Universal Auto-Rewind Spool Holder

by VincentGroenhuis Jan 6, 2019
Download All Files

Thing Apps Enabled

Please Login to Comment

Comments deleted.

If anyone is interested in seeing this in action, I created a short video on YouTube of the filament switching process using this spool holder.


I have it working fairly well, so I'm kinda afraid to touch it. I've had the most luck with PLA+ for the springs, but PLA should work fine. PETG broke easily, but may try playing with that again eventually. Over tightened the clutch on a couple which deforms the spring eventually causing it to fail over time. I have it more loose at the moment and seems to be working so far.

NIce, thanks for sharing!

Is it possible that the clutch and the dial of V3.2 missmatch?

There are three pointers and five tabs. There is one correct way to assemble these together, and four wrong ones.

Align the "free standing" pointer with the dot above the rightmost marker (there is only one such dot) and it should match.

are you suppose to remove the bottom flat part of the spring after printing?

No. The bottommost two pictures in the instructions show how it is prepared.

The spring model does not slice with cura!

EDIT: I got it to slice but its 14HOURS to print just the spring 80% of the print is travel any settings to tune to make this faster?

Yes: turn combing mode off and printing time should drop to under 3 hours. You can find the "combing mode" setting in the Travel section.

I've tried a couple of materials for the R5 springs and this is what I've found so far:

1) PLA+ = Hard to separate the 2 springs from each other but doable. Broke one spring at the center where the spring connects to the hub after installing and using from over tension trying to get it not to slip as much.

2) PETG = Broke 3 out of 4 just trying to get them to release from the back plate. The one survivor is working great.

Getting the clutch to hold as tight as possible, but still allow it to give is most important from what I can tell. The one spring that is working is holding well and only slips about a half inch each time. The PLA+ spring I could get to hold, but would slip like 3 inches when it gave way.

Not sure how to help anyone with this info, but it's my observations.

Going to try a regular PLA spring to see how it does. Will not be trying ABS.

I'm wondering if a ridged outer ring on the spring would help it not to slip so much and still leave enough wiggle room to get past the ridge? I would like to redesign and test, but not sure how to modify the existing ones without just starting from scratch.

Just added a zip file with the v3.02 pre-release files which also include STEP files that can be imported in CAD software. Well, in theory at least, as importing the spring doesn't work with my software yet. Many other files also have been updated to give the double spring more room and reduce friction. Hope this helps!

Ridged outer ring: I tried it once and it turns the rewinder into a musical instrument (rattle)!

Awesome! I’ll check it out. Thanks for the reply.

Well, I have to say, this is pretty amazing.

I have had all kinds of trouble with the original components that came with my MMU2 to the point I had to babysit multi-color prints due to a couple of the original components not catching certain filaments in the front and allowing them to roll up when unloading like they were supposed to. This would cause the roll to get all tangled up with itself because the loose filament would just wrap around the roll and slide off sometimes.

I've printed one of these (took a long time) and it's now sitting on my most problematic position. It has worked well so far. Getting the clutch at the right tension is obviously very important. I'm now using it to print the rest of the Auto-Rewind spools. The spool holders next to it seem to know what's coming and morale is low. ;)

I'll post a make when they are all done. I'm also considering posting a video of how to get the spring loose (R2) as it took me a while to figure it out and I'm an engineer! Looking at the layers in slic3r as mentioned did help to see where they were "suppose" to separate, but it look some tempting to get them separated (and a hobby knife). Make sure you clean up all of the left over plastic that hangs around from the bridging so it doesn't get stuck when winding/unwinding the spring.

Great job on this. I know people have had issues and I get it. Luckily, my printer is tuned in very well (lots of trial and error over a year), so I haven't had issues like most, but I can definitely see where the slightest problems in the print could cause things not to work.

Currently printing them with eSun PLA+. May run out of the main color, so the stands may end up being Sunlu PETG.

Update: Found my first problem. When printing multiple colors and one of the colors is only a very small amount overall on the print, it's loosing too much tension on the spring because the clutch will let loose and it's not enough printing to wind it back up (e.g. printing the dials). Over time, it will loose basically all tension and can't wind it back up unless you do it manually between prints.

It has made me think of a new base with a new winder and axle. This would also solve peoples problem of holding the filament onto the roll while winding it up. You could wind it up after you load the filament! Off to Fusion 360!

I am having a hard time getting the spools to rewind. It starts to rewind but the MMU pauses for a second and the spool doesnt start rewinding again. I wonder if a taller spring would give more torque? Double with it is now. Or even bearings on the axle ?

I am using Hatchbox filament, kinda heavy.

Thank you for the XXL Nut and for this design.

I would say check these 3 things:

1) Make sure your clutch is as tight as it can go and still function properly. I originally had my clutch one less that where it is now because I didn't think my printer could pull the filament well at that strength. I noticed with smaller layer times on that color, it would slip the clutch only once which made it not very tight on the next run. In other words, the slip on the clutch was way too much movement. Making the clutch tighter allowed it to slip a few times and not slip as much (meaning it didn't slip very far) and kept the tension.

2) Make sure the spring is clear of any bridging material that held the spring to the back plate while printing. It can cause the spring to think it's tight if it gets caught and slip around when it tensions. Making sure it's all clean in there allows it to tighten smoothly.

3) Definitely use the R2 stiff spring (if you aren't already). It's just better and can go both ways for tension.

What version of the spring are you using? It should be possible, many users have 3 kg spools without problems.

@Vincent + @Slyice - I am using SpringFlat_Stiff_v1.02_RC5.

@Slyice Thanks for the suggestions.
1) I've noticed on this spring that the coil winds completely at about setting 11 before the clutch slips. So moving it to 12-15 would not really really give it any more pulling power.

2) Yeah, fragments on one of the clutch housings was making it hard for the whole mechanism to function as designed.

However, I have to say, I did not realize you were suppose to add bearings. lol Just added some. Looking forward to testing.
Now I see "bearings" in the details. Geeze.

Yeah, bearings are a must in this design! I updated the instructions to better reflect this and warn about too high friction in the axle.

What difference bearings make..!! hahaha
Rewind is working! -- reprinting using v3 with 2 colors (black/white).

/bow /bow /bow

Hi, just a quick question I haven’t been able to answer from the description and pictures. I printed the soft spring v1.02 RC 5. The spring and the hub in that STL, are we supposed to separate the spring snd the hub? Portions of the spring freed up on its own but the center portion remains connected on the hub.

Check the very bottom of the main page (https://www.thingiverse.com/thing:3338467), the last two pictures show how it is supposed to be separated. Center portion remains connected :)

Universal Auto-Rewind Spool Holder

Have you considered beveling the bottom of the spring where it meets the backer for easier removal like the depression to allow it to separate in two?

Yes, it is beveled, also to minimize the contact surface to avoid sticking too hard.

It doesn't look beveled? Is there a newer version?

You are right, the bottom is not beveled. I now remember the reason: reducing the surface may make it too loose. There is a huge variation in layer adhesion among different printers and materials, one user already reported insufficient adhesion with the RC5 design. In fact I added the curved slits to allow insertion of a flat screwdriver to release the bottom part of the spring. Hope this helps!

Huh, never thought to push up from the bottom. Worked much better. Also may have helped I printed at a .25 layer height. I didn't have any problems with the layers sticking, though I am using Amazon Basics PETG at 240 degrees.

hI VincentGroenhuis first thanks for your work ;)
I just want to know if i can use the spring only but nor the support because i have a threaded rod like support

Yes, you can use SpringRound which fits on your rod!

Thank you for your quick reply
Ok thanks i'm happy to learn it thanks again and very good job :)

You will also need some clamp or holder to keep the spring shaft from spinning. I remember someone else had the same issue and designed a clamp for it, but I don't know if he actually published it.

Just last night I was looking at the stl files and I saw a notch in the spring so I will pierce it too I think
I also watched if someone published the changes otherwise I would publish them myself as I have some time

Did you have a solution for keeping the shaft from spinning but do not tension it too much so that it is locked?

Comments deleted.


The file list is getting a bit long and it's not clear to me what I need to print. Can you bundle the last revision in a zip somewhere

Just cleaned up the files section, there are only three springs left (two in v1.01 and one experimental v1.02_RC5), the others are in a zip file. I'd use the experimental version, it definitely works much better than the older ones. Regarding the other files, either use an all-in-one plate or just print everything if you don't know what to print. Once you have one rewinder up and running, you will understand the differences between the different stands, springs and nuts. Use the multi-material version if you like to print it in two colors.

Just noticed you got pre sliced files on your website and that includes the spring, Though you also have a separate spring for PETG. My guess is that PETG will be a better solution right?

Also there a 3 nuts, Do you have a preferred one to print? Mainly use Esun, Prusament and PM Filament.

PLA should be fine if you don't stress the spring too much. With RC5 and 2-3 rewinding revolutions the PLA spring should retain its shape in the long term and you can always flip the spool holder around to load the spring in the other direction.

Nuts: use at least one curved nut. Add a second one if you have difficulties in centering the spool. If you unwind the spool in ccw direction (seen from the threaded side) then use a flat nut to avoid unscrewing the curved nut. If you big spools then use the XL curved nut. Hope this helps!

