Notifications
Clear all

Bed leveling issue (right side higher)  

Page 2 / 2
  RSS
Bunny Science
(@bunny-science)
Noble Member
Re: Bed leveling issue (right side higher)

I took a dial indicator and measured the drop from the platform down to the y-rods. The distance was about 10 mils more on the right side of the bed than the left. This was true at both the front and back of the build platform.

I suspect the right bearing mount slot is a tiny fraction narrower - enough to raise the right side of the platform relative to the y-rods. However, the amount was the same front/back. My y-rod spacing was within 0.02 mm difference front/back. The rods are parallel (which is most vital). Now, does that mean the platform is parallel to the x-rods? I need to go back and measure the left and right side distance between platform and X-rods. THAT is what must be equal.

The plastic bits and all the metal parts all have tolerance issues that can stack up. Even tightening the screws on the y-rod holders moves things. Adding 3 mm washers between the bolt heads and frame reduced that tendency.

Publié : 31/12/2018 6:47 pm
eBlueSkittle
(@eblueskittle)
Active Member
Re: Bed leveling issue (right side higher)

So after doing lots of tweaking and research I have found a working solution. I don't know if it's the solution for everyone, but it does resolve the bed level problem, which itself is an underlying issue with manufacturing, heat or something else.

All the tests and pieces I was modifying were focused on getting a level bed. I was adding shims under the y carriage smooth rods and shaving the mounts that held the smooth rods. No matter what I did, when I would run Prusa Mesh Leveling plugin on Octoprint the total bed variance would also be in the range of 1.xxx and then if I did -50 and 50 for advanced bed level correction the most I could bring down to was in the .8xxx range.

The reason for this was the items sitting ontop of the y carriage are the problem I suspect. So what was happening underneath didn't really matter because if what was above wasn't perfectly flat then it would always maintain that large variance. I don't know if it's a manufacturing defect or warping from heat, but something isn't level. I have only ever had my bed at 60 degrees C and only printed PLA. Can't imagine it is supposed to warp that quickly from so little use and the lowest temp PLA setting.

I found threads talking about spring methods for leveling but I felt they would possibly weaken over time and require adjusting as the spring tension got weaker. So I went with this method https://github.com/PrusaOwners/prusaowners/wiki/Bed_Leveling_without_Wave_Springs

After installing these nylock nuts and using Prusa Mesh Leveling plugin, I was able to get my total bed variance down to .017mm. It took 10 mesh levels and tightening/loosening between each one to get it to this point.

Hopefully this helps others who are having this issue as well.

Publié : 01/01/2019 8:27 pm
Chocki
(@chocki)
Prominent Member
Re: Bed leveling issue (right side higher)

I too have fitted springs and after using dial gauges, straight edges, multiple strip down and rebuilds etc etc, I finally got my bed variance down to below 0.010 with the bed at temperature, which then allowed me to fine tune the remaining variance (At least as good as I can, it's not perfect, but prints OK) using the Bed levelling adjustment from the LCD.
I still think, no I am certain there is a firmware issue going on here and from what I am seeing, the calculations cause a negative effect on the other half of the bed.
In other words, if you imagine a sine wave or S on it's side for anyone who does not know, if you get a dip on one side of the bed, you get a peak on the other, and these effects are applied by how big your original dip / peak is, so by getting the levels down to below 0.01, any variance is small enough to get away with it.

During all this,I have been through all calibrations many times and have also noticed that sometimes a Z calibration can leave the bed results really skewed, but not every time, you have to check the G81 output and if you have a really skewed result, run another Z calibration. I also printed new upgraded parts, but this made no difference, it was only after the springs and lots and lots of tiny adjustments and checks that I finally have something that can print a first layer over the whole bed (At least one that works more or less).
Also, the Z correction is still being applied even after many cm of height (I can see the lead screws turning), so the errors are being printed throughout the whole print.

I'm starting to trawl through the firmware, but my C coding is not really up to scratch so I do not fully understand the bed levelling yet or more importantly how to turn it off!.

Normal people believe that if it ain’t broke, don’t fix it. Engineers believe that if it ain’t broke, it doesn’t have enough features yet.

