VC4 500 IDEX - 125hz Peak
Man I am so close to finally having Y serviceable, anyone have any idea what the 125hz hump is?
114 Replies
I've had a couple people ask me how I got such a clean Y spike on IDEX, so adding the info here:
NOTE: At these tensions, despite producing a really nice graph, I did have a belt keeper give way. Keeping these notes here for discussion or perhaps find a better way to tension these cleanly.
I think the issue is the toolhead belts and hybrid belts need to be exactly right from one another tensionwise, else it's like a couple people playing tug of war and pulling the rope at the wrong time (which in this case, would translate to X vibrations in the Y movement). Here's what I did:
1. Get your hybrid belts to low-mid tension level and equal to one another,
2. Home your printer, so that the T0 toolhead is sitting middle of the bed like it would be in a shaper test,
3. That 40hz peak is the peak you want to drive way up while getting rid of the others (at least it was in mine in the screenshot at the start of the thread, adjust based on your machine's response). In realtime analyzer, set it to toolhead 0 and oscillate Y axis at that frequency. You'll likely see a lot of Y and also X activity in it,
4. Start slowly tightening your toolhead belts, a couple degrees at a time, evenly. Basically just turn one slightly tighter, then the other. You'll hopefully start seeing the X activity decreasing in the realtime graph, and Y activity growing. You may have to do this a couple dozen times before you eventually hit a point where the X activity falls completely off,
5. Turn off your motors. This is critical, because the motors will hold belt tension unevenly until they are powered off.
6. Re-home the printer,
7. Run the same realtime oscillation of T0 on Y axis at 40hz (or wherever your peak is), you'll notice some of your X activity has returned. This is because you turned off the motors and the toolhead belts were able to balance.
8. Begin tightening the toolhead belts slowly again, until the X activity goes away. It should take a lot less turns this time.
9. Keep repeating this process until you are able to turn motors off, home, and start an oscillation with no X activity left.
After you reach that point, do an input shaper test again and see how you fare. Hopefully you'll have one big Y spike and a little 125hz bump like I do above.
Also doing this got my toolhead belts pretty damn tight. Dunno if I am too tight or not, but the graphs look great so I am rolling with it lol.
Hopefully someone finds the above useful, and maybe someone will help me figure out this 125hz peak.
Thanks for the guide on how you tuned your belts i will try that now. My graphs are not that clean but im in the same boat as you with the 125hz peak. I basically rebuilt the printer 3 times by now and still didnt find the issue. 🫠
Hope it helps, no idea if it's the "right" way or if it's too tight on the toolhead belts, but the graph looked so good I decided to risk it 🙂
Idk if there is a right way but your way got you a better result for sure 🤣 i will try that
But im actually going insane with that 125hz peak 🫠🫠
Can you make a video of how belts look after that? Like is it super tight or loose?
Ratrig should release some tension tool data on initial setpoints or good start tensions to allow us to have a decent shot
Yes I wish RatRig would. I show my tensions and such, but I worry other people will blindly follow the video. I don't know what I am doing, so everything you see here should be taken with a grain of salt. I don't want people snapping or bending things lol
Yeah unless you really fucked up during your build all the printers should within a certain range i think
But I am pretty happy to post this. 5,000 Accel perims at 250mm/s. 🙂
Need a bit more PA tuning but man I ain't complaining at that speed
that sounds really usefull. i'll try that - also struggeling with belt tensions and IS graphs in general.
@TheBrigandier one question (maybe dumb one): if you solely tune based on Y T0 - how does this affect Y T1 and X in general? or are they basically all aligned to each other after you did the x-belt-tuning with Y T0?
That's a good question, you guys may have to experiment and let me know your findings there. Basically I hit a point where I was like "I must get Y better, the others will just have to fall where they will". There's so many variables that I just decided to solve the worst issue.