Thanks, I will start one tomorrow with the RC spring then.

First attempt with the spring ended in a spiderweb, Base went fine, though the springs printed too fast or toom much gap.

What settings did you use? Layer thickness, speed, material, cooling?

0.2mm layers
2 perimiters
25% infill
Cooling 100%
small and external 25mm/s

Tried again with 40% speed, Same mess

Could you check in your slicer that the gap is exactly one layer? If no: what is initial layer thickness? If yes, then how would your slicer normally support a spiral structure?

I'm using Slic3r PE with pretty petg for filament, Quality 0.2 for print and MMU2 single color for printer. The gap is 0.2mm and definitely 1 layer.
I managed to print it by dropping the speed after layer 5 is finished to 10% then after layer 8 I set it to 80%. (percentage of speeds mentioned earlier)

Now I'm not sure what part to cut loose, the spring is loose and I think I need to cut something else too right?

Insert axle halfway and wiggle the top half of the shaft free :)

Then the whole spring would be separated from the back isn't it?

Edit, It doesn't. I do think it's quite loose and not very springy even on 10

No, check last two images on the very bottom of the main page :)

Got it, now it's time to print the stand. Tried by holding the axel and walk backwards though it doesn't roll back very much. I'll see when it's in a stand.

Could be if the axle is too tight or the spring surfaces are still rough. Try grinding the spring smooth, remove the clip from the axle, identify problematic surfaces if there is still friction in the system.

Finished the stand now, the filament guide: is that supposed to drop down on its own? And with 10cm 4*2 ptfe there's too much friction to rewind.

There are instructions on the main page, the ptfe tube is supposed to be attached to the MMU unit (or extruder) at the other end.

Try PLA, or turn off fan with PETG? (PETG with fan results in poor layer bonding in my experience, and apparently almost no bonding if there is a 0.2 mm gap)

Hi Vincent,

First of all, thank you for your work in developing this system. I'm very keen on minimizing the space the MMU requires as I'm planning to enclose my printer for printing ABS.

I'm struggling a bit with the springs. So far I've printed the original spring and RC5 of the dual spring. The original spring works quite well but I've managed to break 3 of them at the hub. There seems to be a very narrow window of 'just enough' tension to provide the necessary rewind without breaking the spring. I'm currently using the latest version of the dual spring but I'm concerned as it just does not provide nearly the same level of tension in rewind as the single spring. I've tried both directions and all of the tension settings on the clutch but am just not getting an adequate amount of torque to wind the filament back onto the spool with any tension.

In your experience how much 'pre-loading' of the spool do you need to do before performing the filament load sequence on the MMU? In other words what do you find to be an adequate number of forward revolutions of the spool to provide at least two revolutions under tension in rewind?

Thanks very much for your consideration.

Sounds like you have significant friction in the spool holder. Try to find parts rubbing against each other, remove the clip, grind rough surfaces if necessary, de-grease the bearings (or buy new ones) etc
Then do this awesome test: https://www.youtube.com/watch?v=EfB9rarOsi4

STL on the support leg used in the video..? :-)

Thnks...I never printed those..I use another stand for my rewinders... :-)

Will do, thanks for the tips.

Hey Vincent, Can you explain why the dual spring is better? For some reason I thought the dual springs would be going in opposite directions, instead of just printed on top of each other. What makes this design better? Just looking to understand.

A single spring has up to 3-4 rewinding revolutions, when fully wound up there is a lot of material tension which eventually deforms the spring.
With two springs in series you have around 6 rewinding revolutions. Als you only need 2-3 actual revolutions the material stresses are much smaller.

Another advantage: the clutch rim can no longer twist or translate thanks to the connection to the base plate and axle shaft, solving a nasty lock-up issue.

Interesting, I was not thinking about them operating "in series", I just assumed it was a .5+.5=1. I clearly do not know much about spring mechanics :D

Ahhh I see what you are doing now, makes sense.

The 1.02_RC1 version has springs in parallel, based on that design you are right! RC2 and later have the springs in series.

Hi. I have printed the rouded versions of the spring. I can easily separate 1 part of the spring once printed it. But I cant separate the part of the spring that is on the base of the spring. I mean: I can see in the slicer (slic3r PE latest) the gap sparating 1 section of the spring (marked as a bridge), but at the bottom of the spring there is no gap separateing the spring from the base. is that ok?

See attached cross-section drawing, the yellow parts are to be broken free and the red parts stay connected.

Look at my pictures this is how i should look like when the springs are separated..

i printed the rc4 version in PLA 0.2 mk3 Quality PLA settings in slic3er latest version

now going to print 4 times the rc5 version.

Yes! Thanks for sharing these photos!

Which version did you print, and with which layer thickness?

0.2 layer thickness and SpringRound_Stiff_v1.02_RC2.STL

If by "base of spring" you mean the attachment in the center: you are correct, this is supposed to stay connected, the whole printed spring remains a single connected body. If not: do you have a photo of the spring showing the separation that you have now?

No, i meant the Spring part that is atached to the base
. There is no Gap between the base and the spring

Could you post a photo of your spring?

Sadly, I still had problems with getting the spools to keep the filament on unwind. No matter what tension I set, it would eventually slip right before an unload and the filament would fall off the spool and tangle.
Yes, I tried your new design. Yes, I tried every of the 17 tension settings and filament arrangement. No dice whatsoever. I'm just gonna have to go back to the stock setup until the MMU2S upgrade comes in.

Pity to hear that! But thanks for letting it know. Things that cross my mind: print a spring at 101% size to get more clutch torque at same settings, change spool locations so that filament path is as straight as possible (no sideways bends, or just very gently only), use stiff version of spring (I will upload stiff version of 1.02 RC5 soon). If you post a picture of your setup then I can take a look at it. It should be solvable, most users did not report any problems.

Hey Vincent, are you still planning on uploading a stiff version of 1.02 RC5?

Done! Check the files section.

Hello. I have printed a few of the 1.02 RC5 stiff spring designs. Unfortunately each spring has broken while in the process of unwinding it after it printed. No matter how careful I am they seem to break at the same location.

I looked carefully at the model in both Slic3r PE as well as S3D and I think I might see why they are breaking. In the stiff spring the "thick coil" that splits into 2 "thinner coils" only has 1 perimeter layer on the inner portion, but has 2 on the outer portion. The soft spring has multiple perimeter layers on both coils. I have attached pics to show what I am referring to.

Would it be possible to thicken the inner coil where it splits off on the stiff spring to reduce breaking?

The particular slices you are showing are at beveled slices which are specifically meant to make separation easier. The real strength of the spring blades are in different layers which have more perimeters. In RC5, stiff spring has at least four or five perimeters everywhere. If the spring breaks at that point just during unwinding after it printed then there must be a problem with the filament or the printer, what about doing some strength testing? Anyway, attached is RC6 which is thicker at the location you zoomed in, hopefully this helps!

Awesome.. thank you!

Also, did you try v1.02 RC3, RC4 or RC5? These springs have a special design which allow for many more rewinding revolutions (five or six) which gives more headroom right after slipping.

can you put in the description all settings you used or what is prefferred i am a newbie sort of speak but its also better for everyone to have a startpoint wich settings to print are used, especially the spring etc.

thanks in advance, i am going to print this later today too:)

Bedankt Vincent voor deze post heb er 5 nodig heb mij de mmu2 besteld duurt wel nog tot mei 2019 voordat die geleverd word:)

I myself use PLA, 0.2 mm layer thickness, 3 perimeters, 30% infill, no supports. I also experiment with PETG springs, both materials work fine.

oke wil try that too and any special temperature settings for filament and the bed?

Just stock settings. I mostly use Fillamentum, Prusa and Colorfabb filament. The Genetic PLA profile in Slic3r PE works fine.

Thanks vincent printing now wish me luck:)

OK, I wish you luck now!

Is there any preferred, or required layer heights for the different bits?

I use 0.2 mm for al parts, but different layer heights should also work fine.

I'm pretty much confused about the SpringFlat_Stiff_v1.02_RC2... it is printed on a solid plate and the outer ring and the inner "nut" seems bonded to the plate. as I can see, the spring itself is seperated by a one layer space so I loosened it. But now I'm confused. Since the spring is loose in the shell and the Clutch presses on the outer shell-case, how does the spring work? the nut drives directly the outer shell... and if i break out the nut from the shell, the spring should be way too loose in the clutch?


I have the same question. First, Yes I have split the spring in two.

I agree, the solid plate connects the center ring, which connects to the axle, to the outer slip clutch ring. The center of the spring is also connected the center ring and the outer part of the spring is connected to nothing. Both SpringFlat_Stiff_v1.02_RC2 and SpringFlat_Soft_v1.02_RC2 have this problem. So I printed SpringFlat_V1.02_RC1 and that one works.

Obviously it was designed this way with a good reason ;)

OH, The center hub splits in half also. I split the spring in half but not the center hub. Now it makes sense.

And RC4 is available, in this one it is much easier to split the center hub.

