RE:
I apologize for my contribution to the belt tensioner rabbit trail. 😆 I agree with both of you. I think it was an important thing to discuss in this thread because of the extensive tension adjustments needed for testing...AND think that now that it has been covered it's time to return to the main topic.
I'll need to do some actual testing, but after running a 2 hour print last night with the new RH motor pulley, and only very slightly increasing the tension from around 87 db to 90 db using the old method, my first impression is that the surface quality appears to be improved. I'm tempted to swap out the LH motor pulley as well to continue the test.
I don't know if it has been mentioned previously here, but the mellow pulleys that I received appear to have a different tooth profile compared to the Prusa ones. They are flat on top rather than rounded like the Prusa ones. It would be interesting to confirm this with others who have swapped to mellow pulleys. This means that there is more than just machining precision that separates them from the original ones.
-J
RE: VFA Artifacts on X+Y Straight Edges
I found the right thread to spam with the tensioner discussion!
RE: VFA Artifacts on X+Y Straight Edges
They are flat on top rather than rounded like the Prusa ones.
I noticed that the bike repairman that lives next to me has a mini lathe and a mini mill. I'll see if I can get him to knock a bit off the outside of the stock pulleys so that there's less opportunity for the belt teeth to catch on the pulley profile. It will decrease the maximum force the belt can handle but I think there plenty of headroom.
I'm getting a bit lazy when it comes to making these elastomer pulleys, I find that I have almost no noticeable VFA as it's settled in. I'll get them done at some stage since I suspect the carbon fibre filament will be releasing fibres as they rub which is quite unhealthy.
RE:
They are flat on top rather than rounded like the Prusa ones.
I noticed that the bike repairman that lives next to me has a mini lathe and a mini mill. I'll see if I can get him to knock a bit off the outside of the stock pulleys so that there's less opportunity for the belt teeth to catch on the pulley profile. It will decrease the maximum force the belt can handle but I think there plenty of headroom.
I'm getting a bit lazy when it comes to making these elastomer pulleys, I find that I have almost no noticeable VFA as it's settled in. I'll get them done at some stage since I suspect the carbon fibre filament will be releasing fibres as they rub which is quite unhealthy.
Oddly enough when I put the calipers on them, they don't seem to measure differently than the rounded top ones. I'll have to do some calibration cubes or something to make sure printing dimensions haven't been altered by using the mellow aftermarket pulleys. 🤔
-J
RE:
On this topic. Wouldn't t it be great if Mellow would be willing to create GT1.5 pulleys that have the exact same effective diameter as the GT2 pulleys, allowing the conversion to GT1.5 belts without requiring a Prusa firmware accommodation.
I guess we need to wait to see what Prusa ultimately decides on for a solution. Really, this is something they should commission and start building printers with. This would maintain compatibility with firmware, open an anffordable upgrade path, and allow new printers to go out the door with significantly reduced belt ripple.
-J
RE: VFA Artifacts on X+Y Straight Edges
What dictates how far the belt moves per revolution of the pulley is not the diameter, but the belt pitch and the number of teeth per revolution. Given that the spacing of the pulley grooves has to match the spacing of the belt teeth for proper meshing, the diameter is a by-product of those two values. The 16T GT2 moves the belt 32mm per revolution, whereas the 21T GT1.5 moves it 31.5mm, hence the 1.5% difference in scale. That's the closest you can get - 22T GT1.5 would move the belt 33mm, which is a bigger discrepancy.
RE: VFA Artifacts on X+Y Straight Edges
I have the belts and pulleys to move to GT1.5 but printing the VFA test tower doesn't show me enough VFAs to be worth the work. It's not VFA free, as I can see artifacts with shiny filament and strong lighting, but it's just not as bad as I had thought it was going to be. I'm going to wait on Prusa.
RE: VFA Artifacts on X+Y Straight Edges
What dictates how far the belt moves per revolution of the pulley is not the diameter, but the belt pitch and the number of teeth per revolution. Given that the spacing of the pulley grooves has to match the spacing of the belt teeth for proper meshing, the diameter is a by-product of those two values. The 16T GT2 moves the belt 32mm per revolution, whereas the 21T GT1.5 moves it 31.5mm, hence the 1.5% difference in scale. That's the closest you can get - 22T GT1.5 would move the belt 33mm, which is a bigger discrepancy.