That said, I did a full shaper test after. My X0 and X1 both looked great and I had a ZV (not MZV) recommendation on those. At work at the moment and can't post them, but it seems correcting Y in this way got X looking pretty good as well in my case lol
It's so difficult to get everything balanced I can't imagine going much further. Anything else you touch causes the X/Y balance to get really wacky fast
One area I would like to play around with is the hybrid tension. You notice I say start at low-mid tension, this was just where I felt they were tight but not too tight. There may be a better place for those to be, but i'm not sure how to determine it without a lot of trial/error.
yeah i feel that, i'm trying to figure this stuff out since 2 or almost 3 months now - got some current IS graphs from my last try here: https://discord.com/channels/582187371529764864/937654723727745046/1331378082371862548 - upping the x-belt tensions helped (after much hardware-fixing. motor plate alignment and stuff) and also @Wetson and others determined that a low to mid y/hybrid-belt-tension seems to help. while others only had success with high belt tensions in general. my guess is - it depends on the machine and especially machine-size. so far for my 300 idex lower y-belt-tension in comparison to the x-belts seems to work better than equal tension on all 4 belts.
I would say machine size is going to make a really, really big difference.
Those long belt spans on a 500 get floppy in some frequencies
This was another reason I was hesitant to post a vid plucking my belts, someone will run with that without fully checking
yeah could be the reason why higher tensions in general work better for the bigger machines - more tension = stiffer belts.
Hands down the biggest breakthrough from those steps is turning the motors off and repeating the tuning process. If a good friend of mine hadn't brought that up, I would still be tuning rather than printing
plucking is absolutely not recommended for idex. maybe just to get the belts in a rough ballpark, but you almost always need measurement equipment. belter, west3d-tension-tool etc.
and yeah turning the motors off and then retry makes sense - lets everything settle in position.
but honestly i'm so frustrated with my machine right now - i have a feeling i'm so close to getting usable graphs but just can't figure out those last few gremlins 🤣 while other just do the ol' "send it" and get great results
Yeah especially on X. If you think about it, starting at the tensioner, the side going straight to the motor is much shorter than the side of the loop going to the toolhead. So when you tighten the tensioner, the short side of the loop gets much tighter than the toolhead side. You'd think it would adjust out through the toolhead path of the belt, but it doesn't lol
I toyed with it probably 20 hours in shaper tests/realtime analyzer, I was feeling pretty horrible with my purchase, glad to find at least some path to good results (even if I may be running too tight, we'll see)
makes sense because the motor is actively forcing the belt to hold in position when idle or pushing/pulling when moving. at least when you tension while under motion it should probably equalize over time? but switching the motors off and re-homing is always a good call when doing belt tension. you could also spot when your tension changes erraticaly or in big steps - that could hint to a motion system issue (binding and stuff).
I had to repeat the process five or six times before I could start it and still have X activity gone from the result
Feels like putting new strings on a guitar
and then when you do Y T1 you get the same result? or for X?
I didn't, after I did this on Y0 with T0, my X1/Y1 were really looking good as well so I sent it
Well, it's kinda tricky. You can't really have a significantly different tuning because the belts also keep the gantry square.
So I fixed my main toolhead, and in doing so got the others in line because they had almost the same tensions to keep gantry square.
hm.. well i'll see what happens. but looking from this angle it makes sense. you tune both x-belts against y T0. and if your y-belts are the same (or close together) then that automatically also applies to T1.
it's a closed system - changing one thing influences all the other things too.
my Y1 still had an additional spike, but it seems to be common on most of the graphs I see here
yeah that 125 Hz spike is an idex mistery. many people have it and no one could figure out what this is yet :kekw:
but it shouldn't really matter for printing because it's way outside the usual movement range.
Nah I mean another inbetween there, I think it might be due to the different position of the X belt on T1
but it was also highly refined after this process, had only Y vibrations
I just ran an 8hr rat pack print so I am sure they've settled and I will need to do a touch up tune, I will post some full graphs here when I get a chance 👍
Would have originally but I was hunting down the 125hz when I made this thread lol
yeah those random spikes here and there could be all sorts of things. looking at my graphs - those spikes around 50 Hz on my X? yeah i BET that's my electronic enclosure. the top plate with the fan starts rumbling like crazy between 50 and 90 Hz 🤣
sec, I got a quick hack for that
I just grabbed some longer M3s and M3 nuts, lifted the lid away
just screwed the bolts in til they stopped turning, then tightened the M3 nuts to the plate
that's just so beautyful and insane at the same time, macgyver would love it 🤣 guess i'll try that and see how it improves things.
but in the long run i guess i'll just straight up replace that enclosure with something sturdier. like a proper junction box housing or something.
straight wall is a bad example for IS, try something with a lot of angles like embossed text
So I ran into a snag with the setup last night. Made it about 10hr into printing and had a layer shift in X I believe (was shifted equally in both X and Y). Might be my method of tuning this runs X too tight, haven't had a chance to check. My worry is if it's even possible to get clean shaper results without getting the belt this tight.
