Belt tensioning issues - firmware tensioning guide and website tensioning "app"
Hello,
I have a prusa core one built from kit.
I've never really managed to get exactly the tension in the belts suggested by prusa. I want to ask a few clarifying questions before i sink a few more hours in it tonight.
The current status of my belts is that, according to the website tuning app my top belt is at 99 Hz and my lower belt is at 90 Hz. The website app is satisfied with that ("just right"), this comes after fiddling with it for 3 hours last night, completely detaching the tensioners by mistake and so on.
Before doing this adjustment, I was at the exact same values - i.e. I only managed to do the best I managed to do when originally assembling the printer; I've always tried to get exactly where prusa says I should be (max 98 hz on the top belt and not more than 8hz difference between the belts) but I cannot get closer than 9Hz between the belts ; also my left tensioner is much tighter than my right one, if that matters.
Now that we have official firmware 6.4.0, I tried the manual belt tuning option in the firmware. It's hard for me to understand what exactly the firmware tuner wants from me or deduce from the GIFs on prusa's website how the belts should move but logic (it's very possible I am wrong, I am wrong all the time) tells me that in the tuner I should adjust the frequency until:
1. I get the belts to do a slow back and forth, one at a time
2. the correct value for the frequency I should stop at is the one where the belt has the highest amplitude or highest range of movement.
Is that correct?
If I do this, the frequencies where I get the largest amplitude in movement for the belt are about 2 Hz above what the web app tells me. At that point the tuner in the firmware cannot help me and tells me to go square the gantry or something ; my gantry is squared so maybe the only thing to do is take everything apart and make sure again I have the proper amount of teeth pulled in through the bracket behind the nextruder.
ALTERNATIVELY I can just leave it like it is and move one with my life.
Are people out there printing with slightly out of range frequencies in their belts?
Thank you
RE:
Your understanding of what to look for during the stroboscopic adjustment (your steps 1 & 2) is correct. In case you have not found it already, here is Prusa's support page describing the process. (And also describing gantry alignment and belt re-attachment if it should ever be needed.)
The overall logic is that the gantry should first be squared without any pull from the belts. Then the belts get tensioned symmetrically. So the belts must not be used to force an otherwise skewed gantry into alignment.
It seems that the belt adjustment wizard "gives up" when it detects that the two belt tensions differ too much from each other. It then concludes that you must have used unequal belt tension to force the gantry into alignment, hence suggests that you should start over with the "Quick Gantry Alignment" as described in the yellow box on the support page above.
In your case, the deviation from the ideal belt resonance frequencies is pretty small. If you experience reliable homing of the tool head (without excessive "head banging"), I would probably leave things as they are for now -- unless you are curious and just want to walk through the whole process outlined on the support page.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
Thank you Jürgen, helpful as always.
That's good to know.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
To add, it's seems to be a (the most) common issue for people to try to force the belts to be 98/92hz. I did this at first and it caused all sorts of issues by skewing the gantry followed by failing Y calibration, layer shifts and homing problems.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
I must say I also found the gifs on that page not very helpful. I wish they had some real videos where they show you how the resonances change over the course of tuning, not just the end result.
Formerly known on this forum as @fuchsr -- https://foxrun3d.com/
RE:
@hyiger thank you for the reply. I also feel like I just made things worse haha. When I assembled the core one I had *no problems* at all with endless homing and banging of the printhead around. It did it twice or four times and that was that. Now I don't know which way is up anymore. But with your reply and the others in the thread, I decided to just take it easier, find something that looks like it works and repair things only when I have print issues.
@foxrun3D , indeed the GIFs help a bit at the beginning and then just leave you (me) puzzled. Thanks for answering, this community helps me relax a lot about the problems (or lack of problems) of my printer.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
I have the exact same "issue". My upper belt is at 98Hz and my lower belt at 89Hz. No matter how precise I try to fine tune every bit, these 9Hz difference stays the same. I also don't experience any issues and that's why I leave it now. The new 6.4.0 method doesn't work for me, because the gap is too big and it wants me to fix my perfectly aligned frame...
