New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
For a few weeks now Prusa has had the new belt tuning frequencies on their web-app belt tuner. They also updated the Core-One-Belt-Tuning-Help-page to reflect the new changes. For the new tuning setup the printhead has to be in the front right corner and then then the two belts running along the gantry need to be plucked in the center of the linear rail. The top belt should be at 98 Hz and the bottom one at 92 Hz.
But the Android / iOS apps are still not updated and have the old settings (85/85). The old tuning method already had a bunch of discussion around it, especially how to properly do it since the official instructions were kind of lacking, so I thought I might make a thread about the new method since I've been struggling with it and I cant imagine being the only one. Another problem was gantry alignment, and how far one could deviate from the perfect frequencies to achieve alignment.
So, now I would like to ask you: What is your experience using these new belt settings? Did you manage to properly tune it? Did VFAs go away if you previously had them? How is the pre-programmed input shaper supposed to know that you've changed the belt tension? Did the resonances change? Should we completely loosen our 85Hz belts and then start fresh, or is it OK to start from the old optimum setting and then tune it up from there?
Here is my own opinion:
The new official manual says to first measure top and bottom belt frequencies, then adjust like this: "Adjust the belts by tightening or loosening the belt tensioner screws by the same amount. A tighter bolt causes a higher frequency, and vice versa. "
This sentence is a bit ambiguous IMO, but I interpret it this way: When one belts tension is too low, make that one higher while lowering the other belts tension by turning the corresponding idler screws by the same amount. Using this method you basically end up tuning the right idler further and further down to achieve the 92Hz, and the right one further and further up to hit 98 Hz. But of course, having one idler more tightened than the other one leads to a gantry gap on the right.
From my experience, the gantry gap works like this: Loosening the idler on one side increases the gap on that side, for example on the right. If you then tighten the idler on the opposite side (left), that will also increases the gap on the right. And that is basically what we're supposed to do with these new instructions. We are tuning one side further and further up, pushing the gap to the other side, and then tuning the side with the gap even further down, which increases the gap even more. I genuinely do not understand how Prusa has made this work in their tests, because I cannot for my life figure out how this is supposed to work. I tried this maybe 5 times, not once was I able to get rid of the gap.
Any ideas? Maybe my kit was fucked up from the beginning, maybe Im interpreting something incredibly wrong.
Just to prove my findings ive made a short video. Ive also confirmed that the measured frequencies are indeed correct by using three different device's microphones: An iPhone 12, a Galaxy S10e, and an old Logitech Headset hooked up to my PC. All the frequency readings are within 1 Hz of each other
IMG_1625
RE: New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
Yes, you're right. You ARE interpreting something incredibly wrongly 🙂
The instruction merely says to turn them by the same amount. It doesn't say to turn them in opposite directions.
The different frequencies on the top and bottom belts actually represent the SAME tension - the frequencies are different due to the different lengths between the print head and the corresponding idlers.
Start with zero tension, check that the gantry is square, then apply gradual and even tension to both sides. Don't let one side get too far ahead of the other because that WILL open up a gap - hence the 'same amount' advice. Aim to start building the 6Hz difference between top and bottom from fairly early on.
You'll find chapter and verse in another thread here - this is a good place to start.
You'll find advice for squaring the gantry with zero tension in that thread, as well advice to lubricate the tension screws before you start re-tensioning.
RE: New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
The frustrating bit for me was that the initial instructions (not the revised ones) were not at all clear. Yes, it was very clear where to move the gantry and position the extruder assembly, and that initially the gantry had to be square. What messed a lot of people up (me included) is the original instructions never said which set of belts to pluck (this should be clearly mentioned in the web app as well but it's not). The end result were people following their interpretation of the instructions and then skewing the gantry so badly that it had to be manually bent back into shape.
Further to that, you can't build up a 6hz difference in some setups without turning one side in the opposite direction. But it has to be done in fractional turns like 1/8 of a rotation. Prusa really need to provide a video demonstrating this, especially if English (or the other published languages) are not your primary one.