Ran fine through a 9hr PLA print, this shift was during my first fully enclosed and heat soaked ASA print, so could also be some binding appeared somewhere. Will report back
Could be overheating motors, what was chamber temp?
56C~
at least according to T1 which was parked to the side
I hadn't considered that
60c is getting close to pushing it for the factory ratrig setup from what I've been reading.
But i think your motors should be fine... Maybe lol... Is your cooling fan directly over the drivers in your electronics enclosure?
Yep
I assumed it was lower than 56C, hotend would be decently close to the bed (though, I guess the motors are too)
Tonight I will loosen the belts off and check for binding while heat soaked. If there's any I will realign rails, and this will give me a chance to try my belt tuning ideas again from scratch
Then I will try a bit cooler
Re-racking wouldn't be a bad idea
Also possible it collided with the print? Especially if using interference style infill
It could be. It's the Rat Pack parts and they print them with a 45 degree overhang, so had some slight curling
I'm still new to ASA, so figuring that one out. Haven't had an enclosed machine until now lol
shouldn't be, LDO HT which are used for the X and Y are rated for 180C
Was not aware, thank you for the info :)
So I found the issue. T1 belt keeper gave way under the tension, so this is definitely too tight. Not sure how to get clean graphs out of Y at this point.
@TheBrigandier Maybe you should dry your filament more
:pepe_british:
j/k...lol just trolling and im entirely not useful here
Because of these issues and the lack of instructions or any clue on how to fix the prints, I feel like throwing the IDEX out and going back to a hybrid.
I feel you man
like for my buisness its not worth anymore to try next 2 or more months just to try print something
Im starting to think the same thing. What is the best Y graph you've seen? I wish they atleast gave us a decent goal to shoot for. Not just a good "Luck"
Not saying I have the best Y graph or know everything on tuning an IDEX, but it's possible.
https://discord.com/channels/582187371529764864/1324491900052439134/1326013966773780582
Input Shaper graphs are a good troubleshooting tool, but is not the end all be all. These graphs will never be perfect (and if you get them that way, they wont stay that way). Every machine is built differently because each person and their own personal tolerances for assembly is different. Input Shaper just says that you can print at that acceleration and below without having ringing. That's it. You can print above that accel no problem and majority of the time never know you dont have perfect graphs. I've seen some phenomenal prints come out of my machine when it says 3k accel and the graph looks like a mismanaged landfill
TL:DR, benefits from high X tension and slightly lower Y tension. If you target greater than 6k accel in MZV, then you can print the basic profile with NO ringing.
what belt keeper gave out? The toothed one or the weird trapezoid shaped one?
thats the highest Y I've seen I thought someone having 5k was amazing. I really hope RR add IDEX belt tuning to there guide. It says they are trying to craft one.4
im done XD i will go back to hybrid on my 500 bc of its size its not worth it to try do it anymore. i have v-core 3 500 i might use some of its component and just build idex v4 300 it will be easer to work with
shorter belts etc.
On my way to the mental house with theese belts I found out that despite IS told me to use ~50Hz for mzv the print result was still not good(big dip after corners but no ringing afterwards). Spent some time to adjust the belts while printing ringing tower but doesn't seem to affect. But what is actually helped is to set Y shaper to 30Hz which almost eliminated the dip(tried from 30 to 50 but 30 turned out the best). My goal was to print at 6k acc so I ran all the tests with that. Now I print at 250ms and 6k acc and so far looks good. Belts are now almost equal to each other at around 6.5mm by the BTT tensioner.
