V4 500 Only Corexy for now.
Is there any guidelines to tune properly the belts?
I've tried more tension, less tension and I can't get a second spike.
86 Replies
Removed the rubbing on the pulleys but nothing changed...
Properly tuned to 90hz ish
125hz peak is the toolhead
But still no first spike
it's corexy right? you don't measure X or Y - you measure 1,1 and 1,-1, looking at your graphs. you measure only one belt isolated - in that mode you won't get that second spike. that second spike corresponds to Y.
I'm using the thing from ratos
Recommended by the commissioning guide
Moved the pulley a bit
Same result
okay right yea i was dumb, with x or y you should have a single peak, with 1,1 or 1,-1 both... hm.. but yeah that would mean you still have some binding somewhere if there's only one peak. can you feel if your y-rails are maybe binding?
In the middle I do have a change of resistance
Which wasn't there before
hm yeah that's bad then, your axes should move smoothly in every direction, no binding/skipping or something. maybe your gantry is racked or your y-rails not properly aligned.
I deracked it before the belts
Was perfectly fine
I'll derack it again
It's slightly skewed now when I bring it to the front now
that doesn't matter - as soon as you insert the belts and tension them, you introduce twist again.
well there's your problem then 😄 if you de-rack it by compensating via belt tension i guess your graph will look better. just don't overdo it, better have a bit too less tension then a bit too much.
could damage your idlers, motors etc.
If that's just that
I'm gonna get really mad at myself
happens. and from what i've read - 70 Hz in those graphs for a 500 is rather on the high side. so maybe you already over-tensioned it a bit. i'd start with as low tension as possible. CoreXY is pretty straight forward in that regard: get your tension on both belts in the rough ballpark so that your gantry is straight and your belts are not loose, but also not too tight, then tune it with IS. your graphs at the beginning will look terrible, but if you add tension bit by bit (say 1/8 rotation for each tensioner screw for both belts after every try) it will get better. if it gets worse again - back off, then you enter overtension-territory.
I've read in #v-core-4 that I should look for about 87 on a 500
I've already built a 2.4, so it's not really a corexy discovery, but the ratrig belts systems is much less forgiving I think
Looks like I found myself a first spike
A small one but if it's over tighten..
it's def too tight, look at this: https://discord.com/channels/582187371529764864/1288589347641102438
Okay
these big machines have waaaay longer belt paths than any other standard printer out there, so you need way less tension to make everything run smooth. too much tension only causes binding and stuff.
Do you have a distance and hz I should aim for?
I mean I could fit my 350 in there 😅
hm i'm not a 500 owner, i have a 300 IDEX (whole other horror story 😂 ) but i'd say aim for around 50-60 Hz ish for the main peak and around 25-35 for the second?
Oh yea
I've dipped the upgrade to idex
For. Now.
But when I see the horror stories to tune it uh
https://discord.com/channels/582187371529764864/1296131270920900720 - have fun :kekw:
those beasts you can only tune with IS graphs. your belt tension graphs are useless there because you need the hybrid belts.
I need to tune hybrid after that
Don't worry I'll ping you
:kekw:
:kekw: we'll see if i can help you... i'm for the most part clueless myself, even after watching the famous "How to read and analyze input shaper graphs generated by klipper"-video from a few years ago for the 4th or 5th time now (https://www.youtube.com/watch?v=M-yc_XM8sP4) - guess the IDEX will properly teach me how to do it.
3D Printers & a Whiteboard
YouTube
How to Read and Analyze Input Shaper Graphs Generated by Klipper
This video is supposed to be a brief overview on what the input shaper graphs can tell us. It is the first video of a series of videos which will give a more in depth view on the kinematic system and input shaping in Klipper.
As always: feel free to jump to those parts, that help you with your problems:
00:00 Intro
00:57 What we can learn from...
I've checked tg73 pdf again
If I read it correctly it recommends 90Hz for the tuning
Which would mean I'm not even tight enough
i don't know this pdf - but 80-90 Hz is generally the recomendation for the smaller machines (300 corexy)
ahhhhh okay 😄 i see your error
that's NOT referring to your graphs - that means literally plucking the belts at the given lengths and checking the frequency on a spectrometer (smartphone-app for example).
Yes
I'm at 87hz
So
Not tight enough in theory right?
no idea. i usually don't tune via "plucking", i only do it roughly and then do the finetuning via tension graphs and IS
i don't think that you can get accurate results with it, noticed this myself when i tried - depending on how you pluck the belt, you can get totally different frequencies out of it. not really reliable, i'd say.