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
I have the exact same "issue". My upper belt is at 98Hz and my lower belt at 89Hz. No matter how precise I try to fine tune every bit, these 9Hz difference stays the same. I also don't experience any issues and that's why I leave it now. The new 6.4.0 method doesn't work for me, because the gap is too big and it wants me to fix my perfectly aligned frame...
This would come down to how many teeth from each belt was inserted into the wiring harness behind the Nextruder. Both belts should have the same number of teeth sticking out.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
I have the exact same "issue". My upper belt is at 98Hz and my lower belt at 89Hz. No matter how precise I try to fine tune every bit, these 9Hz difference stays the same. I also don't experience any issues and that's why I leave it now. The new 6.4.0 method doesn't work for me, because the gap is too big and it wants me to fix my perfectly aligned frame...
This would come down to how many teeth from each belt was inserted into the wiring harness behind the Nextruder. Both belts should have the same number of teeth sticking out.
I can't see how the number of clamped teeth would be critical in that respect. If one belt has a bit more or less slack than the other, you can always compensate with the belt tensioners (within reason). I think the "clamp 4 to 5 teeth" is just a guideline: You don't want to clamp only 1 or 2, since that would not hold the belt securely. And you don't want 10 teeth clamped on each side, because then you could not extend your belt tensioners far enough to even get hold of the belt.
If the belt tensions differ too much, I would rather suspect that the printer started with a skewed gantry, and different belt tensions were used to compensate for that. So it would be a matter of relaxing the belts (via the tensioners), checking the gantry and bending it if necessary. See here, yellow box "Quick Gantry Alignment". Much quicker that taking off the Nextruder to check or reattach the belts!
RE:
The new 6.4.0 method doesn't work for me, because the gap is too big and it wants me to fix my perfectly aligned frame...
By "gap" you mean the difference between the two belt tensions, right?
But how do you know your frame is perfectly aligned? Did you check it without any tension on the belts? If the gantry touches the two ends stops in sync at the moment, with the unbalanced belt tensions, that would rather suggest that it's the belts which pull it into alignment. So you would need to fully relax the belts, possibly bend the gantry to get it aligned, then re-tension the belts symmetrically.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
I have the exact same "issue". My upper belt is at 98Hz and my lower belt at 89Hz. No matter how precise I try to fine tune every bit, these 9Hz difference stays the same. I also don't experience any issues and that's why I leave it now. The new 6.4.0 method doesn't work for me, because the gap is too big and it wants me to fix my perfectly aligned frame...
This would come down to how many teeth from each belt was inserted into the wiring harness behind the Nextruder. Both belts should have the same number of teeth sticking out.
I can't see how the number of clamped teeth would be critical in that respect. If one belt has a bit more or less slack than the other, you can always compensate with the belt tensioners (within reason). I think the "clamp 4 to 5 teeth" is just a guideline: You don't want to clamp only 1 or 2, since that would not hold the belt securely. And you don't want 10 teeth clamped on each side, because then you could not extend your belt tensioners far enough to even get hold of the belt.
If the belt tensions differ too much, I would rather suspect that the printer started with a skewed gantry, and different belt tensions were used to compensate for that. So it would be a matter of relaxing the belts (via the tensioners), checking the gantry and bending it if necessary. See here, yellow box "Quick Gantry Alignment". Much quicker that taking off the Nextruder to check or reattach the belts!
This sounds too familiar for my situation. I can't tell if the gantry is not straight with the belts loosened but I am 100% sure that (in my case), after I achieve my 98-89 Hz result, the left tensioner is much (MUCH) tighter than the right one.
I had tried the "Quick gantry alignment" 3 times the night before I made this post but I had two problems:
1. I cannot see if the gantry is skewed on the left or right side ; it looks ok to me. Only now I realise I can probably stop trying to visually get it and just measure the distance on the linear rods. I will probably try that again this week.
