RE: Failed Y-calibration of a new CoreOne
I'm not quite sure which material the brackets are made off. The easy bendable leg (at the cutout) may be a hint that it's 'just' cut/stamped from an aluminium sheet. In that case it might be possible to bend the bracket by applying only moderate forces to the x-carriage in an already assembled printer, while the other side of the x-carriage is loose.
Would be a more elegant fix than applying the underlays. Maybe someone brave enough will check on his printer 😀 ...
RE: Failed Y-calibration of a new CoreOne
[... ] it might be possible to bend the bracket by applying only moderate forces to the x-carriage in an already assembled printer, while the other side of the x-carriage is loose. Would be a more elegant fix than applying the underlays. Maybe someone brave enough will check on his printer 😀 ...
I'd say leave it to the kit builders to try this first. I would not want to bend the bracket in a fully assembled printer. While the X gantry should provide a nice lever (when loosened on the opposite end), on the Y side the linear bearings would have to take the full force -- not good.
It's better done during the build, when you can clamp the Y end of the bracket (or the whole bearing block it is already mounted to) in a vise. This might well be what the assemblers at Prusa do -- or are supposed to do...
RE: Failed Y-calibration of a new CoreOne
From what I can see in the pictures, I wouldn't try to use the angles in
bend them when installed.
I would clamp them to the part that is attached to the X axle to bend them. The part with the cutout is too easy to bend. The corner with all the material is much harder to bend.
To be honest, I would have shimmed at a different point. Just above and below the screw. Or better still, for a permanent solution with a hole for the screw.
Can it really be bent so precisely afterwards?
I like the solution of shimming both angles independently to minimize pressure on the bearings much better.
I see this as a sensible tuning. It's not difficult to do and if you have the opportunity why not.
I think a few attempts are necessary to achieve this with bending.
RE:
The part with the cutout is too easy to bend.
As mentioned above, I think this is intentional and the cutout is there exactly to make it easy to bend. In production or when building a kit:
- Attach the bracket to the X gantry. Tighten the double screws, leave the single screw semi-tight to allow lateral slippage.
- Clamp the Y end of the bracket in a vise.
- Bend to exactly 90°, using the X gantry as a lever. Check with an angle gauge as you go. (In production, the vise would probably have a 90° reference arm permanently attached to it.)
- Tighten the three screws, double-check the angle, do fine adjustment if needed.
RE: Failed Y-calibration of a new CoreOne
- Attach the bracket to the X gantry. Tighten the double screws, leave the single screw semi-tight to allow lateral slippage.
- Clamp the Y end of the bracket in a vise.
- Bend to exactly 90°, using the X gantry as a lever. Check with an angle gauge as you go. (In production, the vise would probably have a 90° reference arm permanently attached to it.)
- Tighten the three screws, double-check the angle, do fine adjustment if needed.
The attached X gantry would be in the way when you want to clamp the y end of the bracket fully (up to the bending area) in a vise. But since the mounting screw holes are symmetrical, you yould (temporarily) attach the X gantry to the outward side of the bracket. This way you would be able to clamp the y side of the bracket fully with the vise and use the x gantry as a lever for pulling/pushing the bracket to the needed angle.
Yes, I think this would work well for kits.
RE: Failed Y-calibration of a new CoreOne
But then I can clamp and bend the angle over a large area and don't have to use the X-axis as a lever.
If it's steel, as I suspect, it won't be so easy to bend at the short end. And where the notch is, it should stay straight.
And to get it right, we're talking 0.07mm here, it will take a few tries. Of course, this is easier with the kit, as the portal is not yet installed.
With a bit of luck, we'll be able to see the whole thing live in 5 weeks. Then a lot of things will be self-explanatory.
RE: Failed Y-calibration of a new CoreOne
As luck would have it, a video from the competition was suggested to me on YouTube...
Just in case anyone thinks that such things only happen at Prusa.
RE:
It took me 15 Minutes to get it to a perfect fit. Since bending anything and losing warranty was no option for me, i took what I had in the drawer and clipped off a few cm of 0.2mm metal strip.
