RE:
For info for all Someone further up the tread had tried lengthening the belts by moving un tooth at the rear of the head assembly.
On initial assembly I was getting the Y calibration fail problem, The brackets and X axis are good and square so no issue there
The pulleys of the tensioners were protruding from the holder by about 1-2mm I didn't think this was a problem as the X axis was cleanly hitting the tensioner block
Extending the cable by a bit ( Its a interesting job getting to the 4 screws that allow the belts to be adjusted!) , probably 1 tooth , has got it so that now the pulleys are sitting fully inside the tensioner block as well at 85hz.
Calibrated perfectly ! this may pe useful for others
RE: Failed Y-calibration of a new CoreOne
My solution for the Y-calibration fail
I spend hours searching for the problem, adjusting belts, modifing the gantry with all the tips I got here.
But then I removed the right side cover plate and Y-calibration worked. The problem was that the cable for the filament sensor was slightly jaming the X-moving bar. I pushed the cable further trough the hole into the back.
Maybe it helps for some
RE: Failed Y-calibration of a new CoreOne
Hello!I ordered my Core One conversion kit on the first day, it was delivered relatively quickly, sometime in early May.I built the new printer, it passed all the tests at first, at first there was a small hammering in the right front corner, I adjusted the belt settings a little, the hammering stopped.Printing started fine for two weeks, then it thought for a moment and started hammering the right front corner again for 2-3 minutes. Re-tensioning the belts, running tests, recalibrating, the hammering stopped.Everything was fine again for two weeks, now it started again, I tried everything, 5-6 hours of trying with support, no results, it refuses to start properly, it writes different error messages every time it starts. I'm fed up with it, I've never had such a crappy printer before. I want to send it back to Prusa, they haven't been able to tell me how it's going for three days, they don't know how much money they're willing to refund, and the best part is that they're not willing to take back either the camera or the accelerometer, along with the four printer plates. So I have a room decoration, a useless printer. It's all a joke, I'm thinking of suing them in a European Union court.I've read similar posts on several forums, some are already working on forming a lawsuit.
RE: Failed Y-calibration of a new CoreOne
[...] now it started again, I tried everything, 5-6 hours of trying with support, no results, it refuses to start properly, it writes different error messages every time it starts. I'm fed up with it, I've never had such a crappy printer before. I want to send it back to Prusa, they haven't been able to tell me how it's going for three days, they don't know how much money they're willing to refund, and the best part is that they're not willing to take back either the camera or the accelerometer, along with the four printer plates. So I have a room decoration, a useless printer. It's all a joke, I'm thinking of suing them in a European Union court.I've read similar posts on several forums, some are already working on forming a lawsuit.
It seems to me that one of the more effective ways to get Prusa's attention is to make noise on Reddit ( https://www.reddit.com/r/prusa3d/). They don't seem to give a $#!+ about problem reports here in the forum, no matter how many users describe the same issue. But do seem to be concerned about their reputation on Reddit.
RE: Failed Y-calibration of a new CoreOne
Spent the best part of 3 days building the kit version. Calibration tests failed at the Y Axis test. The left side of the rail was around 4mm shy of the stop when the right first touched. Taking the top off the case and loosening the three screws holding the rail to the side bars showed the right side rail bracket was not at 90 degrees. A couple of washers behind the single screw fixed the alignment (and the screw head just misses the cooling fan). All tests now passed. Not ideal but it avoided a full tear down of the nozzle assembly to resolve the bracket.
Yesterday we've received our long awaited CoreOne printer.
It was a batch 2 printer and the assembly of the display was no problem (maybe the display cable was already a little bit longer than in first batch printers).
After linking the printer to our network the self calibration started. What should have been a routine task (especially for a pre assembled printer), became a big problem.
The calibration of the Y-axis failed with a red ERROR message:
PRECISE REFINEMENT FAILED
Consider adjusting belt tension.
More detail at prusa.io/31102
The given link 'prusa.io/31102' leads to nirvana;)...
I've ajusting the belt tension to exactly 85Hz but this didn't change anything.
I've recognized that on Y-movements the carriage touches the stop on the left side about 2-3mm before the stop on the right side would be touched. -> there is a gap of 2-3mm on the right hand side. The opposite gap shows up when the carriage is moved to the maximum Y-position.