Publié : 02/01/2019 9:39 pm
Tinkerwest
(@tinkerwest)
Trusted Member
Re: Bed leveling issue (right side higher)


I too have fitted springs and after using dial gauges, straight edges, multiple strip down and rebuilds etc etc, I finally got my bed variance down to below 0.010 with the bed at temperature, which then allowed me to fine tune the remaining variance (At least as good as I can, it's not perfect, but prints OK) using the Bed levelling adjustment from the LCD.
I still think, no I am certain there is a firmware issue going on here and from what I am seeing, the calculations cause a negative effect on the other half of the bed.
In other words, if you imagine a sine wave or S on it's side for anyone who does not know, if you get a dip on one side of the bed, you get a peak on the other, and these effects are applied by how big your original dip / peak is, so by getting the levels down to below 0.01, any variance is small enough to get away with it.

During all this,I have been through all calibrations many times and have also noticed that sometimes a Z calibration can leave the bed results really skewed, but not every time, you have to check the G81 output and if you have a really skewed result, run another Z calibration. I also printed new upgraded parts, but this made no difference, it was only after the springs and lots and lots of tiny adjustments and checks that I finally have something that can print a first layer over the whole bed (At least one that works more or less).
Also, the Z correction is still being applied even after many cm of height (I can see the lead screws turning), so the errors are being printed throughout the whole print.

I'm starting to trawl through the firmware, but my C coding is not really up to scratch so I do not fully understand the bed levelling yet or more importantly how to turn it off!.

I would try the new 7x7 calibration that up on GitHub it work great for me. Just because the 9 points are level there still can be big changes in between. This firmware compensated for that and not only gave me a great first layer but Also a great top layer. Note I Also heated the Pinda to 30c before starting. Good luck I hope this helps. A lot more information at the link

https://github.com/prusa3d/Prusa-Firmware/issues/1239

Publié : 03/01/2019 3:49 am
Chocki
(@chocki)
Prominent Member
Re: Bed leveling issue (right side higher)

I tried the new firmware and it is better, but boy does it show how bad my heated bed is!, its an m shape on one side and a w shape on the other, no wonder the bed levelling cant compensate.

Now I have the new firmware, I'm trying a few things, and have refitted the original spacers, warmed the bed up to 100 Dec C with the screws loose so the bed can move, then tightened starting from the middle then the middle front and back, middle left and right and finally the corners and I have cone back over the pins and made sure they are all quite tight, the idea being that they will keep some tension on the heated bed once the temperature is dropped, but now I am back to the RHS being higher than the left.

At work now so can't do anymore until tomorrow, but I'm basically going to go through the whol printer, frame adjustment, distance measurments until I find the source of this issue, I read that the tilt to the right may be due to the cutout in the X frame being a fraction too narrow on the single right bearing resulting in the bearing sitting a fraction lower, so will be concentrating on this first, maybe take the edges off with a file if need be.

What I will say is that this modified firmware seems to give G80 N7, G81 results very consistent between runs, much better than the standard 3x3 G80 run, so is much better for getting a level bed. I still suspect I may well have to contact PRUSA for a replacement heater bed, but will give this last go at levelling with the bed pre-heated and under tension at lower temperatures and see what my results are like.

Also, I have an idea from the conversation at https://github.com/prusa3d/Prusa-Firmware/issues/1239 (Thanks john.k7).
It looks like one of the contributors disabled automatic mesh bed levelling altogether, whilst another implemented it with the 7x7 grid probing.
I'm wondering if this can be combined and triggered using the live-z value?.
AFAIK everyone uses a -ve live-z value so I wonder if the firmware could look at the value and always apply a -ve live Z adjustment even if the same value is actually dialed in +ve, but use the sign to enable / disable automatic compensation of the Z axis.

This would be for testing purposes only, basically you dial in your live z as normal setting the -ve value, when happy and you want to test without automatic mesh bed compensation being applied as the head moves across, dial in the same value but +ve.

I need to look at the firmware in much greater depth first and try a few things, but it may be possible. (I hope). The guys in the modified firmware link have already done all the hard work, so this would just enable disable parts, unless someone can come up with an easier way to enable disable from the LCD display without affecting anything else.

Normal people believe that if it ain’t broke, don’t fix it. Engineers believe that if it ain’t broke, it doesn’t have enough features yet.

