Z-offset affects bed mesh creation in Ender3S1 with F4 SoC

[abra5umente]

Describe the bug When levelling with a z-offset of anything other than 0, it affects the way the mesh is produced. Consistently reads one side MUCH higher than the other, seems to flip around from side to side too.

To Reproduce Steps to reproduce the behavior:

1. Perform a manual tramming using a feeler gauge or piece of paper.
2. Set a Z-offset in the menu (either by babystepping or by using the wizard)
3. Perform an auto level procedure (I leave the bed off, but doesn't seem to make a difference)
4. Note how one side will be particularly higher than the other, or there will be random valleys and hills in the mesh that you know aren't there.
5. Clear the z-offset back to 0
6. Restart the printer
7. Home the printer
8. Perform a bed level procedure again
9. Note how the mesh is more "true to life" and the results are much different.

Expected behavior The z-offset should not affect the measurements of the bed.

Screenshots My glass bed with a smooth, flat surface, same as read on my other printer with a BLTouch. I have also verified the CRTouch on that printer which performs the same as the BLTouch

Version (please complete the following information): Latest firmware, 2.1.0.1, build date May 27th 2022. Ender 3 S1 with F4 chip.

Additional context Have tried using BLTouch with same results. Tried with heating and not heating the bed, have verified that there is no power backfeed from Raspberry Pi 4 to printer (using electrical tape over the power pins).

[nilsafendi]

I have been having the hardest time trying to get a level bed with my S1. The bottom left quartile was always smushed and the upper right quartile was the opposite. There was also a random vertical dip that spanned from the middle of the bottom and upper left quartiles.

I'm on vacation without access to my printer but I'm looking forward to trying this.

This issue seems plausible for my situation because I always did the z-offset prior to leveling the bed.

[abra5umente]

Here is that same bed, same printer, same room temperature, but with a z-offset of -2.35 set while probing.

And here it is again after probing with z-offset set to 0 and rebooting the printer.

[elkinator1]

I recently got a CR Touch for my Ender 3 v2. Everything went down hill from there. The stock firmware the CR Touch recommends moves my x-axis in the wrong direction. I switched to this firmware by someone's recommendation and it has been constant failed prints. After all sorts of troubleshooting I went back to manual leveling with a guage. Problem solved.

Prior to going back to manua leveling, I would tram with a heated bed and hotend. I would make adjustments until all 4 corners were exactly the same. Then do my Z offset, then ABL. Everything seemed perfect on the software end minus the center of my bed being lower. With all the youtubers and comments out there saying the beds are wrapped and solved by the ABL. I looked passed the warp bed until I was finally ready to print and it just kept failing.

I kept finding (according to the CRT) that the center was always lower no matter how much I tightened up the 4 corners. At the time, I had 25mm yellow springs that I maxed out the tension on to try and at least bring the 4 corners below center. None of this mattered of course but it was part of my troubleshooting journey. From there I got blue 30-35mm blue springs and maxed them out. I made sure my glass was not lifting from the bed at all and the results were the same, the center of my bed was some how still lower, I couldn't believe it.

From there I gave up and set my printer to its z-home position before disabling steppers. I moved the x and y axis and found my bed was much lower on the right side (which made sense with the way my printer was failing). I decided to just try manually leveling again and then ran the bed traming and the ABL/UBL with 8 touch points.

The results of both showed a fairly level bed (even the center) up until the last row on right side, which according to the CRT it was way to high.

The end, I either have a defective CRT (which I doubt), I'm doing something wrong, or there is a bug in the firmware.

On Sat, Jun 11, 2022, 6:31 AM abra5umente @.***> wrote:

Describe the bug When levelling with a z-offset of anything other than 0, it affects the way the mesh is produced. Consistently reads one side MUCH higher than the other, seems to flip around from side to side too.

To Reproduce Steps to reproduce the behavior:

1. Perform a manual tramming using a feeler gauge or piece of paper.
2. Set a Z-offset in the menu (either by babystepping or by using the wizard)
3. Perform an auto level procedure (I leave the bed off, but doesn't seem to make a difference)
4. Note how one side will be particularly higher than the other, or there will be random valleys and hills in the mesh that you know aren't there.
5. Clear the z-offset back to 0
6. Restart the printer
7. Home the printer
8. Perform a bed level procedure again
9. Note how the mesh is more "true to life" and the results are much different.

