CoreXY belt / idler juddering issue (belt teeth against bearings)

i'm building a folding, portable corexy printer, called the Sandwich200 [reprap.org] this is my first design, and it's based around the Fusebox [reprap.org].

both printers have an "offset" layout which means that the belts miss each other and there are no "crossover" issues. however, something had not occurred to me until i looked very very closely at one of the prints, and i am seeing tiny, tiny jolting marks at very regular intervals of around 0.5mm in every single extruded line. it's as if the extruder is "pulsing" but i feel that it's more likely that the amount of vibration and chatter from some of the teeth pointing "inwards" towards the bearings is enough to jar the printhead and cause regular (quite pretty) fluctuations.

so i came up with this as a potential solution, and i wanted to ask people's advice:



like that awesome all-aluminium printer made by the guy who even made his own extruder (can't remember who it was), i have on the Sandwich200 i put the motors "inboard" so require 2 extra bearings per belt than the Fusebox. the diagram, you can see in red the bearings, except the one with circles, that's a GT2-20. teeth are marked in black dots on the belt. there's a single twist marked green arrow.

now, the yellow arrow is my query. would it be okay to leave that as it is (bearing) or would it be better to replace that with a teethed GT2 bearing or something, and do such things even exist? i can't put a twist before that final corner because the distance is far too small, the only reason to consider safely putting a twist in the belt along the length is because the distance is over 300mm.

the Fusebox would not have this problem, a single twist i feel would be ok, because the GT2 is "in line" with the bearings, there are 2 less (none in the bottom right corner).

thoughts and comments appreciated.

I am using toothed idler pulleys (6mm GT2). At first I made my own from miniature flanged bearings and 8mm bore pulleys, but there's a better solution - Aliexpress or ebay. Just search for GT2 idler pulley, and there will be some with teeth
If that twist is there to reduce teeth contact with the bearings, you could just replace all those bearings with toothed idlers, maybe get rid of the complicated twist. Though I can't tell if that will improve print quality, most printer designs I've seen just use smooth idlers...

I would replace the idler for a GT2-pulley.
In my design i have the motor in the corner, so i don't need the (2) extra pulleys. I have also twisted the belts in the same place.
I can slide the motors back- and forwards for tensioning the belts.

Quote
Dalius98
I am using toothed idler pulleys (6mm GT2). At first I made my own from miniature flanged bearings and 8mm bore pulleys, but there's a better solution - Aliexpress or ebay. Just search for GT2 idler pulley, and there will be some with teeth

aawesome ooo, 18x9mm dimensions, hmm... i'm pushing my luck on space (the final frontierrrr...) - with the rev2 sandwich200 i'm planning, i might not even be able to *get* a twist in, the belt returns so close, but also 18x9mm i will have to think hard about how to fit that in, as well!

Quote

If that twist is there to reduce teeth contact with the bearings, you could just replace all those bearings with toothed idlers, maybe get rid of the complicated twist. Though I can't tell if that will improve print quality, most printer designs I've seen just use smooth idlers...

yeah. it was the rippling that caused me to wonder. maybe most people aren't trying to print at 350mm/sec travel speeds though... which i just managed today (and got a clean print!) with a volcano E3Dv6 upgrade and a 0.4mm nozzle (yaaay!)

but, it was at the more sedate speeds (150mm/sec or less) that the rippling occurred - so thank you, i'll see if i can find those idlers from my usual suppliers, see what they have.

Edited 1 time(s). Last edit at 04/18/2016 10:31AM by lkcl.

Quote
gforce1
I would replace the idler for a GT2-pulley.
In my design i have the motor in the corner, so i don't need the (2) extra pulleys. I have also twisted the belts in the same place.

great - so i'm not the only one

Quote

I can slide the motors back- and forwards for tensioning the belts.

okay that's quite nice. i've altered the carriage so it's easy to unclip the belt and move it along another notch. i say "easy", what i mean is, it's fiddly as hell, with a screwdriver upside-down hoping like hell the retainer bolt doesn't drop out so a way to move the motors along instead would be neat.

How about this? Keeps the motor inboard, puts the teeth on the outside of the idlers, no twist in the belt, requires one less idler.

Quote
FA-MAS
How about this? Keeps the motor inboard, puts the teeth on the outside of the idlers, no twist in the belt, requires one less idler.

nope - that's not a corexy design of belt arrangement. corexy, the key is to go from one corner "down-across-down" and end up diagonal in the opposite corner you do that from the opposite corner as well and you have the +/- effect. the belt arrangement in the diagram you did goes down-across-UP, which means that any up movement the belt will go slack, and any down movement it will go tight and stop movement from happening. you need that down-across-down, it's where the balancing of forces comes from when you have 2 of them in a mirror.

Well the truth be told
The diagram in the initial post
is ONLY one half (or one level) of a coreXY belt arrangement

Correct me if I am wrong, fuzebox CoreXY belt route shown below





You are using a toothed drive pulley on drive motor, smooth (non-toothed) pulleys
to change the belt direction,
a 180 deg twist in belt, to present the belt teeth to drive pulley
all this to move drive motor in, to interfere with axis travel.