Can't be sure if this gap is intented but it look strange to me. It also results in a reduction of the Y-travel by 2-3mm, possibly causing the calibration failure.
I've contacted prusa support (by eMail) 18 hours ago, but no reaction so far.
Maybe someone can confirm that the Y-axis should reach the hard stops on both sides simultaneously.
Thanks, Martin
RE:
Before the build, I borrowed a machinist square and straightened the loose brackets in hopes of avoiding delays during assembly.
Fast forward to assembly, and when I finally attached the X rail/guide I could visually tell that it was still crooked. Straightening the loose bracket is no substitute for aligning them when mounted to the rail. I proceeded to mount each side individually, using the rail to gently bend each brackets to square perfectly with the XY frame, then mounted both and proceeded with the build. When it came to belt installation and tensioning the job was pretty easy. Little by little alternating sides to slowly bring up the tension and things stayed straight. Calibration passed first try and I've been good to go since then. 🙂 Sometimes it bumps more times than others, but never for long before getting on with it.
-J
RE: Failed Y-calibration of a new CoreOne
> Before the build, I borrowed a machinist square and straightened the loose brackets I hopes of avoiding delays during assembly.
I wasn't even thinking about this as I built my printer, but I discovered it was easy enough to fix after full assembly. Unlike aluminum, these steel brackets have plenty of flexibility and will tolerate adjustment. If they were made of aluminum, this business of bending them to align the frame would be risky, because aluminum doesn't tolerate a lot of flexing.
What you describe is consistent with my experience. But I want to add that belt tensioning and XY alignment should be managed separately. The XY alignment is best carried out after the belts have been tuned. What I mean is that adjusting the XY frame's orthogonality shouldn't change the belt tension, but the reverse isn't necessarily true.
RE: Failed Y-calibration of a new CoreOne
> Taking the top off the case and loosening the three screws holding the rail to the side bars showed the right side rail bracket was not at 90 degrees. A couple of washers behind the single screw fixed the alignment (and the screw head just misses the cooling fan).
IMHO this approach isn't really necessary. Your washers accommodate a bracket that's not quite square, but simply bending the bracket is much easier, and these steel brackets can be safely bent (a reasonable number of times). Chances are this will only need to be done once. And if this were not the case, if repeated alignments were needed, imagine having a special set of washers ready to manage any future issues. Imagine the effort required to maintain the alignment using washers of varying thicknesses.
And consider this -- if the frame isn't square, if your Y axis test fails, this means the brackets have already been bent out of square, have already experienced some stress from an unknown source. You won't be subjecting them to a new experience.
I've been reading about this online, and it turns out that Prusa doesn't speak with a single voice on this issue. In one post, the author suggests what I do -- gently and carefully bend the frame back to true. In another, the author recommends packing up the printer and shipping it back to the factory.
But for the latter remedy, one can't be sure the frame won't get re-bent during return shipping, at which point an insightful person might say, "Okay, now I get it." Also, I strongly suspect the official factory cure for this class of misalignment involves ... wait for it ... bending a bracket.
When I fixed my printer's XY misalignment, which was minor but sufficient to fail the Y axis test, I only had to apply fingertip force -- nothing to write home about. And no need for washer-shopping. 🙂
RE: Failed Y-calibration of a new CoreOne
@martinf did you ever get a resolution to this issue? I have just completed the MK4s to Core 1 upgrade and have run into the Y-calibration issue. I have built every printer I've owned, including 4 Prusa's, over my 15+ years of printing and this is the first time I've been foiled, so any insight to a solution would be appreciated. Needless to say, I have followed all the Prusa published information.
RE: Failed Y-calibration of a new CoreOne
@martinf did you ever get a resolution to this issue? I have just completed the MK4s to Core 1 upgrade and have run into the Y-calibration issue. I have built every printer I've owned, including 4 Prusa's, over my 15+ years of printing and this is the first time I've been foiled, so any insight to a solution would be appreciated. Needless to say, I have followed all the Prusa published information.
Are your motor pulleys installed in the correct way (the orientation is different for the X and Y motors)?