Of course. 😆 I should have given the thought my full attention before posting.
RE: VFA Artifacts on X+Y Straight Edges
Diameter is directly related to number of teeth, and a bigger diameter will indeed move it further, but you don't get to pick the perfect diameter; that is constrained by pitch times tooth count, which I think is what you're saying here. There is no way to make it that tiny bit bigger.
What dictates how far the belt moves per revolution of the pulley is not the diameter, but the belt pitch and the number of teeth per revolution. Given that the spacing of the pulley grooves has to match the spacing of the belt teeth for proper meshing, the diameter is a by-product of those two values. The 16T GT2 moves the belt 32mm per revolution, whereas the 21T GT1.5 moves it 31.5mm, hence the 1.5% difference in scale. That's the closest you can get - 22T GT1.5 would move the belt 33mm, which is a bigger discrepancy.
RE: VFA Artifacts on X+Y Straight Edges
I've added my two cents there.
I've now destroyed at least six pulleys trying to make my C1 work.
The only reason of this is that it failed to home after maybe 20h of print time, and I was trying to retune the belts. Seized nut and the rest you know by heart.
I want to make this part from steel, as it should be made from the beginning. And 12.9 screw so it wouldn't break easily.
RE:
Six? Crikey - you must be pretty good at replacing them by now. What's your method for removing the seized nut and bolt from the tensioner mounting block? Some kind of cutting tool, like a Dremel?
Are you going to try a brass insert instead of the square nut?
Maybe take this discussion over to this thread?
EDIT: Sorry, you've already done so.
RE:
Sorry guys, I sincerely mean no disrespect to anyone, but I think I'm not alone when I say this thread has become one humongous TLDR for all of us who can't engage more actively, but just want to see the problem solved or at least be included in an outline of the progress. Is there a chance that someone could be making a weekly excerpt of the findings and a short SITREP every once-in-a-while in a separate thread?
Thanks again to all of you investing your valuable time an knowledge in finding a solution to this pestering problem!
RE: VFA Artifacts on X+Y Straight Edges
Checking back in, not had the time to strip down my printer and change anything recently. But yesterday I had some time and I decided to change the idlers which the toothed part of the belt run over to toothed idlers.
This is the only change I made to the motion system, I did have a hardened nozzle in but I doubt it would make much difference. Recent tests are using the updated belt tension of 98/92Hz.
Result: VFA are improved, but only slightly. More improvement came at the higher speeds which VFA for me are mostly gone in all directions. Lower speeds are improved but still visible in all directions.
I have replacement drive pulleys to test next.
I'm starting to believe one of my motors may be unhealthy or not as good as the other, moves involving both motors seemingly have more noise, produce more VFAs especially in -X and +Y directions.
I thought that phase stepping would quieten this down which when it was going through it's motions made the Y motor very quiet but the improvement figures I get are x:87%/y:23%.
RE: VFA Artifacts on X+Y Straight Edges
What I mean by allowing the inner races to move is it should have some lateral, not rotational, motion. Tension changes due to oscillations should cause the inner race to move away from the front of the printer, that is along the y axis of the printer. You want to allow that sort of movement with a bit of friction.
My silicone hose has an internal diameter of 4mm, so I had to sleeve the M3 spindle bolt to keep the hose centred. And it's such a thin walled sleeve that it wasn't strong. My choices, to eliminate the sleeve, were to adapt the pulley holder to use a 4mm spindle, or just to buy some 3mmx8mm silicone hose. So I've ordered the new hose. Then I'll cut an under-length piece for the bearings to run on - it'll likely be a tight fit, so the hose won't want to turn on the spindle bolt, and the inner bearing races won't want to turn on the hose. I'll then experiment with flanges that have a variable thickness shoulder against just the inner races, to provide a little y-axis friction as you describe. The outer races will be free to rotate without friction, but hopefully the friction on the inner races may take the energy out of the belt oscillations to prevent resonance, somewhat like the rubbing of the pulley holder on the pulley mounting block in your system.
However, since my printed flanges will tend to rotate due to the belt riding against them, the shoulder section will be rotating against the inner races. Hopefully there'll be enough slip between the silicone hose and either the spindle bolt or the inner races to allow this to happen, otherwise the (fixed) inner races will rub on the rotating shoulders and wear them away rather faster than the small y-axis movements would do. Only one way to find out.