Go back to hybrid and wait for official instruction is my option i choose
well.. guess another good tipp is: if your graphs look like crap - check that your electronics box is not adding noise. @TheBrigandier nice hack with setting off the backplate... i did it with 30 mm stand-offs i had left over. for y i'd say i take it - now just figuring out if i can tune x a bit more.
Any tips / guide to how you got these graphs? They are decent I'd be thrilled.
mostly fixing stuff step by step - my machine apparently had (and maybe still has) a few "mechanical gremlins". my mounting plates in the back were miss-aligned (had a 0.1 mm gap on the lower plate of the upper hybrid assembly on the right side), that in term caused my gantry to always be a bit racked when i tried to de-rack with the gantry fully pushed to the back. that gap was small - but over the distance (it's a 300 machine) that added up and caused a few degrees of racking back to front. also screws.. how often i tightened every screw i could reach. make sure your XY-joiners are as tight as can be (only exception: the bolts holding the idlers. those are snug - but not over-tight). i greased and de-greased my rails three times. have even replaced my y-rails with older, known-good 400 mm rails from an old printer that worked until the end (de-comissioned hypercube evolution with rail conversion). i worked on this on and off for almost 3 months now.
and also tried a whole bunch of different tipps and tricks from wetson, xulkal and all the others that hunt these perfect shaper graphs. but i guess in the end the best advice is: make sure that your mechanical sub-system is in good shape and maybe try a few test prints before going in too deep. idex apparently can't have perfect shaper graphs just because it's an idex system. if it's not obvious that something is mechanically wrong with your machine or that you have a defect somewhere... well then i must admit it's not worth it overdoing 😄
i will try one or two things now (tightening/loosening idler bolts in the back - mainly on the right side where T1 sits, because i suspect they are a bit looser than on the left where T0 is, then maybe play with the tension of the x-belts a bit more) but then i'll guess i call it quits if it doesn't change much.
Okay, so you have a 300 IDEX that explains why your recommended is so much higher than mine at 500. Man It's so much harder than single Hybrid. I think Im just gonna shoot for the same tension/ good racking and let my IS graphs be what they be. I feel like equal belt tension is the most important.
I just wish RR had a guide like they did with the single head
yes, for the bigger machines i can't say much apart from that their Hz-values are lower. also their belt tensions. the 300 machines can cope with higher tensions, and i think they also need it because of the shorter belt runs. what was also noted for the smaller machines is that they seem to like it when the Y-belts are on lower tension than the X-belts. with all 4 belts at the same tension i always got worse graphs. lowered the y-belt tensions a bit -> better graphs. for the bigger machines that's totally different.
I'd be happy to just drive the X activity out of the Y graph
and yeah a fully-fledged guide would be nice, but i guess we provided them with so much information in our endeavours that they either compile everything into a big "how to tune idex belts"-guide... or just realised themselves how hard it is to tune these beasts properly.
i'm not sure if this is even possible because all four belts influence each other. look at my graphs - that peak only goes so high because X and Y add up.
It's possible, just need to get X so tight it breaks things lol
yeah but honestly, having a bit of a physics background, i'd rather have both peaks add up at the same point - that means it's overall a sharper peak and IS can filter it out better.
My worry is the X activity is a result of the hybrid belts and toolhead belts not pulling Y in time
Basically the toolhead belt goes to move in Y, but hybrid has either already done so or hasn't done it yet, so it translates to tiny X movement
It makes me wonder if there's a way to adjust the timing of steps between the hybrid and toolhead motors, make this a timing issue rather than a belt tensions issue
hm i don't think so. you'd hear that - in y all 4 motors push/pull at the same time. if just one would lag behind, your whole movement coordination would break down and then the gantry would bind up. if you have ever miss-configured a corexy-system so that both motors spin in the same direction... well.. it's NOT a nice realisation :kekw: can't hit that E-stop fast enough.