Just uploaded RC3 which is much easier to split across both planes.

hmm. I printed two 'soft 1.02 rc4' springs. They seemed well-bonded. I sort of "cracked" one to see if that took the spring free, ended up breaking the whole base side (the side against the bed). But since I had the other one I thought I'd see if I could wiggle the center spinner free. I put a screwdriver in it, it started to split but was breaking the base too. I guess I'll print the old RC1 instead.

Here you go, Vincent. It broke when I tried flexing it to break the spring and especially the hub.

Hope things are okay in your country today.

Okay, printed RC5 and got it split successfully. I thought I was going to break it at a few points, the axle and the "inner wall" were the toughest parts. Here's how I managed to split it:

  • Put a flat blade screwdriver between each "loop", twisted the screwdriver to break the loop free from the base. I had to do it twice- eg working out at the 0 deg plane, then at the 180 deg plane. That loosened most of them.
  • Very carefully split the spring in its lengthwise parts, then used my finger to spread and break it.
  • The inner ring didn't want to break loose. Finally I put the screwdriver between the inner and outer rings and it split fairly easily- starting 180 degrees from the "Y" where the spring joins the inner ring, which was the most difficult spot in the inner ring.
  • I really thought I was going to break the axle. Finally I put a knife between the spring halves and carefully split the rings until I reached the axle. After wiggling it a bit it finally split in two.

A lot of work, really.

Here are the same images that I attached to the post, because I think the "uploaded images" are hard to view: https://imgur.com/a/pqpNqHO

This was printed in PLA- normally I print your stuff in PETG but PLA prints faster. I'm also on a MK3, I wonder if retraction settings have a lot to do with it. I think it's pretty difficult to do this operation, though it's the best spring design yet. Perhaps some of the "skeleton" you used to separate the planes in early RCs would help?

That's odd, I printed a lot of these springs in PLA on a MK3 with stock settings, no issues in separation by bare hands.
Skeleton/spiderweb: another user reported bridging issues. Also, the use of a spiderweb results in a bigger gap and an even smaller spring.
Did you consider breaking the central hub free by inserting an axle and wiggling it?

I was thinking of the spiderweb only on the axle and the inner ring, not across the full area of the springs. With the broken springs that I printed earlier, which were in PETG, I tried breaking the axle loose but was trying to do it assuming the axle needed to be split from the base.

Maybe it's just me that has these problems. We'll see.

In my experience it is very easy to break the central shaft loose by inserting an axle (not deeper than 8 mm) and wiggling it. But not by inserting a screwdriver all way through of course. I will make specific instructions for this later.

Thanks for the pics! Glad you managed to properly break it free now. Do you consider it "too hard" to be included in the next release? I could try further reducing the contact area of the split surfaces, but I am afraid that it might get loose during printing. A spring printed in PLA is already quite loose. Even when printed in PETG (on my Prusa i3 MK3) I do not need any tools to break the parts free. Of course every one's printer has different characteristics. So...?

Thanks for the schematic. I think that's what I did wrong- I assumed it was supposed to free up at entirely different points. I'm printing RC5 and will report back.

Good to hear that the schematic helped!

Thanks for sharing! Also thanks for the last remark.

It broke when I tried flexing it to break the spring and especially the hub.

Flex it to break the hub? This is not possible, the bottom part of the hub should connected to the base. I added a schematic drawing (see instructions) to make it more clear. I don't know how far you tried to flex it, if bent too far it will eventually break but I will consider making the splits in the base a bit shorter.

How exactly did you break the base? If you post a picture then I know which part needs to be reinforced. Which material did you use?
Meanwhile I uploaded RC5 with stronger connection between base and centre hub and easier splitting of the hub.
RC1 is a totally different design with two springs in parallel, while this works fine it does not have the special advantages of the serially connected springs in later experimental versions.

The spring itself should be split mid-plane!

Ahhh now we are talking. I diddn't saw that, the PETG was pretty well bonded together and I just thought, this tiny "gap" was a extruding-failure. Now I see how this works. I wish, thingiverse has a per-file-description.

So, I've spent wayyyy too much time designing/building an enclosure that I feel may be what everyone here is itching for with this parametric spool holder. I'm in the testing phase of it now and all is going well. I should have all the parts/pieces up on Thingiverse in a week or so. Crossed fingers.

BTW: I'm not sure why you have a cylindrical spool nut. I could imagine just a cone on the (as printed) "lower" end of the threaded shaft Unless I wanted to center the spool, there's not a lot of reason not to just have the cone at the end.... I printed two of the cone nuts and just ran one "permanently" down to the end.

I've had some issues with the clutch popping out a bit. With a more robust mounting, I've had fewer problems, for as little sense as that makes. :-)

Indeed, it is a good idea to have two cone nuts for optimal centering. The cylindrical nut can be useful to lock the outer curved nut in place. Just keep the goal in mind: rigidly attach the spool to the hub, well-centered, and use the nuts in the way you feel it is best to achieve the goal.

Alternate to your stands for using less plastic and fixing to surface: https://www.thingiverse.com/thing:3474307

Thanks for the spools, they are pretty cool!

AutoRewind Spool Stand Hanging Screw Mount
by AbeFM

Thanks!! Posted a link on the main page.

All FDM 3D printing filaments are hygroscopic. That’s a fancy way of saying that the material likes to absorb moisture.

This hygroscopicity can wreak havoc on 3D printing filaments. As plastic polymers, they’re made of chains of molecules strung together. Moisture introduces water molecules that break up these chains, ruining the plastic and causing a whole slew of problems while printing.
Source: https://all3dp.com/2/how-to-dry-filament-pla-abs-and-nylon/

Food dehydrators are often used for the purpose of drying filament. See: https://www.thingiverse.com/thing:1618037 and https://www.thingiverse.com/thing:1117970

The difference between "Dehydration" = "Dehumidifying" can be explained as follows:

Dehumidifying: The process of drying air using a device (air-conditioner) or desiccant material like Silica-Gel. This is done at a constant air temperature.

Dehydration simply means removing liquid by using a blast of air to cause removal of liquid from the part being dried. Evaporation of liquid is enhanced through the use of heat and the movement of air over the parts.
See: https://techblog.ctgclean.com/2013/05/drying-the-effect-of-temperature-on-relative-humidity

Filament Dryer / Dehydrator
Three Spool Filament Dryer (Dehydrator)

In the post printing image, you have multi-color parts laid out on the build plate, but there's no wipe tower. Is there really a way to print it with that arrangement?

Yes! There IS a wipe tower ;)

OK - I see it now. It's really tiny!

You also understand why? :)

Well, the two color portions are all at the bottom, which explains the height, but I don't know how you got the Y so small...

Hint: Wipe to Object

Ahh yes, I see hints of blue in the spring. Is it really ok to print the spring in PLA?

Apparently yes, as long as you don't dial too high. The 1.02_RC2 versions should perform better though.

This design is brilliant! We have three in operation and the last two stands are printing.

Has anyone come up with a humidity controlled enclosure that works with this auto-rewind spool holder?

We found a remix for using the old version of the spool holder with a Rubbermaid container (very cool), but this new version of the spool holder is too wide to fit.

 - Brad

I was thinking of modifying the spannerhands dry box to accommodate this. It shouldn't be too much work.

Hello Brad,

I took some time to produce some drawings to explain an idea i have for exactly that where you are asking for. Please have a look at the thing I published. https://www.thingiverse.com/thing:3455378

IKEA LACK for Prusa MK3 with MMU2 and 5 spool Dehydrator

Thank you, nice to hear! I don't know about suitable remixes for dry boxes. The Rubbermaid question was asked earlier and I suggested a workaround by scaling the relevant parts along certain axes. It should work although the spring would be a bit weak. I don't know if the experimental spring versions (v1.02_RC2) can be scaled in the same way, but it might be worth a try.

Hey Vincent,

I took some time to produce some drawings to explain an idea i have for exactly that where you are asking for. Please have a look at the thing I published. https://www.thingiverse.com/thing:3455378

IKEA LACK for Prusa MK3 with MMU2 and 5 spool Dehydrator


That is a cool "Lack Stack." I really like how the filament tray slides out. The best builders know how to design before they build!

The dehydrator is an clever idea - haven't seen that before. My dad used to dehydrate fruit when I was a kid. Our dehydrator seemed to simply circulate hot air around the fruit and to dry it. I wonder if the concentrated hot air will be compatible with 3D printer filament. In other words, does "dehydrator" = "dehumidifier". Interested to hear how it works out.

Hunting around, I found another solution that uses the IKEA Metod cabinet.


Wondering if a dehumidifier like this one on Amazon would work well inside the case.


I printed the SpringFlat_Stiff_v1.02_RC2.STL, in PETG. 0.2mm layers. After looking at where the separation lines are, it's very difficult to get the sprints to separate from each other... RC1, had that web, which made it easier to separate.

While trying to separate the two springs, I ended up snapping the whole spring... I could be the only one... but seeing how RC1 was much easier... RC2 being difficult to get the springs to separate from each other, there just isn't enough room to work with.