ðŸ«
It almost looks like it's more tight
Half a turn less tension
Same thing
I'm at like
75Hz
Which is almost too loose
I'm really tempted to just roll with it
what does a proper input shaper run say? i mean that's what's important at the end.
but looking at the graph i think you still have binding somewhere. maybe an idler not running properly or something like that.
if you give it more tension and that spike disappears again, then there's definitely something wrong with your mechanics
I have an over 100 x noise value
what range? 1e4? higher? lower? paste the graph 😄
That's odd
If I cut the motors
Aka M84
Noise is fine
hmm... that's indeed very high. maybe you have a wiring issue then.
Redid a homing
No issues now
Weird
hm.. intermittend connection maybe? you use a toolboard? EBB42 or something?
Yup
Ebb42 from the kit
The Y looks like dogshit
connected to a 3.0 usb-port or 2.0 (black ones) on the pi? i read somewhere that beacon and toolhead-ports should be connected to 3.0 ports
and yeah your Y is defi binding - power spectral density in 1e3 range means binding most of the time
The X looks fine tho
Urgh
I guess I'll need to check tomorrow for that Y binding..
good look, hope you find something
I hope too
Plan was to print a cube tonight
I guess not
:kekw: and here i am fiddling with my IDEX since two weeks after i finished the mechanical build
For the binding is there a procedure
not really... that's what i'm also fighting with. i have a veery veeeery small range of tension where my power spectral density for X goes to 1e4, everything above and below that and it's crap again. i can't pinpoint if there is binding, if any, i don't "feel" it when i move the printheads around.
what i will try tomorrow (if it's not better now with loose hybrid belts) is maybe loosen all the rails (y, maybe even x) and realign them, then not over-tighten them as to prevent binding of the rail carriages. if that doesn't work and it is the belt paths - no idea how to fix that. my pulleys are all moving, every belt is somewhat evenly tracking (left x slightly goes down on the left side, but only with the left toolhead completely left, bumping the endstop and the gantry fully to the front - tried to push it up a bit, doesn't help).
Looks like we are onto something
@thepete89 IS graphs now or?
It's better but still not too good
from the tension graphs it would seem like you have something loose somewhere. but your IS graph looks fine to me. high, defined peak, 1e4 and 3,9k accel with ZV shaper. i'd try it. i don't know how easy or hard it is to get over 5k with a 500 - maybe could use a bit more belt tension, but not much.
I need to fix the loose stuff on the toolhead yes
I guess I can also cut the excess belts now right?
absolutely :kekw: the more that's flapping around in the breeze during the test, the worse your result.
But like yea
The fact that ratrig doesn't provide
Like goal for optimal IS acceleration doesn't help to find if my result are good or not
hm, in the idex commissioning guide there is a section that states that there will be an extra in-depth guide on belt tensioning, IS and so on sometime in the future. but no idea when that will be available. generally speaking, you can't give universal guidelines and stuff because every printer is different. and the general knowledge what to look for is already out there.
Sure but like
They could give a ballpark
Of like
"a standard built 500 should get about XX on X and XX on Y"
that's the thing 😄 with kits there is no such thing as "standard built" - not even with pre-assembled printers. you have manufacturing tolerances. heck, even tiny tolerances in your freaking bearings could make the difference between "perfect graph" and "oh crap what is this?". i mean, look alone at the millions and millions of variants you could route your toolhead cables. one guy slaps zipties everywhere and calls it a day, another one might just tie the cables down here and there, another one might completely ignore loose cables and zipties in the first place and someone really really really obsessed with this might just slap cable chains on there. you see where i'm going with this?
Yea
I guess I'll settle for that for now
At least i have pretty clean IS graph compared to my last big printer which was a resonating mess
But right now I'm stuck with a little dog on my feet
:ratrig_heart:
Where in the guide do they indicate how much belt I need to keep?
for XY? none, i believe. everything should be cut flush once it's installed.
Rofl
Okay
and the hybrid-kit gives you an extra loop for the hybrid-belts - that in it self would be far too short for corexy.
Got it
I've tied down everything but that spike at 125hz won't go away..
hm, have heard that from a few people now that there's something around 125 Hz - many speculate that it's related to the umbilical flexing or swinging.
I have the same peak at 130 Hz. If I use the realtime analysis and push the toolboard/ oribter to one side, the amplitude goes down. But I guess some play in the carriage is normal?! I have tied down every cable and checked all srecws that are reachable....so no glue where this comes from
With hybrids belts
nice
I can move on right?
yeah this looks fine
Hi Roguyt
My initial graph looked the same yours did, what did you do / adjust to achieve this result?