Yeah I am talking ultra minute differences
something that could be absorbed by the small amount of backlash in the system
I guess what I am saying is why does having X ungodly tight result in this graph?
hm then you'd be in the area of microsteps and that would be something that klipper itself would have to deal with - that's why you have that dreaded "timer too close" error
over-dampening
Basically so tight it can no longer vibrate in X?
your X is SO TIGHT that it physically can't swing anymore
Might be, but my X graphs looked pretty great too
yeah but that's where you get bearing damage, maybe even motor damage (maybe not so on our double-supported motors) if you really overdo it
or - clamps breaking 😄
if it's so tight that it's physically not possible for the belt to swing anymore... you have very high forces on there. but usually you should see this in the X-graphs - if you go into "over-tight" territory, they should get all weird and crazy with spikes and peaks everywhere.
because then they can't "balance" out any kind of noise in the movement path
I still had good definition there. ZV recommendation with 9000+ accel I believe it was, at work at the moment can't check
Obviously not going to run them at that anymore, just seems like a lot of X activity for a Y oscillation test
Maybe difference in belt length too, a lot more backlash in the toolhead belts vs the hybrid belts
just something throwing that timing out I feel
yeah belt "alignment" is also a big issue on idex, on both loops. just one or two teeth off and that's it, you can't get good graphs.
I do feel that could be solved with a design change. Make the toolhead have some type of small length adjustment so you can set tensioners to the same distance, equalize the belts at the toolhead attach points, then do the rest of your tensioning at the tensioners
basically fine tuned length adjustment at the toolhead somehow
it's not so bad on the hybrid belts where you can see how the belt is inserted into the carriage blocks
but trying to do it on the toolheads is a pain
yeah i remember the old EVA days on the VC3.1 where you had to tension the belts on the backside of the toolhead. it worked, but it wasn't easy to do.
but what also adds insult to injury - that whole 4-belt-system is so finicky. you change a small thing (adjust belt tension a TINY bit - and we speak of not even 1/8th turn on the screws) and you throw all your graphs out of the window 😄 i'm testing a bit, adjusted the idler bolts a bit (make looser/tighter) and T1 Y fell down from MZV to EI, altough MZV would still be doable, looking at the graph
maybe i'll try tomorrow and see what happens if i go lower with x-belt-tension again. bit by bit.
i have a suspicion that i'm a bit too high now.
also went higher and higher, and it did improve things at first, but now those X peaks just won't get any closer. my guess is, i'd need to snug up my bearing stacks a bit more and lower the x tension a bit again.
In your opinion should the belts be tighter on 500 or looser?
Im guessing it should be opposite with the Y as well? Like Y should be tighter than X
Did you get these graphs from a 500 IDEX?
judging by what i've read here in this discord: looser. it's a bigger machine, the belt runs are longer.. but in the end it's also just trial and error. same thing with "y belts should be looser than x belts" - we had folks here that had good results with this approach, but also people where this didn't work and they had to bring all 4 belts to equal tension or make x looser than y.
it's a big "depents on your machine and the 'quality' of your assembly", so to speak.
plus IF you have a mechanical issue somewhere - you can tune all you want, until you find and fix this issue, you won't get good results. a normal core-xy-system is more forgiving with stuff like slight miss-alignment and so on.
otherwise i wouldn't have taken apart my gantry 3 times and my motor assemblies two 😄
Gotta turn motors off between tunings as well
Takes forever
@TheBrigandier had a dumb idea now: maybe i'll use the realtime analysis tomorrow (or when i find the time to do it) and tune my belts specifically so that the X- and Y-Peaks line up. basically the opposite from what you suggested. not "kill" the x-influence - but rather add it to anything that Y is doing. and then see what the input shaper says to this.
Try it. I did and couldn't affect the peak enough to get alignment, but there's so many combinations I would hardly say I "covered it".
or do you mean heightwise?
height and/or frequency. i mean, it's "simple" physics. if you have a wave at a certain frequency, you can add or subtract another frequency. you just need the right "phase shift" so to speak. if you have two waves that have the same frequency, but opposite amplitudes, they cancel each other out. if you have the same frequency and the same amplitude (both in the same direction) then they add up.
Yeah I thought you meant frequency.
Dunno if it helps, but I did notice that whenever I got X to go down, Y would go up. I had a thought that I would try to get as much activity out of each one so they could combine and generate a higher value altogether, but they seemed to only want to equalize each other. I didn't exhaust this though
yeah that's the thing with the belts - all 4 work together (or against each other). i think the key here is to find the point where they all swing with the same frequency and amplitude. the point where they all resonate equally.