Expected behavior The z-offset should not affect the measurements of the bed.

Screenshots [image: My glass bed with a smooth, flat surface, same as read on my other printer with a BLTouch. I have also verified the CRTouch on that printer which performs the same as the BLTouch] https://user-images.githubusercontent.com/87623193/173184024-239a7b1e-66b2-44b0-beb5-328789aa0adf.png

Version (please complete the following information): Latest firmware, 2.1.0.1, build date May 27th 2022. Ender 3 S1 with F4 chip.

Additional context Have tried using BLTouch with same results. Tried with heating and not heating the bed, have verified that there is no power backfeed from Raspberry Pi 4 to printer (using electrical tape over the power pins).

— Reply to this email directly, view it on GitHub https://github.com/mriscoc/Ender3V2S1/issues/195, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZI3DIQBKPH3JJ3VIJIUBK3VORTHJANCNFSM5YP357BA . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[abra5umente]

I recently got a CR Touch for my Ender 3 v2. Everything went down hill from there. The stock firmware the CR Touch recommends moves my x-axis in the wrong direction. I switched to this firmware by someone's recommendation and it has been constant failed prints. After all sorts of troubleshooting I went back to manual leveling with a guage. Problem solved. Prior to going back to manua leveling, I would tram with a heated bed and hotend. I would make adjustments until all 4 corners were exactly the same. Then do my Z offset, then ABL. Everything seemed perfect on the software end minus the center of my bed being lower. With all the youtubers and comments out there saying the beds are wrapped and solved by the ABL. I looked passed the warp bed until I was finally ready to print and it just kept failing. I kept finding (according to the CRT) that the center was always lower no matter how much I tightened up the 4 corners. At the time, I had 25mm yellow springs that I maxed out the tension on to try and at least bring the 4 corners below center. None of this mattered of course but it was part of my troubleshooting journey. From there I got blue 30-35mm blue springs and maxed them out. I made sure my glass was not lifting from the bed at all and the results were the same, the center of my bed was some how still lower, I couldn't believe it. From there I gave up and set my printer to its z-home position before disabling steppers. I moved the x and y axis and found my bed was much lower on the right side (which made sense with the way my printer was failing). I decided to just try manually leveling again and then ran the bed traming and the ABL/UBL with 8 touch points. The results of both showed a fairly level bed (even the center) up until the last row on right side, which according to the CRT it was way to high. The end, I either have a defective CRT (which I doubt), I'm doing something wrong, or there is a bug in the firmware. … On Sat, Jun 11, 2022, 6:31 AM abra5umente @.> wrote: Describe the bug When levelling with a z-offset of anything other than 0, it affects the way the mesh is produced. Consistently reads one side MUCH higher than the other, seems to flip around from side to side too. To Reproduce Steps to reproduce the behavior: 1. Perform a manual tramming using a feeler gauge or piece of paper. 2. Set a Z-offset in the menu (either by babystepping or by using the wizard) 3. Perform an auto level procedure (I leave the bed off, but doesn't seem to make a difference) 4. Note how one side will be particularly higher than the other, or there will be random valleys and hills in the mesh that you know aren't there. 5. Clear the z-offset back to 0 6. Restart the printer 7. Home the printer 8. Perform a bed level procedure again 9. Note how the mesh is more "true to life" and the results are much different. Expected behavior The z-offset should not affect the measurements of the bed. Screenshots [image: My glass bed with a smooth, flat surface, same as read on my other printer with a BLTouch. I have also verified the CRTouch on that printer which performs the same as the BLTouch] https://user-images.githubusercontent.com/87623193/173184024-239a7b1e-66b2-44b0-beb5-328789aa0adf.png Version (please complete the following information): Latest firmware, 2.1.0.1, build date May 27th 2022. Ender 3 S1 with F4 chip. Additional context Have tried using BLTouch with same results. Tried with heating and not heating the bed, have verified that there is no power backfeed from Raspberry Pi 4 to printer (using electrical tape over the power pins). — Reply to this email directly, view it on GitHub <#195>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZI3DIQBKPH3JJ3VIJIUBK3VORTHJANCNFSM5YP357BA . You are receiving this because you are subscribed to this thread.Message ID: @.>

I don't think the CRTouch is bad hardware - from my tests it's actually more accurate than the BLTouch. I don't know what is going on with the firmware to make bed levelling screw up so bad though - I've used ABL many, many times and on my Ender 3 Pro (with the same glass bed!) I could just plonk the bed on, make the corners relatively level, run G29 before my print and put M420 S1 in the start up code and it would fire away and I'd have perfect first layers, every single time.