I then put it at the opposite side. To get the right side nearer to the front, you have to put the spacer on the left side. Needed two tries to figure that out. These 0.2mm reduced the distance on the right by ~4mm.
After fastening and tuning the belts i clipped of excess material, did my calibration at 85hz on both sides and voila, thats it.
Glad to hear that you cured the misalignment of your printer too!
From your description I'm not sure if you checked both brackets. I think it's important to do so.
A little example:
You detect a 2mm gap on the right side of your printer. You can get rid of the gap by adjusting the angle of the right bracket with underlays. The gap will be ZERO, but the situation may not be good! You may have introduced static load to the bearings of the y axis.
The right bracket may have been perfectly bent to 90°, but the left bracket may be off. With introducing a 'misalignment' to the good right bracket, you can adjust the 'bad' left bracket by pulling it to a 90° angle but you create static load on the bearings of the y axis. This load may cause additional wear or may even affect print quality.
Therefore I think it's important to check (and correct) the angle of BOTH brackets.
Maybe you've already done so but I think it is important for all future adjustment work.
How would I go about checking which bracket is the one that’s actually off? Right now I just have the observation that the right side is not flush, with a gap of 2 mm and change.
Also how are you all getting in there to measure such precise angles and tolerances if that’s what’s involved…?
RE:
Does the Core One have similarly adjustable pulley's for keeping the belts somewhat centered?
This discussion really showcases the unique challenges of a CoreXY gantry design. Even Bambu, with their reputation of solid engineering, doesn't have it worked out either. The comment section in that video includes multiple people who still have VFA issues despite doing the adjustments.
I take this to mean that corexy, with its increased number of pulleys and longer belts, is simply harder to perfectly align than a bed-slinger is. They have their obvious benefits for sure, but there can be a cost exposed in a certain percentage of printers.
-J
As luck would have it, a video from the competition was suggested to me on YouTube...
Just in case anyone thinks that such things only happen at Prusa.
RE: Failed Y-calibration of a new CoreOne
How would I go about checking which bracket is the one that’s actually off? Right now I just have the observation that the right side is not flush, with a gap of 2 mm and change.
Also how are you all getting in there to measure such precise angles and tolerances if that’s what’s involved…?
If you have an already assembled printer, you don't need any measurement tools.
The correction can be done with processing the following steps:
1. Power off the printer
2. Loosen the belts of both sides completely (so they don't create any pull forces to the X-grantry).
3. Loosen the three screws connecting the left bracket to the X-gantry that you can pull away the bracket from the gantry by about 1mm
4. Pull the print head to the middle position (X-axis)
5. Pull the X-gantry to the front side of the printer (min Y-position). Check for any gap on left or right side.
6. If there is no gap on either side, you can continue with step 7. If you detect a gap, you have now to correct the angle of the right bracket with underlays on this bracket:
6.1 loosen the screws on the right side bracket and insert an underlay left or right from the screws (position and required thickness of the shim is depending on the angle adjustment you are after)
6.2 tighten the screws on the right side bracket
6.3 continue with step 5.
7. Tighten the screws on the left side bracket (loosened previously in step 3)
8. Pull the X-gantry to the front side of the printer (min Y-position). Check for any gap on left or right side.
9. If there is no gap on either side, you can continue with step 10. If you detect a gap, you have now to correct the angle of the left bracket:
9.1 loosen the screws on the left side bracket and insert an underlay left or right from the screws (position is depending on the angle adjustment you are after)
9.2 tighten the screws on the left side bracket
9.3 continue with step 8.
10. tighten the tension of both belts to the resonance frequency of 85Hz on both sides
11. pull the X-gantry to the front side one last time. If there is still no gap, the adjustment on your printer is fine!
Or you may wait until Prusa comes up with some official instructions for checking/correcting the angle.
RE: Failed Y-calibration of a new CoreOne
Does the Core One have similarly adjustable pulley's for keeping the belts somewhat centered?