My opinion on this is that if you are not experiencing serious VFAs (which I wasn't before) then stick with the "old" method. Getting an accurate frequency reading on the lower belt across the gantry/linear rail is hard and the belt-tuning web app even with an iPhone is tedious to almost impossible. In my case, it made absolutely no difference in print quality.
RE:
I'm past the edit time period but in the above when I refer to "initial instruction" and "original instructions" I mean the initial version of the revised (or new) method for tuning the belts. The "old" (85hz) ones were very clear.
RE: New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
I can't keep the gantry align with 98/92. I have to go with 97/93 to keep the gantry align.
RE: New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
Aha! That makes more sense! Didnt think that the belt length difference would explain the frequencies. I checked, and luckily it seems like I didnt bend my gantry.
I've tried the technique you linked and described, to no avail. I tried every conceivable method I could find, in total I spent maybe 4 hours tensioning, measuring, un-tensioning, taking apart the Nextruder to check the belt teeth, re-tensioning, measuring, failing. I am so done.
Tightening the gantry evenly does indeed keep it square, but for me the printer always naturally builds a 3Hz difference in belt tension. Even when I take insane caution to tighten it evenly, I end up with a 3 Hz difference. So I can get 98/95 or 97/94, but anything beyond that causes a gantry gap, no matter what technique I try. 98/95 in the new position is equal to 100/103 when I move the head to the old position. No matter if I use the old or new technique, the 3Hz gap is consistent, but the more-tight belt switches from top to bottom.
What do you mean by "Aim to start building the 6Hz difference" - If the machine works perfectly that should happen automatically, right? For me this is simply not possible
Im pretty sure the belt path is correct, number of teeth in the nextruder holder is equal for both belts. Dunno what else to check. Here are some pictures in case I overlooked something:
Pulleys left and right:
Belt paths:
Belt teeth inside the Nextruder:
Im coming around to the idea that there is something inherently wrong with my machine, because none of this makes sense to me.
There is also this weird phenomenon on my machine: When the gantry is perfectly square at the front, it will not be square when pushed to the back. It will hit the left rear, but not the right rear. Could this be related to my problems? Prusa support was very unhelpful regarding this issue, telling me to disassemble the rear motor mounts and re-assembling them. But this of course didnt change anything as the rear mounts are well constrained and have holes to interlock with. I sent them a video of the issue, here it is:
This annoys me so incredibly much, I did anticipate some tinkering when buying a Prusa printer, but I thought I would just be.... done at some point. Instead this has been dragging on for 2 months now.
RE:
Ah, yes, that's odd, I agree. I wonder if the holes for either one of the motor mounts or one of the tensioner mounting blocks have been drilled slightly inaccurately. Given that you can't get the correct belt tension without opening a gap at the front, I wonder if the correct belt tension would actually close that gap at the rear-right. If so, that could suggest that it's one of the front tension mounting blocks that's not quite in the right place. You could double check the bolts on those are secure, but from memory each one has a locating tab so it's hard to see how the fasteners could be at fault.
Alternatively, is it possible that the mating face of either one of the gantry mounts or the motor mounts or the tensioner mounts has something protruding that shouldn't be there? It doesn't look like it in the video, but there isn't really a long enough clear shot of all the mating faces.
RE:
I just tried re-tensioning to 98/92, and the gap this causes is quite significant:
When I push the 98/92 gantry backwards the gap on the right is now closed and instead appears on the left. So this is yet another unrelated issue, unfortunately.
Im just gonna try using 98/95, as this keeps the gantry square. Gonna re-calibrate the input shaper and phase stepping and i'll see what happens.
The motor mounts have locating tabs, last night I also pulled the right side panel off so I could check the screws of the mount, but they are all tightened and I even when I do loosen them the possible play is very small.
I also took some photos of the mating surfaces:
Im thinking about re-ordering all the 3d-printed parts of the corexy system and replacing them, in case they are somehow at fault. Wanted to print them myself but Prusament PC-CF is out of stock! Ugh...
If that does not help, im not sure what else there is to try. The only other component involved would be the baseplate, but that would be the nuclear option...