The only differences I can see in the firmwares (I compiled my own firmware on the Ender 3 Pro and downloaded the repo to see what this one is doing) is that they have the multiple probes enabled, which I turned off. I don't see how exactly hardware could make the ABL so dodgy, but it could be that it's some strange issue with the F4 chip, or something. I'm going to compile firmware with multiple probes disabled and see if that makes a difference, you typically only need that enabled when your probe is crappy and you need higher accuracy.

[abra5umente]

Alright, so I've compiled my own firmware with multiple probing disabled, and the result is much better.

1. It is literally 3 times faster, which is nice.
2. Z-offset was set to -2.35 here and the result is basically the same as when it was 0 with stock settings.

[abra5umente]

New firmware is here, if anyone wants to test it out. [https://github.com/.../Ender3S1F1-4/releases/tag/20220527_probefix]()

[webmogul1]

If by disabling multiple probing solved the problem, it could be that the probe is not being consistent. Have you tried testing the probe? (Advanced Menu) (M48 Probe Test) It will show the probe deviation.

[abra5umente]

Probe is accurate down to 0.001 or thereabouts. I'd say the actual issue is the multiple probing, but I'm not sure why.

[mriscoc]

M48 should be done at the same position where the errors are occurring, the deviation is not the real accuracy because if one of the measurements is out of the other 10, it does not affect too much to deviation. More interesting is the Range value. The strange is that the same algorithms with the same probe in a V2 give much better values than on some S1.

[abra5umente]

It occurs in random positions every level - sometimes on the right side, sometimes on the left side. I've been through the configuration.h file and I can't see anywhere that would skew any results. A few people have said that simply reducing it to one fixes it - so I'm assuming that it's an electromagnetic interference issue that just compounds with each press. I did try with the bed heated and not heated, and it made no real difference. Do you have an S1 to test with?

[mriscoc]

I don't have a S1, that's why that error is difficult to reproduce and fix for me. But I have reports from other S1 users that don't have errors, maybe it could be a problem with the ground. Please test with MULTIPLE_PROBING = 3 and EXTRA_PROBING = 0.

[abra5umente]

I don't think it's a problem with ground, but I'll have a play tomorrow. I've had another printer with a BLTouch plugged into that same port for years and it's been fine - but I'll see what I can come up with.

[trulow]

I also am noticing an issue with the CR-Touch. I find that probing isn't accurate and the results are jumping around by .02 +/- deviation. You can really notice this when using the tramming wizard as the results are never consistent especially after a printer restart.

[mriscoc]

I also am noticing an issue with the CR-Touch. I find that probing isn't accurate and the results are jumping around by .02 +/- deviation. You can really notice this when using the tramming wizard as the results are never consistent especially after a printer restart.

It could be not only the CR-touch, the mechanical accuracy of the printer is 0.1 mm then if the printer moves it is possible to have large deviations.

[albertoabrao]

New firmware is here, if anyone wants to test it out. https://github.com/abra5umente/Ender3S1F1-4/releases/tag/20220527_probefix

Thank you for sharing the result of your work, @abra5umente !

@mriscoc, my printer is an Ender 3 S1 (F4). I can report having the exact same issue reported by @abra5umente , and his version of the firmware solves the problem.

I assume it has something to do with the lack of memory on the F4 boards, some kind of overflow happening when you try to store too many values? Not very familiar with Marlin, just a wild guess.

The delta between corners and center was 0.1 when perfectly level, now it's 0.02. That's without touching a single screw, just new firmware. All else is the same (temperature etc).

It being faster is certainly nice too. The CR Touch is consistent enough otherwise, as reported by M48 Test - which incidentally also gives me a lower margin of error now.

It would be amazing if we can get a toggle for that option on the next version.

[mriscoc]

The next version will allow to edit the multiple probe value.