Having watched the Bambu video Rainer had linked to, I would say the "adjustable pulleys" are not a feature but a weakness in that design. The idlers should not tilt off-axis at all. In the Bambu printers, the two idlers which sit on spring-loaded blocks, fixed by a pair of screws, are the price you pay for the convenient semi-automatic belt tensioning. In the Prusa CoreXY design, it looks like the rotation axes of all idlers are well-defined (and I very much hope they are!).
The attempts of the gentleman in the video to use this degree of freedom to center the belts are actually quite ineffective. In the beginning, the belt sits low and moves up to about middle height when he actuates the gantry. In the end, the belt sits in the middle and moves all the way up when he actuates the gantry. Nothing gained; the belt seems to run askew along its overall path.
(I also think that his earlier efforts to de-skew the gantry are misguided. With his two triangle rulers, he measures and compares the distance to the rear end of the printer -- which should be quite irrelevant since the reference position is in the front. Then he attaches a sticky-tape "shim" to the front end stop to "compensate" for whatever? I don't get it.)
RE: Failed Y-calibration of a new CoreOne
As an engineer who works at a precision machine OEM, I was cringing at every point in that video. From using the drawing squares, to using tape for shims and imprecise plastic clips to hold the imprecise plastic drawing squares, Nothing in that video is anything I'd recommend to anyone. Also at least on the XL none of the positions of the idlers can be adjusted since they are on dowels, and the gantry position reference is located off of a machined aluminum extrusion.
I know the point of linking the video was not suggesting any kind of fix for the core one, but I cannot recommend taking much of anything from the video. While the general concepts were good (square gantry etc) the implementation and checking of them is completely bogus and not precise at all.
RE: Failed Y-calibration of a new CoreOne
Just wanted to post, that my printer also had the problem. Upon first installation everything seemed fine until I got the idea to check belt tension...
After adjusting belt Tension to exactly 85Hz on both belts, y calibration failed...
On my printer it was also fixable by adding a shim to the left bracket. I think that this actually is a pretty wide spread issue, that most people just have not found yet.
RE: Failed Y-calibration of a new CoreOne
How would I go about checking which bracket is the one that’s actually off? Right now I just have the observation that the right side is not flush, with a gap of 2 mm and change.
Also how are you all getting in there to measure such precise angles and tolerances if that’s what’s involved…?
If you have an already assembled printer, you don't need any measurement tools.
The correction can be done with processing the following steps:
1. Power off the printer
2. Loosen the belts of both sides completely (so they don't create any pull forces to the X-grantry).
3. Loosen the three screws connecting the left bracket to the X-gantry that you can pull away the bracket from the gantry by about 1mm
4. Pull the print head to the middle position (X-axis)
5. Pull the X-gantry to the front side of the printer (min Y-position). Check for any gap on left or right side.
6. If there is no gap on either side, you can continue with step 7. If you detect a gap, you have now to correct the angle of the right bracket with underlays on this bracket:
6.1 loosen the screws on the right side bracket and insert an underlay left or right from the screws (position and required thickness of the shim is depending on the angle adjustment you are after)
6.2 tighten the screws on the right side bracket
6.3 continue with step 5.
7. Tighten the screws on the left side bracket (loosened previously in step 3)
8. Pull the X-gantry to the front side of the printer (min Y-position). Check for any gap on left or right side.
9. If there is no gap on either side, you can continue with step 10. If you detect a gap, you have now to correct the angle of the left bracket:
9.1 loosen the screws on the left side bracket and insert an underlay left or right from the screws (position is depending on the angle adjustment you are after)
9.2 tighten the screws on the left side bracket
9.3 continue with step 8.
10. tighten the tension of both belts to the resonance frequency of 85Hz on both sides
11. pull the X-gantry to the front side one last time. If there is still no gap, the adjustment on your printer is fine!Or you may wait until Prusa comes up with some official instructions for checking/correcting the angle.
one tool I will be digging up if I need to do the above is a feeler gauge, they come cheap at most diy shops and can be sacrificed for shims.
/Anders
RE: Failed Y-calibration of a new CoreOne
one tool I will be digging up if I need to do the above is a feeler gauge, they come cheap at most diy shops and can be sacrificed for shims.