It'll be an interesting experiment, but I don't see it as a permanent fix, because the friction surfaces will wear and become less effective over time. But it'll help us identify the root of the problem.
RE: VFA Artifacts on X+Y Straight Edges
Sorry guys, I sincerely mean no disrespect to anyone, but I think I'm not alone when I say this thread has become one humongous TLDR for all of us who can't engage more actively, but just want to see the problem solved or at least be included in an outline of the progress. Is there a chance that someone could be making a weekly excerpt of the findings and a short SITREP every once-in-a-while in a separate thread?
Thanks again to all of you investing your valuable time an knowledge in finding a solution to this pestering problem!
It's an understandable request -- I also wished for a nice summary along the lines of "what has worked, and what percentage o VFA reduction does it provide?" Unfortunately life is not that easy...
People have tried different things: Optimize belt tension; optimize belt routing (centered, away from the idler flanges); use toothed idlers where the toothed belt side runs over them; use motor pulleys from different brands; switch to belts and pulleys with 1.5 mm belt pitch; calibrate the phase stepping. And, as far as I have gathered -- invariably some people claim a big improvement, some a little improvement, and some see no significant change or the ripple even get worse.
That is unfortunately in line with the experience on other printers, if you search the internet for "belt ripple" and VFAs on 3D printers in general. So the jury is still out... I am really curious what Prusa's solution (?) will look like eventually.
RE: VFA Artifacts on X+Y Straight Edges
However, since my printed flanges will tend to rotate due to the belt riding against them, the shoulder section will be rotating against the inner races. Hopefully there'll be enough slip between the silicone hose and either the spindle bolt or the inner races to allow this to happen, otherwise the (fixed) inner races will rub on the rotating shoulders and wear them away rather faster than the small y-axis movements would do. Only one way to find out.
I had another thought. Maybe it's better to stick with the original pulleys, but adapt the mount so that the spindle bolt is 'bushed' with a thin slice of the silicone hose. The friction control could then be via, say, an oversize washer under the bolt head (and maybe a nut and washer on the other end) - where there's no rotation.
RE: VFA Artifacts on X+Y Straight Edges
Shouldn't the entire tensioner be redesigned?
The belt should be dampened axially in relation to the belt, i.e. in the direction of tension.
If the holder is divided, with a 90-degree bend at the ends. These interlock and their pressure surface is dampened. In principle, like two hooks.
The bearing should be mounted in a long hole so that it can move in the direction of tension.
I hope you understand what I mean.
The tensioner and bearing are separated by a layer of softer material.
I don't have access to my Core One at the moment and can't estimate whether there is enough space for it.
RE:
I may well have overlooked something in the thread. But do we have any evidence that damping on the tensioners (in an effort to damp belt vibrations) can work at all to reduce belt ripple? The only experiment I recall is what Chris did here, and that did not produce convincing results yet. Were there other tests with more promising outcomes?
RE: VFA Artifacts on X+Y Straight Edges
Shouldn't the entire tensioner be redesigned?
The belt should be dampened axially in relation to the belt, i.e. in the direction of tension.
If the holder is divided, with a 90-degree bend at the ends. These interlock and their pressure surface is dampened. In principle, like two hooks.
The bearing should be mounted in a long hole so that it can move in the direction of tension.
I hope you understand what I mean.
The tensioner and bearing are separated by a layer of softer material.
I don't have access to my Core One at the moment and can't estimate whether there is enough space for it.
I think the issue is that the energy that causes the resonance needs to be absorbed, e.g. through friction. I already introduced a silicone layer, and the conclusion was that it acts more as a spring than as a damper - it merely introduced another resonance mode. @malpan surmised that his original mod was successful not solely because of the silicone that he introduced, but because the idler carrier was rubbing against the idler mount and the friction was dissipating the energy.
I've mocked up my idea. The orange parts would be silicone bushes, sliced from a length of the 3mmx8mm hose, and then an oversize washer under the bolt head and another under the nut will provide adjustable friction.
RE: VFA Artifacts on X+Y Straight Edges
I may well have overlooked something in the thread. But do we have any evidence that damping on the tensioners (in an effort to damp belt vibrations) can work at all to reduce belt ripple? The only experiment I recall is what Chris did here, and that did not produce convincing results yet. Were there other tests with more promising outcomes?
Yes - @malpan's original test here: https://forum.prusa3d.com/forum/postid/757715/