[Moobien]

I have had something similar. I will Tram my bed then set my z offset using a feeler gauge. then after G29 has run on a print the nozzle seems to be a lot higher than what it was set at in the z offset, often to the effect of having to lower it up to 0.12mm (and I use a .1mm feeler but also notice a range of -0.12 to even 0 on the mesh

[nmanclank]

I have had something similar. I will Tram my bed then set my z offset using a feeler gauge. then after G29 has run on a print the nozzle seems to be a lot higher than what it was set at in the z offset, often to the effect of having to lower it up to 0.12mm (and I use a .1mm feeler but also notice a range of -0.12 to even 0 on the mesh

I'm having this exact same issue.

[GRRK]

Is it possible that X axis is inverted in auto leveling? I have 3S1 F4. My auto leveling seems to apply wrong offset the way bed is tilted in X. In last test I ran 4x4 auto leveling and then manually mirrored the values on X.... finally I can use whole bed with this approach.

[hayden-t]

to test, adjust bed screws to extreme opposites, to create maximum slant, run abl, and then comand move the head around to see if it follows the slant correctly

[arthurkok2]

on this line in creality's S1 firmware, they have BLTOUCH_DELAY configured. This may cause some differences?

https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L771

[Moobien]

on this line in creality's S1 firmware, they have BLTOUCH_DELAY configured. This may cause some differences?

https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L771

Could that be the difference between HS mode enabled/disabled in this firmware?

[ghost]

on this line in creality's S1 firmware, they have BLTOUCH_DELAY configured. This may cause some differences? https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L771

Could that be the difference between HS mode enabled/disabled in this firmware?

They seem to have HS mode enabled too https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L816

[hayden-t]

HS mode works on my S1 pro, and without it probe does not work. I can do 9x9 or any size with HS, and it matches my bed contour.

[mriscoc]

on this line in creality's S1 firmware, they have BLTOUCH_DELAY configured. This may cause some differences?

https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L771

If you don't define a delay there is a default value of 500ms https://github.com/CrealityOfficial/Ender-3S1/blob/e46e1a85011fb91cd8f8c1cc9c5a1da6887f045f/Marlin/src/inc/Conditionals_post.h#L2926

[mriscoc]

on this line in creality's S1 firmware, they have BLTOUCH_DELAY configured. This may cause some differences? https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L771

Could that be the difference between HS mode enabled/disabled in this firmware?

They seem to have HS mode enabled too https://github.com/CrealityOfficial/Ender-3S1/blob/main/Marlin/Configuration_adv.h#L816

They enable HS because their mesh is only 4x4

[harley44howard]

I'm having similar issues with ender 3v2, sprite pro, dual z. Having a hard time leveling, can get zeros on all 4 corners using bed tramming. When do a mesh, no matter the size, I end up getting obscure high and low spots. Set z offset and print a large flat square to live adjust z offset while in printing the skirt I can see how one side of my bed is higher than the other.. I have squish on one side and to far away from the bed on the other. My only work around was doing what the original post was about and setting zoffset to zero them doing bed tramming, auto mesh, edit mesh manually then set z offset and live adjust.. it Is nowhere near perfect but I'm atleast printing now. I also level the bed turning the knobs what printing because I noticed still one side to high in comparison to the other.. Basically ignored my mesh and adjusted z offset and leveling knobs while watching the first layer of a print... I also have linear y axis and 3 point leveling.. adding the 4 point back didn't change anything.. it is like the probe becomes inaccurate or just doesn't actually set the mesh during the print anymore.. yes I have the proper gcode set, and it does load the mesh before the print. I also tried to tilt the mesh before, and without tilting.

[hayden-t]

"They enable HS because their mesh is only 4x4" HS mode completes upto 9x9 ...

@harley44howard - the problem is the magnetic medium layer is not even (put a ruler edge across it), and even though the probe detects this it does not seem to translate into enough compensation (im still investigating why) . as a manual temp solution just put some foil under between the build plate and mag layer in the low spots

[harley44howard]

"They enable HS because their mesh is only 4x4" HS mode completes upto 9x9 ...