Did you leave the same amount of teeth for both ends of the belts and for both belts when assembling the extruder carriage?
Is your gantry square?
Are your belts properly tuned (after the gantry is square)?
Does the right side of the gantry hit the end stops first? (Try tightening the rights side just a littttle bit more then the left side)
RE:
> I have just completed the MK4s to Core 1 upgrade and have run into the Y-calibration issue. I have built every printer I've owned, including 4 Prusa's, over my 15+ years of printing and this is the first time I've been foiled, so any insight to a solution would be appreciated. Needless to say, I have followed all the Prusa published information.
This issue is covered in my Prusa Core One Web page, it's an easy fix.
Here's the link to the page: https://arachnoid.com/3D_Printing_Prusa_Core_One/index.html
Here's the specific section that discusses this issue: https://arachnoid.com/3D_Printing_Prusa_Core_One/index.html#Core_XY_Axis_Alignment
Basically, the Y axis test usually fails because the free Y travel distance isn't >= 220mm. This failure, in turn, is most often caused by the X and Y axes not being square with respect to each other. And this, in turn, is easily fixed by gently, carefully bending the metal brackets that enforce the 90 degree reference between the X and Y axes, as explained in my Web page. The page shows how to do this, and is pretty detailed.
Also, do not try to fix this by changing belt tensions, as is sometimes suggested -- this only compounds the difficulties. This printer really needs to have both belt loops tuned to 85 Hz, equal to each other to the degree practical. The Y test failure issue is better resolved as explained above, a method that doesn't require adjusting the belts.
After you successful apply this fix, please post again so others can hear how things turned out.
RE:
Did you leave the same amount of teeth for both ends of the belts and for both belts when assembling the extruder carriage?
This exactly is what solved my Y-calibration issues (i've had one tooth difference). I did wrote about it a couple of times, only to get flushed by answers that it doesn't matter, that's supposedly why tension part is for. But, as said, that solved my problems. Yes, i did have to disassemble extruder again, which was P.I.T.A, but it worked (and i still claim that belts SHOULD be of the same length up to the last tooth).
RE: Failed Y-calibration of a new CoreOne
Does the right side of the gantry hit the end stops first? (Try tightening the rights side just a littttle bit more then the left side)
All your other points have been found by many others to help with the head-banging and calibration. But I don't recall seeing this one mentioned as a requirement (or a way to make the XY calibration more robust). Where did this come up, and how much evidence is there?
@mike01hu -- you did not mention where and how your Y calibration fails. In case it's not about the head-banging in the front right corner, but about a failure of the calibration wizard: Please check also that the Y linear bearings can travel freely, and specifically that the cable to the side filament sensor (inside the right side panel) does not interfere with the Y bearing there.
RE: Failed Y-calibration of a new CoreOne
@martinf did you ever get a resolution to this issue? I have just completed the MK4s to Core 1 upgrade and have run into the Y-calibration issue. I have built every printer I've owned, including 4 Prusa's, over my 15+ years of printing and this is the first time I've been foiled, so any insight to a solution would be appreciated. Needless to say, I have followed all the Prusa published information.
My 'solution' to this issue are the washers applied to both brackets. The printer works relatively fine now.
But if I would have had the knowledge that is available today, I would have just bent the steel brackets to the correct alignment. Each bracket on it's own (with the opposite bracket loosen up). In my experience, you only have to apply this correction once. In the end result, both solutions should work the same.
One of the reasons I've bought 6 printers from Prusa was my trust in their support service. But this first time I've ever needed Prusa support, they failed miserably.
RE: Failed Y-calibration of a new CoreOne
Does the right side of the gantry hit the end stops first? (Try tightening the rights side just a littttle bit more then the left side)
All your other points have been found by many others to help with the head-banging and calibration. But I don't recall seeing this one mentioned as a requirement (or a way to make the XY calibration more robust). Where did this come up, and how much evidence is there?
@mike01hu -- you did not mention where and how your Y calibration fails. In case it's not about the head-banging in the front right corner, but about a failure of the calibration wizard: Please check also that the Y linear bearings can travel freely, and specifically that the cable to the side filament sensor (inside the right side panel) does not interfere with the Y bearing there.