Publié : 04/01/2019 4:15 pm
Chocki
(@chocki)
Prominent Member
Re: Bed leveling issue (right side higher)

Preheating the bed and then tightening the screws onto the original spacers certainly helped get a flatter bed, but the original problem with one side not level was as bad as ever, so this method would require accurate measurement and different length spacers.
I even stripped down the complete Y axis, loosened all the frame screws and reset the whole frame making sure distances were even and heights were accurate which did help with the right side being higher, but I just could not get rid of the twist in the right side no matter what I did, so have gone back to the springs.

I have managed to level the bed to within 0.01 variance using the 3x3 G80 G81 method bed preheated to 60 Deg, but when I run the 7x7 G80 N7 G81, the results are a horribly warped bed with a variance of 0.15.

I managed to get a successful full bed print using the LCD bed levelling and the 7x7 G80 N7 method, not perfect, but not bad either, tried the same print with the normal 3x3 G80 method, and got a much worse print with dragged through squares, non joined extrusions, basically, not acceptable as a first layer print.

So to summarise, Standard PRUSA firmware, I cannot print a successful full bed first layer, modified 7x7 or 49 point bed levelling, I can print a successful first layer.

The Github page with the complete discussion is here: https://github.com/prusa3d/Prusa-Firmware/issues/1239
And the firmware I am successfully using is here: https://github.com/prusa3d/Prusa-Firmware/files/2724250/firmware.zip

Normal people believe that if it ain’t broke, don’t fix it. Engineers believe that if it ain’t broke, it doesn’t have enough features yet.

Publié : 06/01/2019 12:43 pm
prusite
(@prusite)
Active Member
Re: Bed leveling issue (right side higher)

Hello
I have now tried to get this correct for the last two days, but alas, no luck.
I have measured everything, very carefully, and as far as I can tell, nothing is off or differs.
the bed to the upper frame when the bed is at the back-position, is exactly the same (or, as exact as I can measure).
26.75cm-ish.
The bed is 90 degrees in angle in all directions, it is also totally flat in all directions and diagonally.
I have tested 90-degrees against the bed and the frame in all directions that I could get in with my L-angle-tool.
My suspicion was that the z-axis got misaligned during first upward-synk, so I really tightened down all motor-screws on both sides.
I then manually put the z-axis in top to see if they stopped on different height, giving some sort of initial mis-alignment. But no, exactly 1cm to the top of the top-frame.
Have re-initialized the thing multiple times, but every time the right side directly catches the paper, hinting that it finds this side higher.
Continuing the setup and flat-first-layer-printing some tests, gives the same result, its flatter and thus 'higher' than the left side (getting pressed closer to the nozzle).

I then tried to get left and right side manually fixed in correct position (the initial step during setup), so that the nozzel almost touches both front-left and front-right, and the nozzel can freely swipe left and right (having a normal paper to protect the base).
If I then measure the top of the two x-axis rods in three places, left/center/right, it gives 18.75cm, so exactly the same basically.

I cannot figure out what is wrong here, all measurable points seems correct.

Publié : 19/01/2019 8:04 pm
--
 --
(@)
Illustrious Member
Re: Bed leveling issue (right side higher)

@prusite: If the PINDA probe is too high, it will force the nozzle lower and leads to the nozzle catching the paper prematurely.

The PINDA probe is also temperature sensitive. Despite the cal procedure, the longer it soaks, the more its sensitivity changes. Short of a new temp regulated heater on it there will be variation as it moves about. Generally, the variation will be repeatable, but leads to bed calibration error. A weak point in the PINDA design. Using the 7x7 cal firmware is one option, at least it removes the non-linear errors the interpolation in the 3x3 cal add. But for the folks who make the effort to manually dial in the bed flatness, it would be nice to have an option to turn off the bed leveling (easy to remove G80 from the startup code), but add your own cal constants to the 7x7 grid (no G29 options here).

Generally, and especially when using grabby plastics like PET or ABS, a layer difference of 0.05 mm is a don't care; but when layer 1 fill is important, or using PLA where adhesion can be problematic, a "constant and repeatable" layer 1 height is important; and it gets frustrating to have to keep "dialing-in" layer 1 on a weekly basis.