First, thanks for sharing your valuable feedback! It is unfortunate that you have troubles in separating the spring parts. Another user had troubles with the bridging in RC1, that's why I made a version that required no bridging. I'll think about creating a version for RC2 with spiderweb support to fix your issue, but it has the drawback that we end up with more and more different versions of the same spring (flat&round, soft&stiff and now also with&without spiderweb support). A different approach might be to print the spring in separate parts and clip them togheter. I will think about it. If you have suggestions, let me know. Meanwhile you could try printing the RC2 spring in PLA, this is easier to separate. Or tweak your PETG printing settings around the interface layers.

Printed the last version yesterday...first thought was that the whole bottom plate and the side only are there for printing reasons and should be removed.. So I bent and cracked a little and the spring got loose...but the center part is still fastened to the bottom plate... Should I just use more force...or is the bottom plate and sides a part of design and supposed to stay in place?

All are part of the design. You can check the layers in your slicer software to find the gaps that are supposed to separate. Hope this helps!

So the center of the spring should remain fastened to the bottomplate while in use? Everything goes in there..? :-)

You are 50% correct! I know it is complicated, but I like to make complicated things. The "spring" in fact consists of two separate springs put in series, connected together at the outside (the next-to-outermost rim). See if you can find the gap to split the spring in two. As for your original qustion: the bottom 50% of the centre of the spring remains fastened to the bottomplate, but the top 50% should be set free.

Good thing to have your own 3D-printer when trying to break apart this..I can just print another one..

I´ll give it a try!!

Having said that...I really like your work! Thank´s a lot for sharing,,and having the patience with us while answering endless question...

It worked!! :-) I managed to separate it exactly as it was meant!! :-)

Yay! Hope it also works as intended! The clip may be too tight, in that case just leave the clip out to reduce friction.

Hi Vincent,
thanks for your work...
Is there a recommendation which spring to use with PLA - stiff or soft?

Thanks, Henrik

I'd try soft first, keep friction to a minimum and dial low to reduce tension on the spring.

I will test that - thanks for the quick response...

Hi Vincent,
I'm still printing parts for these but need to organise the bearings.

You mention ZZ type but in your pictures, looks like RS.

I understand ZZ type will spin easier so assume they are the best ones to get?

Both are fine! I am using the 608 bearings provided by Prusa.

Thanks Vincent.

First spool holder is now complete and assembled. I will be using these with my MK3 with MMU2 and currently have the spools on top of a cabinet with the festo mod in relation to feed path.

So I'm thinking I'll need to run an additional length of PTFE tube from the auto rewind spool to the end of the PTFE tube from the MMU. I can print out a couple of joiners; I'd rather not have to replace the existing PTFE tubes from the MMU with longer ones so am hoping the use of joiners would work.

In relation to the tube holder, is it just the PTFE tube that keeps it in place as I notice it falls from side to side if trying it in the upright (top of spool) position.

Thanks for your help with all my questions.

Yes, the PTFE tube keeps the tube holder in place. This makes it easier to reposition the spool holder if needed.

I like the design and the idea of your spool holder but I had an issue yesterday while tightening a spool on the hub. I tightened the nut and after some minutes the hub broke. As you can see in the attached picture the material is very thin between the thread and the plate. I think it is not necessary to have a thread until the end. In my opinion, it might be better the stop with the thread 5 to 10mm before the plate and chamfer with a 5 to 10mm radius.

Yes, those walls are quite thin. The thinnest side is approx. 1.6 mm, which still fits four to five standard perimeters. In theory the cross-sectional area and strength should be comparable to that of a solid 30 mm cylinder with three perimeters and maybe 5% infill. Most users (including me) have no problems with tightening spools by hand as tight as possible, so your issue might be related to suboptimal layer bonding. You are not the first one to report breakdown of the hub on that location though, so I will think about adding a reinfoced hub besides the standard one in v1.02. One user already published a remix which includes a reinforced hub: https://www.thingiverse.com/thing:3425043

Remixed spoolholder

Unfortunately I have a broken spring V1.02_RC1. I would say total printing time was 20-25 hours, I had it set at 10 on the dial. I printed the spring from Amazon PLA, I'm thinking next I should try to print in PETG?

Oops! Thanks a lot for sharing these pictures! So let's see, the spring is broken at two locations, first at the outer rim and secondly near the axle shaft. I see two possible causes: 1) the clutch was operating normally but the stresses were too high, or 2) the clutch jammed.

In case of 1) it definitely helps to print the spring in PETG as that material is stronger under stress. On the other hand, I have several very old PLA springs from the small original rewinder (diameter 50 mm) which are still operating fine. So if 1) was really the cause then the dial might have been set a bit too high.

In case of 2) we have to figure out how exactly the clutch jammed. I know that in the v1.0 and v1.01 versions of the spring the clutch can jam if the spring is wound around the axle and pulled too tight, while it should not happen when wound towards the outer rim. Another possibility is that the spring caught a piece of the dial, blocking further movement of the spring, but if it had been operating fine for 20 hours then it seems unlikely to me.

So my conclusion is: yes, try PETG, and set the dial a bit lower than normal (aim for 1.5 rewinding revolutions). Meanwhile we hope to receive feedback from other users as well to see how the spring design can be improved in the existing universal rewinder.

Here are my pics from my first 1.02 version spring showing the support. PETG.

I am currently printing 2 springs in PETG now, I will report back on how those springs work later this week. I was aiming for 2 rewinding revolutions to ensure it pulled all of the slack back from the MMU in order to minimize the possibility of tangles, but I will shoot for 1.5 going forward.

A question about case 2, if the spring jammed, wouldn't that cause the spool to temporarily lock up and then the entire filament stand would be pulled towards the printer until the spring fractured?

I have a feeling PETG will help the issue as they were working fantastic until the fracture, I love the design and functionality.

Are you using the tube holders? If yes then the stand cannot be pulled towards the printer in case of a spring/clutch jam (except if the tube holder itself got loose first). But I hope your PETG spring will work well, especially when dialed not too high.

Meanwhile I modified the instructions, which now mentions to aim for a minimum of 1.5 revolutions (instead 2) plus a warning not to dial it too high.

Yes, I am using the PTFE tube holder, and that makes sense what you've said.

The first 2 PETG springs did not turn out, the bottom half of the springs are beautiful, however the spider web support caused bridging issues on the upper halves of the springs.

I'm trying to print another with the fan turned on full right after layer 18 when the bridging needs to start.

I'll keep you posted.

Just added a new spring (1.02 RC2) which does not need supports or bridging and also performs much better. I printed it in PETG and it turns out fine. Might be worth a try.

Great, I’ll give it a try soon. Thanks!

I finished the first one.
Great construction. Thanks a lot for that.
I have one question though:
How is the tubeholder fixed to the stand?

it just goes onto the axle and is placed on when placing the spool.

Hi Vincent,

Just printing the new spring design now.

Is there an easy way of removing the divider or is it just a matter of going around the spring with a hobby knife to remove it?

Should be easy, no knife needed. Just bend the spring up&down and most part of the divider should pop off.

Thanks Vincent. It was easy to remove - I didn't realise the spring was in 2 sections.

I have a couple of minor imperfections in the print but I'm hoping it will be ok. I used Prusa PETG filament.

Those "imperfections" are normal, the divider is quite minimalistic and not fully supporting the top spring. It should not affect the functionality, but if it does then it should be easy to smoothen it.

I'll print the new spring in the next few days. Hooray!

Vincent, do you have a way that you notify people about designs or updates to designs?

You could try the "Watch" button on this page!

Is PLA OK for all the parts?

Short answer: yes. Long answer: the new spring (1.02_RC1) has not been extensively tested yet as it was uploaded just a few days ago, so I'm not sure. At least the new spring is symmetrical, so it is possible to flip the system around without issues.

Vincent, is the divider in the middle of the new spring supposed to be removed entirely, or just broken free?

The divider acts as support during printing and must be removed.

I printed spring out and getting support out makes spring split in two , is that the right outcome ?

The axle shaft and the outer rim should still be in one piece, connected together by two different springs. So yes the "spring" (actually two different springs) are split, but being connected to axle shaft and rim it should still be a single physical body.

Hi, just want to check that the bottom spring of 1.02_rc2 does not need to be connected to the base? When I was removing the support, the spring split (which is good), but the bottom spring closer to the base was also removed from said base. They're still both connected to the axle shaft and outside rim.

Correct! 1.02_RC2 consists of two springs connected in series. This way you have more rewinding rotations without stressing the material too much. Hope it works!

This design is great except for one little issue, which is that sometimes the filament will come off the spool, especially when the spool is new. I assume it happens when the printer unloads the filament right after the spring slides.
There are no bends in the PTFE but I think I was able to mitigate the issue was to rotate the PTFE holder back so that the filament was entering the tube upward and behind (instead of on top).
It would be nice to have a rail that essentially allowed any loose filament to stay contained within the bounds of the spool.

Thank you for the feedback! I fully acknowledge your issue. If you (or someone else) designs a custom stand with rail as you suggested, then I'll post a link to it here. I myself am currently focusing on the spring following multiple reports of broken springs, so there is a new design (v1.02_RC1) in the files section. This new spring design possibly allows for some extra tension without breaking stuff which might actually be sufficient to solve your issue.