@harley44howard - the problem is the magnetic medium layer is not even (put a ruler edge across it), and even though the probe detects this it does not seem to translate into enough compensation (im still investigating why) . as a manual temp solution just put some foil under between the build plate and mag layer in the low spots

@hayden-t I'll take a gander at that. It was quite severe though and if I remember correctly it's the same bed and magnetic pad from the firmware before this without any issues. Also tried different pei sheets (2 new ones) I have 2 new magnetic sheets as well I guess I could try a new one.

[albertoabrao]

I noticed that the original firmware for the S1 uses the following feedrates:

Recv: echo:; Maximum feedrates (units/s): Recv: echo: M203 X300.00 Y300.00 Z5.00 E25.00

Professional Firmware uses Z20 or so.

I've been getting more consistent results by using the same values as mentioned above, on the Professional Firmware, of course - Creality's is just terrible. Tried raising Z to 10, and it was significantly worse already.

[hayden-t]

that is your max z speed, do you think the leveling runs at max z ? im not so sure

[albertoabrao]

that is your max z speed, do you think the leveling runs at max z ? im not so sure

I did not think much, I just changed it to see what would happen.

The end result was that I got a slower probe movement, both up and down the axis, than I was able to by reducing the probe speed - there, it would reduce the downwards speed, but not so much the reverse.

It also resulted in more consistent probing results, so I decided to keep it. And share, in case it helps someone else.

Now, as to why this is the case - or not, in case I am wrong - mriscoc or someone else more knowledgeable than I am may be willing to elucidate.

[jordanpenard]

I'm glad I found that thread, because I was about to loose my mind with the auto bed leveling on my S1 (F1) . I'm seeing similar issues than what @abra5umente reports, even though I didn't try to run the ABL with zoffset at 0. Where I am so far is that I ran multiple 9x9 ABL, plotted the results in octoprint bed level display, and tried to find the tallest and lowest reported points on my machine while the level adjustment is enabled, and everything is out of place, the nozzle distance to the bed changes massively between those points, by around 0.2/0.3 mm, it even caused a few holes in my bed surface. My next step is to do a manual bed leveling and recorde le values to compare with the ABL values. I will try the FW that @abra5umente posted earlier too. What I'm still unsure of :

• is the ABL amplifing the values measured and ends up over compensating?
• are X and Y swapped around?

I shall come back with my findings once I get the time to do more testing.

[hayden-t]

please look at my 2 latest videos regarding abl investigations https://www.youtube.com/user/hthring

[mriscoc]

please look at my 2 latest videos regarding abl investigations https://www.youtube.com/user/hthring

You need to consider the weight of the head and the impact on the deformation of the X axis.

[hayden-t]

i think there are several reason the plane of travel can become skewed

[JFingerle]

Is this still an issue an F4 boards? Thank you!

[DanMan32]

Could the problem everyone (or most) are seeing is due to too much slop in the Z axis, especially on the right side? The z axis on the Enders is supposed to be 0.04mm. When manually tramning corners even before using this firmware, I noticed that there was a z-offset deviation when moving the Z up vs down. I've also noticed that when printing 5 squares bed level/z-offset test, the left side may be perfect but right side not as good. Most Z movement during a print is in the up direction, but mesh building using ABL is read in the down direction, so right there there is a deviation in the mesh offset vs what actually exists during print.

Then again the mesh display is based completely on ABL readings, but if there's Z axis slop, that would explain the deviations between readings.

Also though I didn't notice it with this firmware, with stock Creality firmware for BL/CR, you'll note that 0 will show twice when moving Z axis. IOW going down, you'll see 0.2,0.1,0.0, 0.0, -0.1, -0.2. And yet the two zero points will yield a different physical Z clearance from the bed.

[mhaas]

Just to add one data point to the discussion. I am trying to figure out some bed adhesion issues on my S1 and suspected that the ABL is an issue for me. I tried the "disable multiple probing" fix described here on latest master with the S1 UBL config.

Results with stock professional:

Results without multiple probing:

The actual numeric values are different, but the overall shapes seems to have been retained.

Will go back to stock professional for now; my first layer is still not sticking and I will try other means of fixing that.

[DanMan32]

You do seem to have a lot of variation, but as you said, the pattern is close.

Not sticking to the bed typically would be due to z-offset with too much of a gap, dirty bed (clean with soap/water and/or IPA) or bed/nozzle temp too low.