Publié : 20/01/2019 8:31 pm
prusite
(@prusite)
Active Member
Re: Bed leveling issue (right side higher)

Sweet, thanks for the reply.

I also noted (during my 10-ish retries) that the PINDA needs to be lower, it is now way lower than what the instructions indicate it needs to be, otherwise it always catches paper to the right.
This can also cause fast printing and printing leaving threads to collide with the probe, and rip the model off or skew it, this happened once already.

Regarding 3x3 or 7x7 I'm not so sure my issue will be better using this, since the issue is clearly always on the right-most end. This should be visible in the sensor-info on the rightmost row (in theory).

I'm pretty sure the bed is rather flat, there is no light coming through under my L-angle (that is made of metal and veery straight and shiny 😉 ).
As far as I have understood, recalibrating the bed using nylocks is one option, but maybe not in this case. The original poster said he/she tried this, and I'm pretty sure my sensor would also just follow my bed down and give the same error. (since there is as you point out, no off-switch for this).
I clearly see the motor adjusting Z downwards during printing to the right, so I'm fairly sure my issue is software/sensor-based by now.
However, I did actually manually compensate with max -50 and + 50um in the menu/software to make it a tiny bit better.
It is not perfect by any means, but it is actually sticking to the bed using PLA for now at least. I will probably see the errors get worse as I print larger models.

Have any-one else in this thread with similar issues tried the 7x7 software?

I would really like to see the Prusa software-team respond to this, as it seems, very general and wide-spread levelling issue?

Publié : 20/01/2019 9:02 pm
Chocki
(@chocki)
Prominent Member
Re: Bed leveling issue (right side higher)

You should really try the 7x7 mesh bed levelling software, nyloc or spring bed levelling using g80 /g81 pronterface, calibrate the PINDA temperatures and use the modified script which pre heats the PINDA whilst holding nozzle temp below printing temperature, does bed levelling, then brings nozzle up to temp fully, I would also do a filament retract move at the end of the print and maybe before starting the print send the print head up to say 100mm and wait for a button press, this is so that you can clean the print head before starting the print or bed levelling routine..
Yes, it takes longer, but seriously, the difference, I wen't from an almost unusable printer, to perfect prints anywhere on the print bed.

Normal people believe that if it ain’t broke, don’t fix it. Engineers believe that if it ain’t broke, it doesn’t have enough features yet.

Publié : 21/01/2019 10:21 pm
prusite
(@prusite)
Active Member
Re: Bed leveling issue (right side higher)

Sounds good, I will probably try both the 7x7 and the nylock (bought some) later on.
But my first action now (took a few hours) was to remove the bed, the y-axis-rods and everything.
Re-seated the bearings, oiled them, re-did basically all the bed.
Also took off the plastic rod-holders, measured with callipers, filed off some uneven edges on all four, so the rods seated better.
Put everything back together and now the initial measure actually can continue without catching (so much) paper, with the PINDA probe a lot higher.
So, yeah, some effect at least.
Re-calibrated and trying it out now, but it looks a bit better with this standard non-intrusive operation.
One major benefit of this operation was that the PINDA is now at standard-height according to the manual, and it does not hit everything all the time 😉

Publié : 22/01/2019 1:24 am
hetman
(@hetman)
New Member
Re: Bed leveling issue (right side higher)

Hi Guys, sorry for a small off topic.
I have been thinking of buying mk3 kit, but table alignment issue holds me from that. Is there an official Prusa statement for this issue? Or is it that everyone just has to deal with calibration of the axes - not software table leveling, but actual mechanical calibration of the frame, table and overall setup. Does the build design allow you to adjust squareness of axes? I understand from this topic that at least you need to modify table adding jacking screws for leveling/flatness.

Thanks!

Publié : 22/01/2019 11:32 am
prusite
(@prusite)
Active Member
Re: Bed leveling issue (right side higher)

I tried to get prusa support to look at this forum-thread, but they did not have time (at that very moment at least due to lots of calls/chats).
It seems that the levelling issue is a real thing though.
Regarding an official statement the support-personel seem to ignore that this is a real issue, and just keep telling you to redo the steps and retry calibration etc etc.

Update to a theory, this guy has the same theory as I had, that the z-axis-top-stop is not even (even though my measurement in top-position is IDENTICAL).