2. "3. Grip the gantry firmly on both ends and apply a gentle counter-force in the opposite direction of the tilt until it straightens." I had no idea what this means. But now if I can identify if the thing is crooked by measurement along the linear rods (unless someone has a better idea) , I can just apply forces in all directions until the measurements are fine.
If I manage something, I will report back.
RE:
1. I cannot see if the gantry is skewed on the left or right side ; it looks ok to me. Only now I realise I can probably stop trying to visually get it and just measure the distance on the linear rods. I will probably try that again this week.
The idea is that -- with both belts loose -- you just gently push the whole gantry towards the front. If both sides touch the end stop at once, and stay in touch with the end stops if you stop pushing, it is straight. If you see a small air gap on one side (looking down from the top), the gantry needs to be straightened.
If you remove the top lid of the enclosure, you get a clear view of the end stops and the gap. If you are unsure, you can also slightly push against the gantry on either side when it is supposed to be touching the end stop. If it still wiggles, it's not touching the stop properly.
(To correct the alignment, you don't bend the X rail itself of course, but the soft-metal mounting brackets which attach the X rail to the Y bearings.)
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
Posted by: @jurgen-7
I can't see how the number of clamped teeth would be critical in that respect. If one belt has a bit more or less slack than the other, you can always compensate with the belt tensioners (within reason). I think the "clamp 4 to 5 teeth" is just a guideline: You don't want to clamp only 1 or 2, since that would not hold the belt securely. And you don't want 10 teeth clamped on each side, because then you could not extend your belt tensioners far enough to even get hold of the belt.
For a given belt tension, the frequency is dependent upon the "free length" of the belt. If the belts are installed incorrectly (i.e. a different number of teeth engaged in the nextruder) then the belt tensioners will necessarily be in different relative positions to the gantry pulleys (closer or further) resulting in a different "free length" to be plucked and a different frequency for the "same" tension. The priniciple is the same as any stringed instrument. The frequency to which a string is tuned is dependent upon the length of the string between the nut and bridge and the tension on the string. Tune a guitar string to an E note when open you can then, for a given string tension, create different notes (frequencies) by pressing your finger against the string and varying the "free length" of the string.
Regards,
Steve
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
But with the current tensioning instructions, you no longer pluck the belts on the sides, adjancent to the tensioners. You pluck them at the X gantry (or, with firmware 6.4.0, you let the motors excite them and watch their oscillation, but still at the X gantry). There the free length is defined by the distance between the tool head and the left-hand idlers, and remains the same independent of the tensioner position.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
2. "3. Grip the gantry firmly on both ends and apply a gentle counter-force in the opposite direction of the tilt until it straightens." I had no idea what this means. But now if I can identify if the thing is crooked by measurement along the linear rods (unless someone has a better idea) , I can just apply forces in all directions until the measurements are fine.
Let me try to explain.
Printer off (or motors off). Slide the gantry towards the door.
Observe which end of the gantry hits the stop first. Let's say that's the left side. What you do is to move the gantry back, then place something like a screwdriver handle (or perhaps the multi-wrench) between the stop and the gantry on the left.
Pull the gantry towards the door, pulling on the right hand side. You'll want to put sufficient tension on that side (with the handle keeping the left side spaced) so the gantry is pulled so it's square. Pull a bit, remove the handle and try again. ALWAYS do this with the belts loose.
You could also pull with your right hand on the right side of the gantry and push with your left. Either way, you're just trying to get that gantry square so when you pull it forward you hit the stops at the same time on both sides.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
I've pondered this a bit and offer the following-
1) It's essential that the X gantry be aligned, IOW hitting the stops at the same time with the belts loose.
2) The gantry alignment is a sensitive indicator of when the belt tensions are equal.