This design should solve that problem:

You can see it in combination with your auto-rewind spool holder:

Prusa i3 MK2/S/3 Filament Guide ClipOn
by Area51
Prusa i3 Mk3 top stands for Universal Auto-Rewind Spool Holder by VincentGroenhuis
by Racx

I ran into this as well. The trick for me (on the new springs) was to carefully adjust the tension so that when it slips, it only slips a small amount and doesn't completely unwind. One click can make all the difference.

I find that I need a higher setting (12-13) with the new springs.

I have lots of trouble getting these to work reliably. The adjustments on the springs don't seem to make much difference. The slack falls off the spool then tightens around the axle. I'm using the bearings and I've tried adjustments from 0 to the max. Not sure what I could be doing wrong. Any help is appreciated.

I have lots of trouble getting these to work reliably. The adjustments on the springs don't seem to make much difference. The slack falls off the spool then tightens around the axle. I'm using the bearings and I've tried adjustments from 0 to the max. Not sure what I could be doing wrong. Any help is appreciated.

Do you have a photo of your setup? Sharp bends to the left and right are one possible cause of the problems. Also, did you pre-tension the springs?
I don't have a video of the universal rewinder design yet, but there is one of the original design. Did you do the experiment: https://youtu.be/1Cki6mBVA8s?t=28 ? If yes, how many rewinding revolutions do you get? Does it change at different settings of the dial?

How do you pretension the springs with this design. I am finally using this design on 2 spools but one of the things I really liked about the old design was that I could hold the spool and twist the rewinder to pre-load the tension.

What I do is: put the loose end of the filament through a hole in the rim of the spool and then twist the rewinder (with spool) a few times before feeding the filament in the printer. An alternative way is to first install the spool+filament, next lift the axle from the slot, rotate the axle two turns and put it back in the slot.

i have to join the band of broken springs, it has worked fine and i only needed a tension setting of 7'ish

not sure where it broke but i will try and get it apart while printing and then post a pic

Could you please try one of these? It is a new design which hopefully solves the issue.

will do, it will be the round axle. but it will not be before tomorrow i can print them

pic attached

The new design should prevent this type of cracks.

1.02 RC2 is available in the files section, it is now a serial double spring with good performance.

Hi Vincent, just sharing my experience so far.
I have broken a couple of springs. what i noticed is that for some reason the spring get stuck and then the obvious result is that it fails. i tried to figureout why, but it's not clear to me so far. i tried both wounding in inner and outer direction, but still there is something missing.
I'm also trying another path, using the universal hub with the parametric spring and then add a couple of spacer, similar to what Cabbey did, but both sides so that i can make sure the spool is free to move when the spring goes to the end. it's an hybrid of the two models you created.
if interested i will share my results after some testing.
thanks again

I will do and give you a feedback.
thanks a lot

I see. Winding the spring in inner direction results in imbalanced forces and torques on the outer rim which then may get stuck if dialed too high. Winding in outer direction results in high stresses around the axle. I tried to mitigate this by making the spring thicker and by adding reinforcements around the axle, but apparently this is not enough in your case. Maybe grinding the axle, spring and dial might help so that the parts slide more easily; you could also try leaving out the clip to get it running more smoothly. We see from the comments that you are not the only one with this issue. I'd like to have the system fully working for everyone, but apart from some very minor tweaks it is a bit difficult to find the best way forward within the current design philosophy: many ideas require a lot of work and/or a total design overhaul. Your approach to use the 100mm parametric spring combined with a self-constructed slipping element is a good one, I myself tried it as well a week or two ago. Two curved nuts plus one flat nut make for a suitable wedge-shaped space with sufficient friction for the (free-rolling) spool. Make sure that the nuts are well tightened on the hub, otherwise it might again cause jamming and destruction of the parametric spring.

With this design, am I right that the Nut (curved or otherwise) that supports the spool only goes on the side away from the spring? What is supporting the spool on the spring side, just the pressure that the nut is providing?

You can put a nut on the other side as well. A few of my spools work great with a curved nut spun on flat side against the back of the clutch, then another curved nut facing the opposite direction from the far side. However that did not work well for several of my spools, so I've modeled out a set of spacer discs that go inside to support the weight of the spool.

(Note I'm not affiliated with @VincentGroenhuis or the University of Twente or RAM... just a happy user of these spools.)

Spool Hub Spacer
by cabbey

That is exactly what I was expecting would be needed.

On a side note, one thing I always had problems with was when changing a roll, if you rewind and accidental let go of the filament, it will loosen up and can get wound over itself when re tightening. I was envisioning a system that clamped the filament and stopped it flying loose, but don't know how it would work precisely.

There's a relatively cheap and easy solution to that: small binder clips from any office store. :) As soon as the filament comes out of the PTFE tube, it is immediately clipped to the side of the spool. The binder clip doesn't come off again until the spool is reinstalled in the holder and you're ready to insert the end into the PTFE. At which point the clip goes into a small dish of them near the spools... ready to go on the next spool to come out.

Good idea! I usually use the holes along the rim of the spools and have to work around the pieces of filament sticking out. The binder clips can also be printed of course.

Note to self:
Always have it so that the spring is tightening and not loosening. One broken spring later and I now know.

You mean the spring should be wound around the axle and not towards the outer rim? This makes sense as the torques are highest in the center. I read more reports about broken springs, so I guess I may have to adjust the instructions accordingly. The drawback is that winding the spring around the axle causes unbalanced forces on the outer rim which is then deformed to an elliptical shape. Setting the dial too high then leads to a lock-up with again a distaster as result.

I think it makes sense. I think my spring must have snagged somewhere but the extruder has enough power to keep the spool turning which meant that the spring only had one thing to do and it started doubling back on itself. Sorry not explaining it too well. Perhaps the tensioner could also be reversed not that I’ve noticed the tension setting changing. I will keep an eye on it.

Summed up experience's:

i can't dial to 0 nor can i get it to the max mark, the lowest i can get is 2 and the max 13

more number markings would be nice in case the ones you see point to where you cant see them

bearings are a perfect fit, no need for a clip to hold them in place

hub thread and nut thread was a VERY tight fit, PTFE grease from superlube helped

I have checked that my extruder feeds precise 100mm when told so
A single wall cube of 20mm in all directions messure precise 20mm and the wall is precise 0.4mm thick, same width as nozzle

The stop block on the spring is held in place by the friction of the nut on the 8mm threaded rod i use, there is another nut in the other end

in my case it seems that a tension setting right in the middle does the job fine for me

for me the scale is perfect and the threads are really nice. Absolutely no grease required. Maybe you need to check your printer settings.

there is only so many times that you can check and recheck settings before it would be called insanity

All I can say is that the threads are some of the best threads I've seen, the "nut" just spins straight onto the thread. I've printed 3 of these things now and they have all been perfect.

Good luck finding the answer.

I have the exact same experience..

i can say that i have tried a single wall cube and the wall comes out precise 0.4mm thick as they should
i have made sure extruder feeds precisely 100mm when asked to

i have checked everythig to the "fly fu**'ing" level as we say here in DK

and if you read some of my post i kind of hinted that i will replace the cooling fan to a 5020 that should push a lot more air than the current 5010


Try printing a 3DBenchy. Lots of overhangs to challenge your printer. Good chance that the result won't be as smooth as standard Prusa ones with its huge radial fan blowers.

hmmm i looked arround to see what prusa uses on the mk3

seems like they use a 75xx fan

i have a 7515 fan but the duct i use needs some modding and i'm not that good with 3d modelling, even thou my extruder has a step file for each of the fan duct options

i'm told its easy to mod a step file, but my brain do not agree at all

yep, allready done and what i found is that i can barely get by with 0.2mm layers, anything less and the overhangs suffer, the thinner the layers the more they suffer and i get a tiny amount of stringing

so yes the radial blower i have (5010) is not good enough so i'm hunting down a 5020 instead, and my extruder mount allready have a fan duct for that, i'm using this one for extruder mount: https://www.thingiverse.com/thing:2065461

Prusa i3 MK2 Upgrade kit for E3D Titan Extruder by S±E

Thank you for posting your make and it is good to hear that it works! So locking the spring to the axle is easier than I thought, with just a pair of nuts and nothing else.

At the moment I don't know exactly why your hub and nut are so tight. One possibility is the steep slope of the overhangs (60 deg) in the threads. If your printer has suboptimal part cooling then it causes these overhangs to sag a tiny little bit. Setting fan to max and some overhang printing optimizations might help here.

