Maybe something you can see with a quick check ?

Then your whole build can be perfect but bed is tilted. But you probably checked this when realigning everything i assume ?

Re: Tried everything, can't get bed level.

Hey Dimitri!

Thank you for your suggestion. Yes, I double-checked the bearings underneath the bed as well, and just earlier I loosened and re-tightened all of the brackets and ends stops for the Z-axis and motors as well. Absolutely not change. Left side of the bed appears to be .100 to .150mm higher than the right side of the bed.

What is *particularly* upsetting about this is that I cannot understand what good the "auto-leveling is" if it doesn't actually *level*!

I understand that 3D printers (especially of the RepRap variety) inherently require some tinkering, but I bought the Mk2 specifically because of the enhanced resolution, the new auto-leveling, and because of the famed "support", none of which has been particularly stellar so far.

In fact, following a chat conversation, I emailed support about missing manuals in my box, a missing extra print nozzle that I had ordered, and also this very issue, along with photos. That was three days ago and I have yet to hear a response. I imagine I'm going to have to spend more time emailing them again about these same issues tomorrow.

In the meantime, I will still bang my head against whatever nearest wall I have available while I try to figure out what's going on.

Re: Tried everything, can't get bed level.

Are you sure that the auto bed level feature is working for you? You should be able to see the Z-axis rotate (watch the small black covers on the motor) when it's moving across the bed. If it doesn't, check if you have the "G87 ; enable auto-level" in your gcode (not sure if needed in the latest fw)

Re: Tried everything, can't get bed level.

.1 mm is the measurement of a hair. The auto leveling feature should make the adjustment for you.

Re: Tried everything, can't get bed level.

I agree completely that the auto-level *should* be making the adjustment for me, but clearly it is not. 🙁

I am estimating the .1mm part. I am basing that on estimated layer height, and the fact that when the left side is a bit too squished the right side is not quite squished enough.

I am using the V2Calibration from the SD card. Shouldn't the auto-leveling command already be present?

Re: Tried everything, can't get bed level.

Babak,

Would you please confirm, that the print nozzle does not touch the print bed during the bed leveling procedure?

If the PINDA probe is too high, then it is not uncommon, that the PINDA probe triggers first after the nozzle touches and actually bends the print bed low. This is very easy to overlook and it has clearly a detrimental effect on the bed leveling accuracy.

Vojtech

Re: Tried everything, can't get bed level.

>I have tried using the new beta firmware with bed calibration and without. No matter what at least one or more points will read "Z- 9.99" during calibration.

The number "Z-9.99" is the target Z value when lowering the probe. This value will never be reached, the movement will stop once the sensor triggers. So you may safely ignore the confusing "Z-9.99" message.

> Isn't that *exactly* what PINDA and "full mesh leveling" was supposed to solve? I am having a difficult time understanding why this can't be adjusted for with a combination of PINDA and firmware.

The firmware estimates the print bed shape from the measurements of the 9 sensor target points. As long as the printer is built correctly (axes move freely and linearly, the belt pullies do not slip), the bed leveling works. If some of the assumptions do not hold, the firmware is blind: It cannot detect these deviations from the model from just the 9 measurements. I hope this explains why the bed leveling is not a panacea.

> Left side of the bed appears to be .100 to .150mm higher than the right side of the bed.

The firmware works in a following way: It measures the 9 points, which it then interpolates by a smooth tensor product 2D quadric function. Because it is an interpolating function, the extrusion width shall be correct over the calibration points. What happens between the calibration points depends on how the actual machine matches the model. The 2D quadric function, with which we interpolate the 9 points, cannot possibly model a twist of the Y table for example.

I understand the bed calibration issues are frustrating, but the firmware could only correct for what it is able to measure. And there are many ways, how the printer geometry could get deformed.

Vojtech

Re: Tried everything, can't get bed level.

... Is it possible that the cutouts in my frame aren't correct?

There's always the possibility of a manufacturing error, you would need to email Prusa and talk to them about the measurements to take to check it all...

I have tried using the new beta firmware with bed calibration and without. No matter what at least one or more points will read "Z- 9.99" during calibration, I'm sure that's related to my y axis not sitting in the frame right. I can actually tighten/loosen the pinda probe 2mm to the left or right and the points it reads z- 9.99 will move around.

I've spent many hours trying to resolve this and I'm really starting to lose my patience. 😥
I feel for you as I had lots of problems with trying to get my MK1 level and never did to any degree of satisfaction. It's so frustrating. Especially as you just know that it is probably something really simple that's causing it!