Made it home, here's the T0 X graph that I had to go along with that Y graph at the start of the thread.
nice graph. but you even see it there: your peaks are all aligned. y is swinging with x, at the same frequency (or roughly the same) - so the combined peak (X+Y+Z) is higher because they all add up.
Yeah I agree, but it's at least a single peak, one sec
this is what I was fighting before going gorilla on the X belt
I mostly had single peak on X through all this, Y was being ultra stubborn
i see. and i had graphs like these on Y when my Y belts were too tight.
once i lowered the tension a bit and went up with X, that went away.
I tried all tensions on Y, loose as can be, incremented all the way up to very tight by like 5 degree turns
interesting. and i guess you checked your mechanics, made sure your plates are aligned, idlers not too loose/too tight and so on.
Best I could get with normal tensions was something like this
Yeah I reracked everything, new lube, reassembled the motor stacks with emphasis on alignment, got belts off idler edges, etc
and you have what kind of machine? 400? 500?
500
so it seems those bigger machines are even more tricky
they have really long belts that like to vibrate against the 3030
I would assume the 300/400 don't have this as much
at some tensions the hybrid belts wobble as much as half an inch lol
hm before i had my plates aligned my X-belts happily vibrated and slapped the extrusion sometimes. also my y-belts vibrated like crazy. i guess it's because of the slight miss-alignment i had.
maybe something is binding slightly somewhere in your machine, what speaks against this is that your power is in 1e4 range. but your graphs look suspiciously similar to mine when i hadn't done the plate re-alignment.
and my gantry was racked slightly, because of this.
I even did a bunch of tests with X gantry carriages loosely attached, just in case there was some binding there
but it ended up in worse results, got slop
yeah with loose bolts everywhere you won't get good graphs. too much vibration from the screws.
With belts off I had it pretty butter smooth, and with belts on and correctly mounted (length down to the tooth), gantry sits square on its own
no need to pull it either direction
Assembled with machinist square, etc lol
on a granite surface, best flat spot I had
I think I will just re-belt, get some decent tensions, and roll with it awhile in hopes someone has a breakthrough (or Ratrig releases a tuning guide...)
I also have the Mandala motor plates coming, might tear into that down the road
well at this point it's just a bit comedic how some of us try all sorts of stuff to find the "holy grail" in idex belt tension 😂 while all the "send it!"-guys also get good results. it's a bit unfair. maybe i should also just gloss over all the IS stuff and print stuff...
It's hard when you see someone post some really good graphs
but I question how well they maintain that
first i just wanted to make sure everything's mechanically sound... then i fell down that rabbit hole :kekw:
yep, it's like heroin except not enjoyable
:kekw: 3d-printer-belt-tuning-addiction...
Got a GT2 belt wrapped around my arm, biting it
injecting that MZV
dances around the printer grunting "Tiiighteeen... All.. The.. SCREEEWS!"
but in all honesty, it's sometimes frustrating. but on the other hand you always learn something new. when we're done with our machines, i guess we will know them really well. or just sit there and say "i have no idea what i'm doing but i guess it's the right thing?"... yeah.. i'll see what i can come up with with the realtime analysis stuff.
I can say I am the guy who actually did tighten shit until it broke, for science
I just really, really hope there's a solution and not "Welp, 500 IDEX just be that way"
as long as we don't have an official answer, i'd say that could also be just the case. the bigger machines always had their problems and kinks, even in the VC3/3.1 era. that's why i personally always go for the "smaller" sizes - with a 300 i roughly know what i get, mainly because my previous printers also where in that category.
400 idex
I at least wish they gave us some "Shoot for numbers" I decided im goin for 5k in x AND 3k in y If its all square ill be content. I can still print fast, It'll just be haunted by ghosting lol. Which doesent matter for some parts ya know.
well look at my graphs.. on a 300, Y can easily reach 10k and higher (i've seen people here where they reached 13-15 k on Y) - but X is a real pain to get right.
From what I've read 300 and 500 are totally different beasts to tune. Your smaller size means you can print faster with less vibration
Yeah thats why im going back to hybrid on 500
And
It’s easier to do hybrid from idex xD