All this to maximize the use of smooth pulleys?

You believe the belt teeth moving over the smooth pulleys is causing juddering (jolting, pulsing)
in your X Y movement?

Have you considered larger diameter smooth pulleys

TIMING BELT DESIGN AND INSTALLATION SUGGESTIONS: GENERAL GUIDE LINES
Idlers, either of the inside or outside type, are not recommended and should not be used except for power takeoff or functional use. When an idler is necessary, it should be on the slack side of the belt. Inside idlers must be grooved, unless their diameters are greater than an equivalent 40-groove pulley. Flat idlers must not be crowned (use edge flanges). Idler diameters must exceed the smallest diameter drive pulley. Idler arc of contact should be held to a minimum.

oops just realized this is ikcl post ---- sorry for being sarcastic and not humbling myself to your mechanical design excellence

cozmicray, your post is informative and funny, and your apology and... and... how to put it... view of my "random excursions into unknown 3d printing territory with cheerful complete abandon and recklessness" as being "design excellence" is as funny as it is appreciated.

Quote
cozmicray
Well the truth be told
The diagram in the initial post
is ONLY one half (or one level) of a coreXY belt arrangement

Correct me if I am wrong, fuzebox CoreXY belt route shown below

[attachment 76676 Fuzebox_Belt.jpg]

yep that's how the Fusebox does it - there is a design of all-aluminium printer, i wish i could find it again, he got 9000mm/s travel speeds and designed all the parts to be simple-CNC-machinable as well as 3D-printable, he added the extra (2) bearings per side so that the motors didn't stick out by 1/2 their width.

Quote

You believe the belt teeth moving over the smooth pulleys is causing juddering (jolting, pulsing)
in your X Y movement?

that's the *hypothesis* that i am currently investigating as to whether it could be true. other hypotheses include that the hardboard panels are acting as resonance chambers, amplifying the sound of the motors sufficiently to cause the Z-bed to vibrate.

Quote

Have you considered larger diameter smooth pulleys

i did... but i've prioritised space considerations as a high-priority design criteria for the v2. larger smooth pulleys would mean larger outer dimensions: i'd prefer to try alternatives and only go for larger diameter pulleys if absolutely necessary. good suggestion though.

Quote

TIMING BELT DESIGN AND INSTALLATION SUGGESTIONS: GENERAL GUIDE LINES
Idlers, either of the inside or outside type, are not recommended and should not be used except for power takeoff or functional use. When an idler is necessary, it should be on the slack side of the belt. Inside idlers must be grooved, unless their diameters are greater than an equivalent 40-groove pulley. Flat idlers must not be crowned (use edge flanges). Idler diameters must exceed the smallest diameter drive pulley. Idler arc of contact should be held to a minimum.

yeahhh nophead used bearings and then washers on the belt-loop for the x carriage on the mendel90. he also added twists to the belt but that was so it would avoid hitting bits of plastic on its travels.

what i *have* seen - on the original reprap - is to add a convex "ring" on the bearings. what that ends up doing is, if the belt travels off-centre, the convex curvature makes it go back to centre. a bit like when you're driving a car along worn ruts in a motorway and the steering gets ripped out of your hands (which happened to me when i had a '99 ford mustang cobra - 235mm-wide tires and sports-style power-steering with only 2 and a half turns from full-lock left to full-lock right: hit the edge of the rut, the power-steering took over and i was half a metre over the white line before i got the car back under control. salutory engineering lesson, that one...)

but, the side-effect of those convex rings is that the belt tension will subtly change.... which on a corexy design i feel is a baaad idea.

so... yes... i knoooow... edge flanges on the flat idlers *sigh* because yeees, the belt is rubbing on the plastic - this is on the sandwich200v1 and also the fusebox. i'll see if i can get hold of some thin 20mm diameter washers.

oops just realized this is ikcl post ---- sorry for being sarcastic and not humbling myself to your mechanical design excellence
[/quote]

bow, bow, scrape "in the country of the blind, the one-eyed man is king...."

Mechanical excellent grasshopper

Hicups in X , Y or both?

Pure X and pure Y prints should be done and examined
Diagonal print may also show something

Difficult to separate X , Y axis issues in CoreXY system

For resonance -- you may want to put a large weight on printer (avoid crushing printer)
see if that changes things.

What stepping is selected on stepper drivers? Are they all the same?

Tape a red line on pulleys --- does it rotate smoothly?

in a country of the blind --- everything is heard / felt !
-- even the errors from the 3D printer of the one eye king

Quote
cozmicray
Mechanical excellent grasshopper

ahh glasshoppaa...

Quote

Hicups in X , Y or both?

both.

Quote

Pure X and pure Y prints should be done and examined

a single piece with square corners printed straight on x and y showed 0.5mm warbling on both sides.

Quote

Diagonal print may also show something

that's a good idea. will try that.