I'd suggest you look for brass sheet material instead. Much easier to cut and trim than the steel of a feeler gauge. Plus, even if feeler gauges are cheap, it's a sacrilege to cannibalize a good measurement instrument! 😉
Model shops should have brass sheets and foil in a few strengths. The thinner material can be cut with scissors and trimmed cleanly with an Xacto knife.
RE: Failed Y-calibration of a new CoreOne
Very nice video, thank you -- and some good design decisions on display.
I assume that the plastic mounts for the three Z motors are not for acoustic decoupling (they seem rather stiff for that), but to allow some tilt of the lead screws independent from the mounting plane of the motors. Since I have this mild obsession about avoiding over-constrained mechanisms where possible, I am very pleased with that detail! 😎
Does anybody know what the end switch (?) in the CoreXY subassembly is for? I was under the impression that Prusa drives the motors into the mechanical end stops throughout, using the Stallguard feature of the Trinamic motor controllers to detect these home positions.
RE:
Hey All!
Prusa support got back to me, (I had actually shared this thread with them in a note that I sent). This is the procedure that they are recommending:
Thank you for reaching out to us and choosing our products.
It looks like you're experiencing layer shifting in your prints.
To fix this, we first need to ensure that the frequency (tension) of both belts is close to 85 Hz.
Please measure them following the instructions in the next link and share the results with me: https://help.prusa3d.com/es/article/adjusting-belt-tension-core-one_845048
Once you have the results, please send me a photo similar to the following:
To take a photo similar to the previous one, please remove the top acrylic panel as follows:
Remove the four rivets from the top panel.
The previous photo is to determine whether the gantry (X-axis) is properly aligned with the printer frame.
If it is aligned, you will notice that both sides of the X-axis are fully tight against the front end. However, if you find a gap on either side, it indicates that it is not parallel, so you should follow the procedure below to adjust it:
- Loosen the belts almost completely. Carefully, as the square nuts can fall off the plastic parts when removed completely.
- Check which side sits against the front end (tight side) and which one has to be pushed towards it to touch it (loose side)
- Put an uniwrench or something that is at least 5mm thick between the tight side and the front end and slightly push on the loose side to correct the skew. Video showcase
- Check the alignment again and repeat if needed. In case the tight and loose sides switched, use something thinner and try to balance it out. Tighten the belts again
After making the previous adjustments, you can repeat the print that was experiencing layer shifts and let me know if the issue was resolved.
We will continue to follow up on your case until it is fully resolved.
RE: Failed Y-calibration of a new CoreOne
- Check which side sits against the front end (tight side) and which one has to be pushed towards it to touch it (loose side)
- Put an uniwrench or something that is at least 5mm thick between the tight side and the front end and slightly push on the loose side to correct the skew. Video showcase
- Check the alignment again and repeat if needed. In case the tight and loose sides switched, use something thinner and try to balance it out. Tighten the belts again
Now that's interesting, thank you for sharing! So the support team has moved away from the "set uneven belt tension to mask the skew" advice, which is good. But I am not sure I like their new advice, which seems to be "bend the gantry brackets into shape right there, in the assembled printer".
The bending action shown in the video does not look like it will change anything -- looks to me like they are staying well within the elastic regime. And I don't like the idea of putting that bending force onto the linear bearing that rides on the Y guide rod, using a pretty long lever which acts on a short bearing. But maybe those 90° brackets are indeed from soft aluminium, and there is a force range where one can bend them slightly (and permanently) without overloading the linear bearings?
RE:
But maybe those 90° brackets are indeed from soft aluminium
They are magnetic, so some kind of steel.
I have the same problem with the gantry not touching. In my case the gap (~2-3mm) is on the right side when standing in front of the printer. I decided to follow the prusa suggestion from above. But when it came to actually bending the bracket, I did not dare to push hard enough to cause plastic deformation in the bracket. Instead, I looked for a shim and found a 0,6mm wire. After several trial and error clamping attempts I have gotten it close with this arrangement: 
My first attempt at printing with the wire in place is running. Previously, the print always made it past the head-banging stage. Sometimes it took only 10s, sometimes several minutes.