Has anyone had an issue with left to right adhesion and actually fixed it? Or experienced my issue with my bed not settling in the frame correctly?

I was very lucky with my upgrade as it went very well overall. When I first tried to calibrate it I was having lots of adhesion problems with the V2 Calibration print and the first layer of any object. In the end I took the Live Z adjust down until the plastic could barely get out of the nozzle as it was so close and then took it up steadily until I was happy with the look and adhesion. It turns out it has to be closer than you'd imagine! So far every single print has worked perfectly! I've printed more things in the last weeks than we did in the whole 8 months we've had the MK1!

I would recommend taking a day off fro it, do something completely unrelated to 3D printing and then come back and try too lock at the problem from a different angle...

Hope you get it fixed soon!

Re: Tried everything, can't get bed level.

Vojtech,

The nozzle is definitely not touching the bed at all. In order for me to be able to print I have to have a Z-offset of -1.100 or so. Nozzle has never crashed into the bed once, and I have the PINDA probe matching the directions exactly in height.

Re: Tried everything, can't get bed level.

Hummm

To my mind, having a Z offset of -1mm or greater seems to imply that the probe is too low.

Yes, the probe sensitivity can vary and the position referred to in the manual is just a starting point, but probe height is something that can easily be changed to get everything spot-on without having to use Z adjust to the extent you have.

This is what I did:

1. Set Z adjust to 0.000
2. Auto-home
3. Check nozzle gap
4. Raise Z and adjust probe
5. Repeat from 2. until nozzle gap is close to 0.18mm (or 0.13mm for firmware 3.0.5)
6. Run V2 Calibration and use live adjust to get final height - should need very little adjustment.

I use feeler gauges to check the nozzle gap, but remember that a sheet of copy paper is about 0.1mm thick and can be used to get close approximations.

Peter

Re: Tried everything, can't get bed level.

Peter,

Good advice, and I will certainly do those steps, but I don't see how this would help solve the problem. The point at which the probe will "detect" the bed will not change. It seems like this will only give me a smaller Z offset, but I am having a hard time understanding how this would help the issue being described.

-Babak

Re: Tried everything, can't get bed level.

Babak

I agree, but we are still running RC firmware and I have never trusted live adjust, even with the Mk1. Quite honestly, I am of the opinion that we really are starting to ask too much of a little 16MHz microprocessor... I take the view that the less it is stressed out, the better it will work (same as humans really).

And there only needs to be a very minor and hard-to-find fault in the firmware in the "Live Adjust" code for there to be a major issue with the results.

I think it is worth you trying to get the probe set as precisely as possible and to see if it makes a difference. If it does, then the programmers will have something to look for.

Peter

Re: Tried everything, can't get bed level.

I had one print tonight, where I think the Pinda probe knocked off a structure, due to stringing and retraction with PETG. I think the electronics struggles with the demands made upon it.

I am thinking of applying 32 bit Duet electronics with David Crockers DC42 differential I/R level sensor.

Re: Tried everything, can't get bed level.

Babak

I agree, but we are still running RC firmware and I have never trusted live adjust, even with the Mk1. Quite honestly, I am of the opinion that we really are starting to ask too much of a little 16MHz microprocessor... I take the view that the less it is stressed out, the better it will work (same as humans really).

And there only needs to be a very minor and hard-to-find fault in the firmware in the "Live Adjust" code for there to be a major issue with the results.

I think it is worth you trying to get the probe set as precisely as possible and to see if it makes a difference. If it does, then the programmers will have something to look for.

Peter

Peter,

I understand your wanting to help and I appreciate your earnestness, but I have to respectfully disagree on some points. RC = Release Candidate, which means that to the best of the developer's knowledge, the firmware is considered stable and ready, so my first assumption will not be that just because it's RC that they are going to fix it. In fact, if RC firmware has a problem it is more likely that it will NOT be fixed before "official" release.

Humans and processors are nothing alike. Microprocessors don't work any differently when they are "stressed out", and least not in the context your are using it. I promise you that any problems we are currently having would also exist if we swapped the processor for a faster one and recompiled the same code for it. The issue is with the code, not the MCU.

Yes, while, small flaws in firmware can result in large differences, I do not see how such a bug could skew the Z axis in this way. The adjust offset is clearly working, as it does move the first layer higher or lower accordingly. The issue is with the mesh bed leveling, not the "live adjust". There is no reason not to "trust" live-adjust.

While I don't consider the position of my probe to be "less precise", I think your suggestion of trying to lower it and move it closer to the print bed doesn't hurt. When I get back on Monday I will give it a shot and see what the results are.