I will try to actually stop the left z-carry a bit earlier, maybe using some distance/piece of tape maybe, that sets up the actual z-axis sides correctly.
Perhaps the software will then store the both sides' motor-positions differently and maybe that will stop the right side to appear higher up in the virtual space in the software.
The small difference I now have though, would probably be correctable in the menu/manual bed-levelling.

Publié : 22/01/2019 11:59 am
dennis.s20
(@dennis-s20)
Active Member
Re: Bed leveling issue (right side higher)

Just wanted to give an update on my issue
I was also plagued with the leveling issue on my MK3. But interestingly, after I bought the Bondtech MK3 upgrade kit and assembled everything the 7x7 mesh bed level suddenly showed exactly what i was suspecting all the time.
The right site showing was 0.2mm too high and the left side was 0.2mm too low. Now after adjusting the Nylocks I can print awesome first layers and everything sticks wonderfully. So the question is now what does the Bondtech print head do differently than the stock one?

What I believe is that on the upgraded print head the PINDA now sits in a slightly different location and I think it is also slightly rotated conpared to the stock one.
So maybe this has to do with the PINDA after all and for everyone affected it may be worth trying to rotate the PINDA Probe, just as a test.

Publié : 28/01/2019 7:28 pm
J3D Design
(@j3d-design)
New Member
Re: Bed leveling issue (right side higher)

Exactly the same issue here. Printed fine for the first 4 months then after that my right side of the bed seems to be higher, calibration and other fixes doesn't seem to amend this. I'm having to adjust z live for the first layer, bit of a pain!

Publié : 30/01/2019 10:01 am
MichaelK
(@michaelk)
Eminent Member
Re: Bed leveling issue (right side higher)

I've just got a pre-built MK3 and checked the bed level with the Octoprint plugin.

My findings can be found here

https://shop.prusa3d.com/forum/user-mods-octoprint-enclosures-nozzles--f65/octoprint-bed-visualisation-t27454.html

Is this similar to what everyone else is finding ?

Regards
Michael.

Publié : 31/01/2019 7:34 am
3DALHEMIJA
(@3dalhemija)
Estimable Member
Re: Bed leveling issue (right side higher)

I got my MK3 week ago ,and i havent solved this problem jet..
Tried several wizard and regular calibrations, loosing and retightening screws of heatbed etc.
Same live Z adjustment settings given me different results. When i fix one side, another one makes problem... not even to mention that after new firmware, i got problem reading gcodes from SD...

Publié : 31/01/2019 10:07 pm
prusite
(@prusite)
Active Member
Re: Bed leveling issue (right side higher)

Update-time again.
I have had the printer for a while now, doing all sorts of stuff to it, but I ended up double-checking the frame once more.
Tried to loosen all screws, really making sure it was flat doing the re-frame-alignment again, even rotating the alu-extrusions to make sure they were not skewed.
The result, same right side a bit high, no different, veeery not expected, giving that all potential skewed parts now was rotated.
Anywho, filed down the inaccuracies on the rod-holders (as mentioned I think), re-assembled, and using manual bed-level-correct gives an pretty ok level, not perfect, but not that bad.
Have not tried 7x7 3.7.0-beta yet, but I guess the next step for perfection would be nylocks.

Publié : 23/03/2019 8:49 pm
idotf
(@idotf)
New Member
RE: Bed leveling issue (right side higher)

I have had my MINI+ for a few months now, since they new versions started shipping. One of the two minis has just started doing this right side higher thing. I don't know what to do. I originally thought the tube was shrinking and maybe having an extrusion issue. I changed that, made sure the PINDA was good and did the first level calibration. As everyone has described above, if you z the right side to where it sticks, the left side of the bed smashes the filament onto the plate and ruins the quality of the prints on the left side of the board. This is 2021 and I see this thread goes back several years... what the heck are we supposed to do since no one has an answer???

Publié : 29/05/2021 11:47 pm
rhys does 3d printing
(@rhys-does-3d-printing)
Membre
RE: Bed leveling issue (right side higher)

Bruh I just realized I had this issue because one of my z motors wasn't plugged in, I had just switched motherboards and was going through trying everything before I figured this out

Publié : 26/09/2023 9:50 pm
Page 2 / 2
Partager :