I don't recall where I have read about tweaking the left and right tension to make it juuuust uneven. It's somewhere on this forum and it was specifically someone who did all of the other steps perfectly. Squared gantry, perfectly tuned but still issues with banging. Someone suggested that maybe if both end stops don't hit at exactly the same time (or the "wrong" side hits first), it might improve homing. And that worked by tightening the right side (I think) just a bit.
It might have something to do with the fact that the Core One does not use end switches but instead relies on a motor stall. The travel in the Y direction requires two motors to be active and the motion has two possible end stops (left and right side). I assume that the machine uses one of the two motors for the stall detection and because of the belt layout, there is most likely a difference in stall behavior for each side of the end stop, for a specific motor. So maybe if end stop with the longest belt path to a specific motor is hit first or simultaneous with the shorter belt path end stop, the stalled signal is less clear, requiring more bumps. Just my hypothesis...
You bring up other good points:
Try homing from the menu instead of during the print. There are some cases where homing fails or takes more time when the heater is active.
Try loosening the y axis bearing mounts, move the Y axis back and forth with the motors turned off. Then, gradually retighten them. This makes sure the bearings are nicely aligned with the rods, possibly having a positive effect on homing.
RE: Failed Y-calibration of a new CoreOne
I think it was either me or I was one of them, suggesting to make the right side belt a tiny bit tighter.
I had both sides hitting exactly in-sync (to my eyes and fingers) and the belts tuned by ear, after Prusa Tuner & Spectroid showed identical Hz.
Ran the Auto Home without heating and the endless banging started. While it was banging, I very slowly tightened the right tension screw and at some point, the banging stopped. (and didn't come back yet).
Spectroid showed a steady Hz for the left and 0-1 Hz higher for the right belt.
It's not about the right end-stop being hit before the left stop, but rather about the left stop definitely not hitting first, but both still visibly in-sync.
RE: Failed Y-calibration of a new CoreOne
I have resolved my issue by rethinking the gantry assembly process. It is critical that the gantry is exactly square to the Y-axis, but the assembly instructions provide no guidance on how to check or measure this need. Indeed, the mechanical design of the connection to the Y-axis fails to take this into account and tries to achieve this with two bent sheet-metal brackets. Having completed the Core-1 conversion and listened to the poor thing bashing its brains out whilst trying to achieve Y-calibration, I went back to the drawing board, figuratively speaking, and decided that during assembly critical measurements of the gantry geometry had to be made. Fortunately, this can be achieved post build with careful bending of the gantry brackets, so to do this follow this procedure:
- Switch off the Core-1 or disable the drives
- Slacken all belts
- Move the gantry fully forward
- Centre the extruder
- Whilst gently holding the gantry forward, but NOT forcing it to take up the uneven slack, measure the distance from the inside front edge of the drive plate to the face of the linear bearing at both ends of the bearing, in my case it was around 60mm.
- Determine which side was closest because this is the side that will need pushing back in the next procedure
- Move the gantry to the middle and whilst firmly holding the furthest side push back gently on what was the closest side. TAKE GREAT CARE WITH THIS.
- Move the gantry forward and measure both sides again. There must be no difference.
- If there is a difference, repeat the procedure.
- Readjust the belt tension and check the Y-calibration
- Print a 200mm square that lays down a simple 1mm wide perimeter and measure across the corners whilst still on the plate. It should measure 282.8 across the corners in both directions, and if so, you have succeeded!
This worked for me, but Prusa needs to adjust the design to ensure squareness and adjust the assembly instructions to address this shortfall.
.
.
.
How does one post pictures in a post???
RE: Failed Y-calibration of a new CoreOne
Thanks, Martin. You will see I have resolved the issue for me.
RE: Failed Y-calibration of a new CoreOne
Same here, hat to carefully bend the brackets straight.
This helped me: https://www.printables.com/model/1143218-vernier-skew-test/files
RE: Failed Y-calibration of a new CoreOne
Thanks a lot for putting this together, much appreciated.
In your depiction of the print skew, it is depicted as a parallelogram. In my case however, I end up with a gap on the left side of the X-carriage both on the front and the end of the printer (Y-axis max and min), so this would rather result in a trapezoid depiction of the print skew. Am I looking at the same issue?