A workaround is to scale the nuts to 101-102% in X and Y directions, adding 0.15 mm of additional horizontal clearance all way around for each percent increase.

the ziptie can do the trick at low tension settings but will not hold the spring at higher settings, so settings are lowered

not so much rewind pull but i think it will do the trick

so if anything a version of the round spring with a small hole in the stop block, or a small dent so the drill bit will not walk away

well the fan is allready running at 100% so if anything the fan i have is just not blowing enough, its a 5010 and i do in fact plan to upgrade to a 5020 instead

i spoke to fast on fixing the sprint to the axle, the nuts compress the spring to the dial/clutch part and hence the whole thing rotates... i will try with a simple zip tie and if that does not do the trick i will drill a 2.5mm hole and use a 3mm bolt as a set screw

Prusa uses a radial blower fan which is much more powerful than a 5010, this could indeed make the difference in overhang quality.
Fixing the spring to the bolt: I think the best way is to make a part which interlocks with both the spring and the nut. Some solid object with 8mm hole through-all for the axle, a hexagonal cutout for the nut and a rectangular cutout for the stop block. Fixing the nut to the rod is the last step which is very easy.

give me a few days to thinker with the idea... for now the zip tie at low tension does the trick

not elegant but works :-D

yeah, it has to be thought through so that you dont end up pushing the pring towards the hub, it has to be able to rotate as freely as if i held the stop block with my finger, but your idea is food for thought, i might end up doing something similar

maybe a metal nut in the other end and a solid block with a cutout for the stop block and i think instead of a nut i will use a set screw of some sorts,

Sorry Vincent, I don't know WTF I'm doing wrong. The filament on the fuller spools is going everywhere except back onto the reel, it's just not rewinding sufficiently. Am I supposed to have it tighter (higher number) on full reels, or looser (low number)? I've tried both & neither seem to make much of a difference. At least the PETG springs haven't broken yet.

The system might work for a little while, then I come back to it & filament is everywhere! Any assistance would be great, thx.

Thanks guys, I understand what you're talking about Vincent, but as you can see, I'm rather limited for space.

I will experiment with different filament placement & go from there. Thanks again!

It is not your fault, it is because of the sharp 90 degree turn to the left which causes the filament to twist (to follow the bend). During rewinding the twisted filament is strong enough to push itself off the spool as you see. One solution is to rearrange the spool setup so that the paths are straight and you only have up/down bends and very little sideways bends. Another solution is to install very long filament tubes, so that these curve very gently. You could also experiment with stronger springs but I don't know if it would work for you.

All I did was move the spools to make an 'arc' around the MMU so the PTFE tubes are straight(ish) & so far so good. I can't believe something so simple can make such a difference, it makes so much sense now. Thank you!

also the direction that the spool rotates has a say in how it should be setup, look at end of instructions

have you tensioned the spring? when the printer pulls on the filament it tensions the spring, but if that is not done enough there is not enough tension to pull it back

Thankyou for a great design & its working well.
I have a problem with the spools size as my new filament have wider spool openings. Can you create larger nuts or advise how to customize the nuts. The new spools i have a centre width of 73mm & the curved nut is cracking so not holding the spool securely.

A curved XL nut is provided in the files section which easily fits your 73mm spool!

in the description you find a link where you can put in your own numbers

takes 10 sec to find :-D


Parametric Auto-Rewind Spool Holder

I had already checked that link & as Vincent points out there is no customiser for the nut. Check your facts :-D

There is no customizer for the nut in this link...

Thankyou Vincent....
i missed that as i was using the gcodes you had supplied & had not downloaded all the files from this post.

Hi, has anyone made a Creality Ender 3 stand for this?


Hey Vincent

This is amazing. Works perfect! Printed mmu in black and white petg and it looks pro! Thanks for sharing this.


Thank you! I read that the universal rewinder is not a perfect solution yet as some users report issues like broken springs etc, so it is good to hear that it does work in most cases like yours!

Hey Vincent,

I am using five of this spoolholders and they work great! I've printed them in PLA+ and the springs in REAL PETG but today to of the springs snapped right in the center where the first rotation would start. What infill do you use to prevent this?


Did you check the functioning of the clutch? it should start slipping after about 2-3 rotations and not much higher, otherwise the material stresses get too high. You could also try reversing the rotation direction (and recalibrate) to let the spring rewind around the axle instead, just to try.

Pity that two springs broke down! I use three perimeters and 10% infill. You are not the only one with this issue. will look into it.

i would say that the infill does not matter much since most of the spring are perimiters anyway

maybe make the spring thicker? not sure if that would work well, but just a thought

Ciao Vincent, i'm trying to reproduce the thread in fusion 360 to create a different Nut. I'm struggling to find the right thread type. would be possible for you to share the exact hub diameter and thread type ? thanks again for sharing this. ciao

Of course! 30 mm diameter, 4 mm pitch, ISO type thread.

thanks. looks like Fusion doesn't have this option. max is M30x3.5. what do you use to model it ?
thanks again

If you model the part at half height, then thread a hole with M30x2, then apply a 2x resize on the axis the hole follows you'll get M30x4. I did that last night and left a part to print, this morning it spun on perfectly. (I'll be uploading a "remix" with a variety of spool centering rings sometime today I hope.)

Edit to note VincentGroenhuis points out this will put the bearing surfaces quite out of whack, see thread below for better numbers to plug into the above idea.

Good idea! The perfectionist might say that the angles as defined by the ISO thread profile (60 deg) are not preserved when scaling in Z only, but if it works then it works. A more accurate alternative (if needed) is to use M30x3.5 and scale Z by a factor 1.142857.

hmm... good point, I hadn't thought about the impact to the bearing surfaces last night, just did it in a way to make easy math. ;)

I just altered my drawing to use that approach (extrude at final height divided by 3.5/4, thread with M30x3.5, scale Z by 3.5/4) and the resulting threads still match the imported stl from the flat nut, so looks good!

Edit to add: they're live now

Spool Hub Spacer
by cabbey

I use Solidworks, create a 30 mm rod and make a 4 mm pitch ISO thread profile on it. It is not a typical thread size of course, hopefully you can find a workaround in Fusion!

thanks Cabbey, Vincent. i will try tonight!! ciao

Hey, i tried printing the spring but my printer wouldn't do it. I printed out the other parts that worked fine. But when printing the spring flat v1.01 he wouldn't go past the first layer. He kept repeating the same movements but he didn't go up a layer and did not printed out any fillament. So stuck in first layer. I imagine it is a slicer problem. I am using Cura 3.6.0. My printer is a Anet A6

also, the Spring is printing in an oval. I did some more research on this and found some answers but it still hapens. it apears that the model is just oval. since a other part i printed at the same time is perfect round

I'm having the exact same problem. Did you figure it out and, if yes, could you help me by telling us how? Thanks.

Try the RC5 design, this is easier to slice and works better than the original one.

That is very strange. Could you try a different slicer?

Do you have any slicer recommendations?

Slic3r, both stock and the Prusa Edition, slice the springs fine, just slowly. (On a 4 year old mac book pro it takes about 10 minutes to slice the spring and clutch.)

Do you have any slicer reconmondations?

I just installed Ultimaker Cura 3.6.0 and the spring slices just fine at default settings. You can check the projected output in the layer view, does it look circular or oval-shaped?

Just wanted to give you an update. I finally got all the parts printed & gave a MMU print a go. Everything was going great & then about 25% into the print, two of the springs (printed in PLA) broke around the centre axle. I am reprinting all the springs in PETG instead which should hopefully be much stronger.

The one's that broke had relatively full spools of filament on them if that makes any difference.

What was the setting of the dials? Do you think lowering them would have preserved the springs and still give sufficient rewinding revolutions?

IIRC, it was set at around 10. Any less than that & it wouldn't rewind much at all.

Apparently the centre of the axle has too high stress. What about flipping the stand so that the spring is in the other side and rewinds around the axle?

In your case I'd also do a filament strength test, to narrow down the search for possible causes.

you can replace the spring with a propper metal spring and you could a heck of alot more then 4 full rotations

not needed, the spring included is plenty good enough and you dont need 4 rewind rotations

What to do with all those extra rotations?!

@VincentGroenhuis , great job thanks. i'm using this on a different stand an i would like to use the dual knob also with 88mm spools, but i cannot due to the actual length of the hub. are you planning a 100mm length version ? thanks

The 3 kg spool stand thing has a longer hub (over 100 mm). You can also generate your own hubs with the customizer of the parametric rewinder thing.

Found!! thanks a lot. ciao

Hi Vincent, i printed the 107mm hub from the large spool, but the clutch and also the spring size , taken from the universal, don't fit.Both are too small in diameter. In the large spool i don't see those elements. am i missing something? thanks again

Never mind..my fault. i tried to assemble too early in the morning..) ciao

@VincentGroenhuis i wonder if this could be modded to work on other spoolholders also?

on my Mendel90 i have an 8mm threaded rod and i use this one: https://www.thingiverse.com/thing:767317

problem is that when the nozzle goes down to the bed it pulls filament off the spool fine, but after print i home the Z up to get it out of the way, but the spool does not rewind automatic and over time it becomes a mess

so i'm looking for something that can rewind a bit automatic but still be on a 8mm rod

EDIT: you said this 2 days ago: "All rewinding spool holders designs use the same axle size (8 mm) and accept both round metal axles and 3d printed flat axles."

so i guess i'm good, just wanted to be sure before i start printing out parts :-D

Parametric universal spool holder
by rowokii

Yes it should be possible. Print SpringRound and see if you can fix the stop block somehow. All other parts are used in the same way as the normal spool holder.

i guess the "stop block" is the part of the spring that is highest when printing?

are the parts as step files inside on of the zip's? if so i could pull them in F360 and figure how to solve it before print

1) yes, 2) no I did not include STEP files, only STL and .AMF. If you need a STEP file of SpringRound then I can export it for you.

think i got it

where the assembly instructions say insert axle that mens i slide in my 8mm metal/threaded rod

but what is the purpose of the clip? to hold in the bearing?

i think i have a few ideas on how to fix the stop block, make a 2.5mm hole in it and make threads for a small 3mm bolt

but i will thinker with the idea, i might also get a plastic storage box and just place the spool holder as you made them inside so i can keep it try

its funny how one idea can make other idea's pop up

at any rate if you are going to make new versions of this one that can sit on a metal rod by itself could be usefull, not for me alone but also for others

Clip: yes, it is to keep the bearing in place! Looking forward to see your solution. Others also combined this rewinder with storage boxes, see the parametric rewinder page.

will give this a go tomorrow, to be sure i need to print these parts right?

nut curved xl
spring round

thanks, will give it a go tomorrow :-D
i assume that i use to of the curved xl nuts? or does one side have to be the normal nut?