3) If the gantry is not aligned when you tighten the belts, you must adjust one and only one tensioner until the gantry is again aligned.
4) At that point you can adjust both tensioners by the same amount to raise or lower the belt tensions per the recommendations of the resonance utility.
5) You cannot independently adjust both resonant frequencies- they change as a pair. You are in control of alignment and one resonant frequency. The other frequency follows wherever it must, if alignment is to be maintained.
RE:
I've pondered this a bit and offer the following-
1) It's essential that the X gantry be aligned, IOW hitting the stops at the same time with the belts loose.
2) The gantry alignment is a sensitive indicator of when the belt tensions are equal.
3) If the gantry is not aligned when you tighten the belts, you must adjust one and only one tensioner until the gantry is again aligned.
I disagree with #3. (And, as of the latest alignment instructions, Prusa seem to disagree as well.)
You need to first make sure that the gantry is aligned, while the belts are fully loose and do not affect its position. This is done by actually bending the (soft) brackets which hold the X gantry to the Y bearing blocks -- see "Quick Gantry Alignment" in the yellow box on the support page above. Once the gantry is straight, you begin to tension the belts, and you always do that symmetrically.
This will result in the same tension on both belts. (Which corresponds to slightly different observed frequencies, due to the different free belt lengths between print head and idlers during the "tuning" process.)
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
Since I can no longer edit my post above:
I had actually overlooked the "with the belts loose" condition in your item #1. So we are not that far apart -- you don't recommend to deliberately adjust the belts to different tensions in order to straighten the gantry (as I had misinterpreted it), but to use the gantry alignment as an indicator for correct adjustment to equal tension. That may work as well -- I have never done it that way, and found it easier to just turn the tensioners by equal amounts once they have taken the slack out of the belts.
For what it's worth: My Core One (a very early kit build) has never had problems with unstable homing or extended head-banging. I had straightened the gantry at the time of installing it, and that seems to have done the trick, throughout the various iterations of belt tensioning recipes provided by Prusa. The VFAs (belt ripple) have however not improved significantly, even with the most recent belt tuning wizard and balanced speed profiles in PrusaSlicer.
RE: Belt tensioning issues - firmware tensioning guide and website tensioning "app"
Yes, I think we're actually on the same page. What I wrote was a quick summary of something longer that you can read here- https://conradhoffman.com/CoreOne_assy.htm The belt stuff is near the end. Prusa was also reading it, and I was hoping to hear their comments, but nothing yet. I too have slight VFAs and have never used the balanced profiles. I'm going to try that today just to see what happens. This page from Josef got me intrigued- https://blog.prusa3d.com/new-in-prusaslicer-consistent-surface-finish-and-nerfing-vfas_120400/ There's just so much information it's hard to absorb everything!
RE:
Hello again,
I had some time today and thought I would try everyone's recommendations. tl;dr; I couldn't make it work, my left belt tensioner now is spinning along with the nut that its threaded through and I need more repairs.
So first of all thank you to eveyrone who pitched in more ideas, Conrad, Jürgen, k1mu and Steve.
First I tried again prusa's recommendation:
- I loosened the belts completely and I can see the gantry is slightly not square , having about 2milimeters distance from the stop on the left side (my left, the printer's right).
- No matter how much I tried to push the gantry in the opposite direction, to make it straight, that gap remained
I then tried everything I could from the tips above, including visiting Conrad's page (nice write-up) the best result I could achieve was ... exactly what I had had before:
- 98 Hz top, 89 Hz bottom according to the prusa web app that uses the phone's microphone
- 101.5 Hz top, 89.5 Hz bottom according to the manual belt tunning
This is the place where i gave up last time because really, it's probably good enough.
This time I had ambition so I loosened the belts fully again and was probably not careful enough, now the left tensioning screw is spinning in place.
I have to do more repairs now so I will not be able to come back very soon to update on the belt tensions. When I have a new result, I will definitely complain/celebrate here.
See you soon, I hope.