Remember, the probe is measuring z height of the bed and the probe, not the nozzle. The z-offset sets the difference between the prone '0' and nozzle.

I found with 5 squares level/height tests that my nozzle has to be closer than 0.1mm that one would normally set with paper gap. Otherwise the stripes in the squares are too far apart.

[albertoabrao]

After much tweaking, so far, the best settings for my S1 with the F4 board are the following:

1) Do not use more than 4x4 for mesh points.

2) Go to Advanced -> Probe Settings -> Multiple Probing. Set that to 0. For some reason, it's more consistent that way... and faster. You'll appreciate that, because of what you'll do on the next step.

3) on Control > Motion > Speed, use the following values (taken from stock firmware):

• Max X Speed: 300
• Max Y Speed: 300
• Max Z Speed: 5
• Max E Speed: 25
• Save settings, turn printer off, then on again, just for good measure.

That allows you to get consistent results between measurements.

4) To measure Z offset, I usually do the following.

• First, remove the rubber nozzle protector.
• Now, warm up the hot end - and only that - to 240-ish. Unload filament (if needed), and use a brass brush to clean the nozzle really, really well.
• Let it cool to room temperature. Grab a snack or something.
• The rest should be run with the bed and nozzle COLD.
• Clean the bed really, really well. If using the spring steel bed that came with the S1, spray some isopropyl alcohol (70% or 99%, if you can find the latter) on it, rub with a clean toothbrush, then wipe with microfiber towel.
• Now run the tramming wizard until it is level. You don't need to be OCD about it, as long as it passes, it's OK. Ask me how I know... :)
• Get a piece of "Wax Paper", the one you line your trays with when baking. It's just the right thickness, and does not absorb moisture, thus giving a very consistent "drag" every single time.
• Now, Move Z to Home. It will probe the middle of the bed - twice.
• Just before it goes down after the second probe, slide the wax paper between nozzle and bed. Then adjust. You are aiming for a confident bite of the nozzle on the paper... you're able to slide it back and forth, yes, but you can feel it's there. The feeling is slightly different for glass and the spring steel flexible bed that came with the printer. Aim for a "confident" bite and you'll get the gist of it after a few tries.
• Now, before you call it quits, remove the paper, hit Move Z to Home again, and make sure it's somewhat the same "feel" for the drag.
• If you do that twice - or thrice - you'll find that either a specific tiny range - say, 3.54 to 3.56 - or even a consistent number - 3.55? - is where you get satisfying results.

Now, let it cool again, install the rubber hotend protector, then go ahead and test! For PLA, that's all you need to do. For PETG, I add 0.02 extra Z offset. That's it.

I find that I only have to mess with it when I change the bed surface, and even then, it does end up around the same number I settled on before switching bed surface.

If you're lazy like I am, you can install two Cura plugins that will allow you to not have to add the extra Z offset between PETG and PLA:

• Z Offset Plugin
• Material Setting Plugin

After you install these from the Marketplace, just go to Settings -> Materials, then select the pane "Print Settings". On the bottom left corner, there's now a button called "Select settings". Select all the ones you'd like to customize for each material - e.g. brand of filament etc. One of the options I use is Z Offset, then I just add the 0.02 extra that I use for PETG filaments. No need to change every time - or should I say, forget every time? :)

Let me know how it works!

[albertoabrao]

Oh, three more things.

1. AFTER you do all of the above, before each print:

Heat bed to temperature, then run tramming wizard.

If the material requires a different bed temperature (e.g. PLA vs PETG), heat it to the temperature you'll use, then run Tramming Wizard.

That is especially true for the spring steel bed. Glass tends to be fairly consistent, you can usually get away with levelling it cold and it won't change much.

Spring Steel, on the other hand, varies wildly: level it cold, and when you have it at 75C it'll be completely out of whack...

...at least that's the case with my printer. :)

2. If you get a mesh that's just terrible with the spring steel bad, relax. It's just that it sucks. It'll print fine if you tune it all up, even if the mesh shows up all red and weird. Seems to be a common theme with the Creality spring steel beds.

Try a glass bed, and you'll see the difference right away. Pretty much perfectly level.

Again, both print fine despite that. I am mentioning it so that you don't get alarmed.

3. If you want to try a glass bed, watch out for the probe tip vs. the clips you'll use to secure it to the bed.

I am still looking for a clip that lets me use the whole area of the glass bed without being involved on collisions with the retracted probe.