Depends on your spool dimensions and rotation direction (CCW as seen from nut/open side needs an additional locking nut). Just experiment what works best.

i got a tube of ptfe grease from superlube, that and carefully turned on the nut made it go fine

i guess there was a blob in the nut threaded that made it go tight

after half of the nut was on it got better and only at the end of the hub thread it got tight again

will check if i need a nut more and then i will write a combined post of my findings, but the clutch works but with limited range on the dial

no broken spring

so first curved nut xl printed, the hub is from https://www.thingiverse.com/thing:3369189 Hub_107mm_A7.stl

the nut goes on about 1 turn before it locks up completely

are there a way to "fiddle" with the parametric one so its just a fraction more loose?

Large stand for 3 kg spools on Universal Auto-Rewind Spool Holder

So got the clutch, dial, spring and hub printed

the only "bad" experience is that i can't dial to 0 nor can i get it to the max mark

the lowest i can get is 2 and the max 13

and i would agree with another that said it would be better with 2 set of number markings on the dial so the numbers can be more easy seen
if possible have 4 set of marking spaced evenly on the dial

the good thing is that the bearings are a perfect fit, no need for a clip to hold them in place

i will try and see if i cant picture it in my head first

Sounds all good! If nut is too tight: you may be overextruding or having difficulties in printing the steep overhangs of the thread (try increased cooling). You could try grinding down the threads with some coarse sand paper if you know the difficult spots.
If you cannot dial lower than 2 or higher than 13 then it is another sign that you may be overextruding. I'd print a calibration cube or equivalent and measure the actual dimensions as precisely as possible, this will also improve your future prints.

hehe, funnt you mention the cube thing, that is the first thing after i have done anything to the printer

or if i change filament or change slicer settings

i just hope that things work as they are

now i just need to figure how to fix it to the threaded rod, most likely i will drill a 2.5mm hole and tap threads for M3 and use a small 3mm bolt as a set screw

Having worked with these awesome spools for a few days of non-stop printing now, I'd like to suggest an edit for whenever you do a new version. Please replicate the tension indicator 2 or 3 times around the diameter of the clutch and the pointer around the dial. It seems like every time I need to adjust the tension setting the dial is on the opposite side of the spool from where I'm working. I think once or twice maybe it's been in the right place to see it.

In order to maintain symmetry I would have to replicate it four more times (five total). Not sure if I will do that, it would get a bit cluttered. Adding just one at opposite side would make assembly ambigious, also not sure if this is a good idea.

4 indicators spaced evenly on the dial would be ok, not sure why there should be 5?

Because of the 5 fins on the dial that insert into the clutches. In order for the indicator to work correctly you need the fin to be a the correct spot in the clutch when the indicator pin on the dial is aligned with the correct number on the clutch. If there are 5 then you can rotate it however you want, they'll all align correctly all the time. With only 1 pin it's also easy, only 1 of the 5 ways you can put the fins into the clutch will result in the indicator pin even on the number range on the clutch, making it very easy to see if it's correct before you snap things together. But if you have any other number of arrangements, then you have a 4 in 5 (80%) chance of not getting it aligned correctly and the indicator pointing to an incorrect number, or possibly not even being in a number range. With 4 indicators there's a very high likelihood of putting it on wrong. With 2 indicators there's a reasonable chance you'll be able to tell it's not right to begin with... but it would make the directions much more complex.

But the position of the fin is in relation to the number on the scale. It doesn’t matter how many scales you have, they will all read the same number and therefore the fin will be in the same place no matter which scale you set it with.

It's not the position of the 5 fins that engage the clutch that is the problem, it's the position of the 1 pointer that is. The ratio of pointers to fins being other than 1:1 or 1:5 is problematic.

I can't see how that makes any difference either. The pointer bears no relation to the tabs that engage the clutch. You could have 100 pointers if you wanted.

Actually the pointer has a very direct bearing on the tabs that engage the clutch... it is indicating what position in the clutch slot the tab is at.

100 is divisible by 5, yes, but the adjustment range of one dial is larger than 1/100th of a full circle. So you would end up with multiple pointers on the same dial. The technology may be advanced, but it is still no quantum mechanics... :)

of course.... did not think of that :-/ had a washing machine blow out smoke from the electronics so have had my mind another place

3 Pointed dial and clutch

Those are very good points. I didn't think about what having 3 indicators and 5 tabs would do to the assembly instructions. In the end I just used a fine point sharpie to transfer one set of markings 180º around the clutch plate. (well, more like 175º, I put the indicator in the flat bit between two of the bumps.)

I cant find the multi color stl's for use with mmu2 or palette2. Are you only releasing multi color gcodes only?

Check Parts_MM_v1.01.zip to get the multicolor STLs for each part.
Or use UniversalAutoRewinder_v1.01_MM.zip.amf , you should be able to open this in Slic3r (PE) and see the individual multicolor objects. Same for Stand_RewindTime_MM_v1.01.zip.amf . Does this work for you?

zip within a zip.....missed that sorry....

Just a quick question , is the tube holder suppose to be open ? I seen you did two version and there was a blue strip in the middle of fork ..

Is optional, it makes swapping spools easier if tube holder is installed under the spool.

Any tricks to get more tension , I load spring , but it doesn't want to backwind when MMU unloads, any Ideals ? With spool on it it will spin back and forth ..

Could you send a photo of your setup?

Thanks for uploading the pictures! Very easy now to spot the issues.
1) The tube holder causes problems if you do not use a tube at all. So leave the tube holder out.
2) Spring/clutch system works better if wound in the other direction. So rotate the setup 180 degrees. Do not set dial too high, otherwise the clutch cannot slip.

Just to clarify your point 2 here... the spring/clutch works best if the spring is pushing to the outer diameter when under tension?

I took tube off to see if that was causing a issue , not sure what you mean 180 deg (having it pull from top ?) Can you send picture on way you think it should be .. Thank you

I just mean, put the spring on the other side of the spool.

I just made assembly instructions a bit clearer on point 1.

Hi! This is a great design. Any chance of you doing a wall mount holder so I can bolt these to the wall?

There are a number of wall mount stands on the parametric rewinder page. Look at https://www.thingiverse.com/thing:3227387 , https://www.thingiverse.com/thing:3268015 and https://www.thingiverse.com/thing:3291542

Screw-down tabletop, wall, and ceiling mounted axle stands for MMU2 Auto Rewind Spool Holder
Hanging Holder for Parametric Auto-Rewind Spool Holder
MMU Spoolholder for Stuva / Besta

Oh Ok. Sorry about that. I wasn't sure if the axle/pin would fit these.

Which axle to use depends on the specific stand. Just check the designs, print and experiment. All rewinding spool holders designs use the same axle size (8 mm) and accept both round metal axles and 3d printed flat axles.

Thanks for helping us to solve space issue with MMU2, great job!

In your “Post-Printing” Summary, under Pict #3, you mention: “Insert spring, note the correct orientation (spring shaft pointing down)”

If you use a 608zz bearing as mentioned under Pict #4, there is no space for extra spring axe size?

I think we should have the same spring shaft height as the spring itself to fit in the space remaining when the bearing is inserted.

Am I wrong?


You have to turn the spring upside-down! Yes, the instructions were a bit unclear so I changed the last sentence to "Insert spring, note the correct orientation (spring shaft going through the dial)".


1) Excellent
2) Spool shaft needs strengthening have attached -- patch in slic3r
3) First layer, it would print easier if the numbers started on the next layer (if you have mediocre adhesion this whole part will fail) -- I was not able to patch in slic3r
4) A lot of thin parts that can break easily. Another place that could use reinforcement is the tabs that go into the clutch.

1) Thanks!
2) Could you upload it as a new model (remix), then I'll link to it from the main page! I cannot replace the original hub as some users use a curved nut adjacent to the hub backplate to keep a large spool centered.
3) You are right, I will consider your suggestion in a next revision.
4) Did those parts break off during actual use, or before/during assembly? Together with 2) I guess that your printer's layer bonding is suboptimal and could be improved (slower speed, more heat, lower fan, new/dry filament etc). You are not the first to report this issue, but in my printed parts the tabs are quite strong.