Quote

Difficult to separate X , Y axis issues in CoreXY system

For resonance -- you may want to put a large weight on printer (avoid crushing printer)
see if that changes things.

i bashed the boards in the middle, they resonate clearly and loudly (hardboard, after all...) with a distinct audible note (somewhere arouuund.... 50hz, it's similar to mains-hum) i think i need some cushions...

Quote

What stepping is selected on stepper drivers? Are they all the same?

200st/sec, yes.

Quote

Tape a red line on pulleys --- does it rotate smoothly?

if it makes any sense, they "sound" fine when moved by hand.

Quote

in a country of the blind --- everything is heard / felt !
-- even the errors from the 3D printer of the one eye king

beauuutiful

Edited 1 time(s). Last edit at 04/21/2016 10:44AM by lkcl.

Quote
lkcl

Quote
cozmicray
Mechanical excellent grasshopper
Pure X and pure Y prints should be done and examined

a single piece with square corners printed straight on x and y showed 0.5mm warbling on both sides.

Quote

Diagonal print may also show something

that's a good idea. will try that.

[/quote]

that's iiinteresting.... ok, the diagonally-done print still shows minor warbling - i'm now using a 0.6mm nozzle - the warbling is about 1mm apart. on the straight X-Y print, the warbling is much more pronounced, and it's about 0.6mm apart. my guess is there's sqrt(2) in it. it's on all sides, so not an "X only" or "Y only" issue but definitely both.

half-way up the print i put cushions on the boards: no change in the effect, the warbling is still there on both parts (just very very faint on the diagonally-printed one)

hmmm....

Grasshoppaa

"What stepping is selected on stepper drivers? Are they all the same?"

Are the driver jumpers (under drivers full step, half step, 1/4 step, 1/8 step,1/16 step,1/32 step)
set the same

maybe try another step setting

Perhaps try a much slower / faster printing speed?

The line on pulleys (maybe a toothpick) may show a / all pulleys juddering!

Extruder juddering? or Build stage shaking around?



Perhaps placing printer out in well groomed zen garden next to Bonsai tree with proper fung shui?

Quote
cozmicray
Grasshoppaa

"What stepping is selected on stepper drivers? Are they all the same?"

1/16th and yes

Quote

Are the driver jumpers (under drivers full step, half step, 1/4 step, 1/8 step,1/16 step,1/32 step)
set the same

maybe try another step setting

*sigh* yeees, bugbear about the duet, you can't software-select the driver microstepping easily, you actually have to cut tracks to do it.

Quote

Perhaps try a much slower / faster printing speed?

i ramped down to 200mm/sec, acceleration was quite slow so it was doing less than 200mm/sec for sure, and it's still present.

Quote

The line on pulleys (maybe a toothpick) may show a / all pulleys juddering!

ahhh yeah, that makes sense. no, i am putting my finger on the pulleys and they're definitely vibrating quite heavily. i get the feeling it really is the teeth-facing-bearings. trouble is i have to re-print the damn carriage (3m of PLA) in order to turn the teeth round (so that i can grip an inverted belt, as described at the beginning of this thread) or i have to redesign the idlers to take GT2-idlers.... grrrrr.

that, and i'm redesigning to a sandwich200v2 which will use a flex3drive - i'm reluctant to spend effort investigating something that i'm going to replace in only a couple of weeks.

Quote

Extruder juddering?

could _possibly_ be the extruder.... i had to design it myself (geared 56:13, MK8 at the centre, 3mm bowden with a 6mm OD tube... not exactly normal in other words) and early versions the herringbone gears would be off-centre, causing "skipping" at specific points as it rotated - very annoying.

anyway yes, possibly it's down to a "jump" in the extruder on each revolution... let me think.... no, don't believe it would be that. if it was, the bumps would be a lot further spaced out (like, one every 50 to 100mm and i would never notice them).

_could_ be high-frequency juddering of the extruder, but it seems highly unlikely. i'll try to put my finger on the carriage next time it's moving, to get a feel for what's happening.

Quote

or Build stage shaking around?

mmmm.... don't think so. it's too regular. if it was "shaking" i'd see variations - this really is way too regular and too close together. like, it's at.... let's say that the speed is 150mm/sec, say (due to acceleration reductions), the frequency is very very specifically at.. 300hz (because there's around 0.5mm between bumps).

thing is, even when the printhead is accelerating, if it was the extruder (or build stage vibration), there would be variations. but it really _is_ very very exact and specific and regular, all of which is pointing, for me, towards it being the teeth against the bearings. at a 45 degree angle the spacings increase by what looks like sqrt(2) which kinda clinches it for me.

Quote


Perhaps placing printer out in well groomed zen garden next to Bonsai tree with proper fung shui?

ahhhh goood philosophyyyyy [fifth element quote - little chinese guy selling thai food to corben dallas...)

sadly this has PLA: if it gets lots of sunlight it'll melt.... i've made that mistake before... oops

Edited 1 time(s). Last edit at 04/21/2016 04:48PM by lkcl.