[mhaas]

I did some experiments this morning and I thought I would share the raw results. I haven't had time yet to analyze.

I tried all combinations of the following:

• Z Offset 0, Z Offset -3.55
• HS mode on / off
• MULTIPLE_PROBING on/off

Results:

Professional Firmware

z_offset: 0.0, highspeed_mode: 0

Regular:

-0.09   -0.09   -0.07   0.02    0.05
0.10    0.05    0.00    0.04    0.04
0.04    0.03    -0.03   0.04    0.06
-0.02   -0.06   -0.11   -0.04   -0.02
-0.02   -0.09   -0.18   -0.13   -0.14

Normalized:

-0.07   -0.06   -0.05   0.04    0.08
0.13    0.08    0.02    0.07    0.06
0.07    0.05    -0.00   0.06    0.09
0.00    -0.03   -0.08   -0.01   0.00
0.01    -0.07   -0.16   -0.11   -0.11

z_offset: -3.55, highspeed_mode: 0

-0.11   -0.10   -0.07   0.01    0.05
0.11    0.05    -0.00   0.04    0.02
0.03    0.02    -0.03   0.03    0.06
-0.03   -0.04   -0.10   -0.03   -0.01
-0.03   -0.11   -0.19   -0.15   -0.15

Normalized:

-0.08   -0.07   -0.04   0.04    0.08
0.13    0.08    0.03    0.06    0.05
0.06    0.05    -0.00   0.06    0.09
0.00    -0.01   -0.07   -0.00   0.02
-0.00   -0.08   -0.16   -0.12   -0.13

z_offset: 0.0, highspeed_mode: 1

-0.11   -0.10   -0.07   0.00    0.06
0.09    0.04    -0.01   0.02    0.03
0.02    0.01    -0.04   0.02    0.06
-0.04   -0.05   -0.10   -0.03   0.00
-0.04   -0.12   -0.20   -0.16   -0.16

Normalized:

-0.08   -0.07   -0.04   0.04    0.09
0.12    0.07    0.03    0.06    0.06
0.05    0.05    0.00    0.06    0.09
-0.00   -0.01   -0.07   0.00    0.04
-0.00   -0.09   -0.17   -0.12   -0.12

z_offset: -3.55, highspeed_mode: 1

 -0.12  -0.11   -0.08   -0.00   0.06
0.07    0.02    -0.02   0.01    0.02
0.01    -0.00   -0.04   0.01    0.04
-0.05   -0.06   -0.10   -0.03   0.01
-0.05   -0.13   -0.21   -0.17   -0.16

Normalized:

-0.08   -0.07   -0.03   0.04    0.10
0.12    0.07    0.03    0.06    0.06
0.05    0.04    0.00    0.05    0.09
-0.01   -0.01   -0.06   0.01    0.05
-0.01   -0.09   -0.17   -0.13   -0.12

Professional Firmware, no MULTIPLE_PROBING

z_offset: 0.0, highspeed_mode: 0

-0.12   -0.11   -0.09   -0.01   0.04
0.04    0.02    -0.02   0.02    0.02
0.00    -0.00   -0.05   0.01    0.04
-0.08   -0.07   -0.11   -0.02   0.01
-0.07   -0.13   -0.21   -0.16   -0.16

Normalized:

-0.07   -0.06   -0.04   0.04    0.09
0.09    0.07    0.03    0.07    0.07
0.05    0.05    -0.00   0.06    0.09
-0.03   -0.02   -0.06   0.02    0.06
-0.02   -0.08   -0.16   -0.12   -0.11

z_offset: -3.55, highspeed_mode: 0

-0.12   -0.09   -0.07   0.00    0.06
0.03    0.02    -0.02   0.02    0.04
-0.01   -0.01   -0.05   0.01    0.06
-0.07   -0.06   -0.10   -0.03   0.01
-0.08   -0.13   -0.21   -0.17   -0.16

Normalized:

-0.07   -0.05   -0.03   0.05    0.10
0.07    0.06    0.03    0.07    0.09
0.03    0.04    -0.00   0.06    0.10
-0.02   -0.01   -0.06   0.02    0.06
-0.03   -0.09   -0.17   -0.13   -0.11

z_offset: 0.0, highspeed_mode: 1

-0.12   -0.09   -0.06   0.00    0.05
0.02    0.01    -0.01   0.02    0.04
-0.01   -0.00   -0.04   0.00    0.06
-0.07   -0.06   -0.10   -0.04   -0.00
-0.09   -0.13   -0.20   -0.16   -0.14

Normalized:

-0.07   -0.04   -0.02   0.05    0.10
0.07    0.06    0.03    0.07    0.09
0.03    0.04    0.00    0.05    0.10
-0.02   -0.01   -0.06   0.00    0.04
-0.04   -0.09   -0.15   -0.12   -0.09

z_offset: -3.55, highspeed_mode: 1

-0.12   -0.09   -0.06   -0.01   0.05
0.02    0.01    -0.02   0.00    0.04
-0.02   -0.01   -0.05   -0.01   0.06
-0.06   -0.06   -0.11   -0.08   -0.03
-0.09   -0.13   -0.18   -0.15   -0.11

Normalized:

-0.07   -0.04   -0.01   0.04    0.10
0.07    0.06    0.03    0.05    0.08
0.03    0.04    0.00    0.04    0.11
-0.01   -0.01   -0.06   -0.03   0.02
-0.04   -0.08   -0.13   -0.10   -0.06

gcode: https://gist.github.com/mhaas/a1e9866803bc87815e1bee93b12346c2

[mhaas]

I compared the meshs I generated above to understand which setting has the biggest influence. I compared both the average difference between the meshes as well as the biggest absolute (single-cell) difference. Here's my results: https://github.com/mhaas/s1-ubl-debug/blob/main/mesh.ipynb - scroll down to the diagrams.

Overall, it seems that indeed the MULTIPLE_PROBING has the biggest influence by far.

On the weekend, I will manually generate a mesh as a ground truth and see which of the different settings comes closest.

Additional things one might investigate based on the data:

• Influence of z-offset?
• comparing normalized meshs instead of absolute meshs
• comparison to ground truth

[mhaas]

I spent a bit more time looking into this. On my printer, even at Z offset 0, there is a marked difference between Multiple Probing enabled/disabled.

I am collecting all information in a wiki: https://github.com/mhaas/ender-3-s1/wiki/Automatic-Bed-Leveling-Issues

You are cordially invited to contribute, edit, update etc.

My current hypothesis is that the Z probe feedrate has an impact on the probe reliability: https://github.com/mhaas/ender-3-s1/wiki/Automatic-Bed-Leveling-Issues#the-z_probe_feedrate-hypothesis

I don't have access to my printer right now. I think the next step will be to get the devug leveling output (see second link). If you can help out here that would be great!

[mriscoc]

Be aware that BLTouch and printer mechanical repeatability is no better of +-0.05 mm, so the difference falls within the error margin. M48 standard deviation is not the real accuracy more information about repeatability gives the range value.

[mhaas]

That's a good point. It is even possible that my finding that multiple probing has a big impact on the mesh is exactly because MP gives more accurate results. Still, the patterns that I see are between MP on and off do not seem to be enrirely random.

Going forward, I will include a test print in my experiments to understand if a given configuration has a positive or negative impact. Of course that is harder to automate.

By the way, I earlier found an issue in the Klipper repo indicating that BLtouch results depend on Z speed. Of course, that is a completely different firmware and architecture.

[grrmisfit]

something is off, doesnt matter what settings the mesh is wrong. I even swapped probes to verify it wasnt a probe issue. corners are usually off on the left side. i can clearly make a mesh and watch it be off when printing a test print. i manually level with paper and wizard tells me im off but when i adjust and go back to manual the height is clearly way off. short of going to klipper i dont know what to do.

[hayden-t]

@grrmisfit please see my above post https://github.com/mriscoc/Ender3V2S1/issues/195#issuecomment-1221262298

[grrmisfit]

@grrmisfit please see my above post #195 (comment)

Read it, tried it, still major variance on 2 left corners. I even disabled rods and made sure x gantry was level on both sides. Test print has left side grinding bed and right side sticking to barely sticking. I'm going to do a test print without using mesh and see the results. I've taken this printer apart multiple times and center prints are hit and miss with outside being non existent without wheel adjustment