Im not sure if the drilled holes in the baseplate could even be at fault, as its most likely stamped. If it was CNCed, maybe the machine could have skipped a few steps and then drilled a hole wrong, but I seriously doubt that they would CNC a flat piece of metal in this quantity.
RE: New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
That's puzzling. As far as I can tell, everything looks the same as my printer, but you have what appears to be a 2mm gap front-right.
I've just used a tape measure to measure the distance between the mating surfaces front to back. It's not an accurate measurement because the straight line distance has the gantry in the way, but with the gantry roughly in the middle so that the tape measure only needs a gentle curve to pass over it, I measured 277mm on both sides. If you have 2mm gap I think that should be visible in those measurements if it's down to one of the blocks or the baseplate being out of spec.
RE: New belt tuning frequencies (98 / 92 Hz) - Discussion (is it even possible to align the gantry with these settings?)
My tape-measure measurements are roughly the same, but the gap at the rear right (which is always there, independent of front gantry square-ness) is very small. I dont have feeler gauges handy right now, but its in that scale of distance.
Puzzled and angried by this, I've spent a lot of time this weekend basically disassembling the entire CoreXY motion system, down to removing the belts and comparing the # of teeth on both of them.
My main findings are, that the CoreXY motion system is aparrently VERY VERY sensitive to any of the four corner pulley blocks being placed slightly differently, down to the tolerance of the holes for the screws that hold them in place. From my findings and experiments, this is the main cause of not being able to reach the given belt tensions, which seems to be a problem a bunch of ppl have.
Initial setup and issue: With the gantry square at the front, pushing it to the back created a gap at the right motor mount. To even achieve a square gantry at the front, I needed to overtighten the right belt and it was impossible to achieve Prusa's given frequencies (85/85 or the new 98/92), because this would misalign the gantry so it would have a gap at the right when parked at the front.
First adjustment: To try to close the rear right gap, I loosened both belts completely and made sure the gantry was square at the front with no belt tension. Then I used this "truely square" gantry to align both rear motor mounts and tried to get them to be as perfectly aligned to the gantry's square-ness as possible. I was indeed able to close the rear right gap mostly, but I instead caused it on the left by being a bit uncareful while tightening both blocks. The left rear gap was smaller than the old one on the right, and when I then tightened the belts with this setup, the left belt was the one that needed to be over-tightened to close the front gantry gap, but the amount of over-tightening required was less.
Second adjustment: I then, very carefully re-aligned both rear motor mounts so the rear gap would be gone. I then reassembled everything, and now I can get to 99/92 without misalinging the gantry in the front! I can still feel a smudge of a gap at the rear left motor mount, but im not gonna pull off the side panels again just so I can get at the screws of the motor mounts. 98/92 seems to be elusive, sometimes I can get there but somehow the belt tension always seems to fluctuate a tiny bit.
So, my findings would suggest that:
1. If you cannot reach Prusa's belt frequencie without opening up a front gap, you’ll need to overtighten one belt to compensate.
2. The side that needs overtightening will correspond to the side with a rear gantry gap when the gantry is pushed back against the motor mounts.
3. Inversely, if meeting Prusa’s belt numbers causes a front gap, that gap will be on the same side as the misaligned rear motor pulley block.
This assumes both belts are the same length (equal # of teeth at each end inside the nextruder holder) and the tightening is done as reommended: Slowly, in small steps, and equally for both sides.
Now I think I need to confirm/deny these findings. Anyone else with similar issues mind checking for rear gap once the gantry is aligned at the front?
Idk if my findings are new, but this ceartainly surpised me. I didnt think that the tolerance of the screw holes would have such a big impact on the correct belt tension. Maybe thats the solution to all the tension problems we're seeing, as its certainly esier to misalign the motor mounts than to align them: Blindly, without knowing that such a thing could be important.
I've also noticed that right after a hot chamber print (Polycarbonate for example) the belts have a higher tension. I just re-checked this: Room temperature :99/92, after PC print: 104/96. Now, after the print the tension is suddenly 97/90, so I guess there is something entirely different afoot, AGAIN!