(2) I don't have an actual model, just a patch. I made a small torus (40mm tall 10mm thick, inner radius 22.5mm I believe) and then put a chamfer on it. I placed that as an extra object for the slicer to handle. It's not perfect, but it's practical. The space it patches is very thin, just a mm or two, it has broken on pretty much every model if I am not careful by either dropping the part, or either tightening the spool moderately.

(4) The tabs break when handling them during assembly, they are fine in actual use. But I have had several of them break. They are thin in the direction they do not receive stress, but they pop right off if they do. They are only about 1mm thick, even a bump of 0.5mm would improve strength substantially. (My understanding is that it varies as a cube, so 50% thicker is more than 3 times stronger).

Is it possible to export these as step? I can more likely make the edits you request then. I can't edit STL in my cad software (which is why I went for the cheap patch in slic3r). I see what you mean if you have a really heavy spool you'd want the thread to go all the way to the end. However, you have some other options to make sure that the bearing hole doesn't cut so close to the threading. (One is to make the thread even larger inner diameter). I'm sure you'd know better, since you've gotten this far with the design. Once again, kudos!

p.s. the stand is also a bit thin near the top and has been a breaking point as well. But it's easier to patch with glue successfully. I'm using new matter hackers filament, btw and default settings for my prusa printer. I haven't had these issues typically, but I can investigate on that end.

Attached two photos of the dial: one is straight and other is bent by pulling two tabs to each other with my fingers. So the question is why your tabs cannot handle such forces.

Hub_perimeters figure shows hub perimeters at the height where it breaks in your case. The thinnest parts is four perimeters wide and the thickest side six perimeters plus a lot of infill. In my prints it is so strong that it is impossible to break this part with my bare hands. I just tried to tighten a XL nut on a hub (without spool) with all the force of my hands and was unable to break the hub. (The nut is now so tight that I cannot even release it anymore...)

Also attached the hub as a .step file. Hope this helps!

Thank you very much Vincent. I will use a broken dial to test the flexibility when I am home, my guess is that the plastic I'm using is just standard PLA and is more brittle than yours, but let's test it. (PLA+ or PETG would be less brittle, but it's not what I'm using).

I have added a curved reinforcement that might also serve the purpose you were describing when you said, "I cannot replace the original hub as some users use a curved nut adjacent to the hub backplate to keep a large spool centered". Maybe the curve can be accentuated more, let me know (I'm sure you can reproduce the idea more accurately).

To be clear, I'm not breaking this by adding a nut to the shaft by itself, I am putting a spool on the shaft, then using the nut to tighten it to the shaft. Some of my spools have a very large inner D (especially near the edge), so this is creating a lever action when I am tightening it from the other side. (between the contact of the spool on the outer diameter of the flat part of the shaft, and the weak connection of the flat surface to the shaft). That's what I'm experiencing. Or, you could just take the shaft and try to bend it with your hand while holding the flat part.

(the 1mm and 8mm is just a measurement from the top of the ball bearing to the top of the reinforcement). Thanks again!

Just freed the XL nut from my hub using locking pliers, put a spool on it and tightened the XL nut again with all the force of my bare hands. The blades of the hub backplate deform significantly (see attached picture) but refuse to crack. I leave it tightened for another night to see if there are any structural deformations in the long term, but I guess it holds fine.

I am using standard PLA from different brands: Fillamentum, Colorfabb, Makerpoint and Prusa. Standard PLA profiles, no special measures such as filament drying etc. So I can only conclude that the strength of the designed parts is fine for the application, even under extreme loading. You should really be able to obtain the same strength if your filament, extruder and temperature setting are all as good as mine.

Hi Vincent!

No, I didn't mean to be confusing, I should explain: When I use some pretty plain PLA (the one I have used the most of is that matterhackers pink) it is just plain brittle. However, I have had time to ensure that it's not the printer or settings, the PLA is just brittle (they are also new rolls). BUT, I have lots of PLA here and most of it isn't just plain PLA, it's PLA+ or whatever you want to call it, and it's not as brittle. Any of these rolls work better for these parts. But my original feedback was based on using the more brittle variety and how the thinner structures fail under mild stress. I agree 100% that I am not getting the kind of material properties you have demonstrated, but I can get them by switching materials (which for the cap I have had to do since I cannot change the model to add reinforcement).

So, what I'm trying to say is that to use this model you have to have less brittle plastic, otherwise there are a few places where the model is simply too thin for the application.

But circling back, the design is really excellent and I have 8 of them working so far. We obviously have the same printers and are using the same settings etc, so it's just that PLA is not a homogenous material with the same abilities. I apologize if I was a bit confusing, and I also want to thank you for your feedback and also the step file.

That makes sense, I had not thought of the different strengths of PLA across the various brands before. Many people consider PLA for non-engineering, visual applications only and switch to "engineering plastics" like Nylon/Polycarbonate/PETG/ABS+/ToughPLA whenever significant material stresses are involved. Here we are challenging these basic principes by using plain PLA for engineering applications like the auto-rewinders and high-performance pneumatic motors. So that's how we end up in this situation :) :)

Next they'll be telling us PLA isn't suitable to get a man to the moon!

This is an absolutly fabulous design on so many levels. Thanks for sharing and the inspiration!
One question: The curved "Bolt" piece can be seen on the opposing end of the spring. i assume that the other end is just adjacent to the spring, but dont see this mentioned anywhere. Did I miss something?



This is an absolutly fabulous design on so many levels. Thanks for sharing and the inspiration!
One question: The curved "Bolt" piece can be seen on the opposing end of the spring. i assume that the other end is just adjacent to the spring, but dont see this mentioned anywhere. Did I miss something?



Happy to hear that you like it! Three different nuts are provided to mount the spool to the hub, and how exactly depends on the spool dimensions. In most cases a single curved nut is enough. The straight nut could e.g. be used to secure the curved nut in case the spool is unloaded in counter-clockwise direction (as seen from the nut).

I also use larger spools like Colorfabb 2.2 kg spools. Do you have a stand that works for those?


Yes, will publish it soon! (Tonight?)

Cant wait

Can't wait? Ok, here is it: https://www.thingiverse.com/thing:3369189

Large stand for 3 kg spools on Universal Auto-Rewind Spool Holder

I'm curious as to what type of material you used to print this? I'm thinking about using PETG, but if it works with PLA I would much rather use that. Thanks. =)

PLA works fine! Not sure about the long-term stiffness of the spring, so I also printed two in PETG.

Thanks. I didn't even think about just having spare springs on hand. Great idea!

Is anyone else having an issue where the retraction spooling is lagged or slow to the point that the MMU ejection overlaps it? Note this is a brand new 1kg spool, so there's very little space left on the spool, and it's heavy. Watching it while the MMU was loading and unloading it seems to behave correctly, the rewind was just late to start, to the point that it was probably one third ejected before the rewind motion started... by then the filament was already well outside the sides of the spool, and it just wrapped it around the clutch. I'm a little fuzzy on the clutch adjustment... it only seems to adjust how much it rewinds, not the speed of it, or when it starts. (I adjusted it while the filament was still attached to the side of the spool, not feeding into the tube... after spinning in the outfeed direction about 10 revolutions, it spins back about 2... so I think I have it set "correctly".)

had the same issue. but the filament is not very tight on the spool. i rewinded the spool 2 rotations when i install new filament. That works.

Yes I have and then the torque is too low. The default approach is to increase dial setting by a step or two. This increases the slipping and rewinding torque and should result in higher angular acceleration and quicker rewinding when filament is being unloaded.

Reversing the hub might also help. If spring is on other side and wound towards the rim then you can set the dial higher without locking up.

If this method does not work then you could try printing a spring from the parametric rewinder (bigger spring is an advantage here) and position three nuts such that the spool can slip on the curved nuts. Hope it helps!

I think I found an even easier solution. :) When first installing a new spool onto a holder, prime the spring by spinning the spool 5 or so rotations before unhooking the filament and feeding it into the tube. This pre-tensions the spring, allowing it to be ready to retract immediately.

My experimental test for this was just threading the filament in and doing the ptfe length calibration a few times. When the spool was installed and only enough filament to reach the printer had been pulled off, it was always slow to start re-winding. If between those two actions I spun the spool a few turns, then it rewound the filament perfectly. If I spun it too many times however, it risks the clutch releasing just at the worst possible time... in one case it was while I was stepping the filament down to properly align with the extruder gear. That one still rewound "ok", but only barely.

Good point to pre-load the spring when mounting it! I guess I should add it to the instructions. Also, it should rewind at least two times consistently, also right after slipping. (I just updated the instructions.)

More torque (i.e. higher dial setting) may also be needed if you have a sideways bend in the tube which may cause twisting of the filament, making it harder to rewind correctly.

First off this sounds awesome and I look forward to trying it. However there's a huge amount of files in this download. I'm not complaining I just have no idea what I'm actually supposed to print for this. Can anyone give me some info on that?

I think the only answer to this is: "It Depends."

Everyone will need:

  • 1x Hub
  • 1x Dial
  • 1x Clutch
  • 1x Stand (of some sort, either the one of those included here, or one of the many variations)

Almost everyone will need:

  • 1x Tube Holder. (if you have some other means of holding your ptfe tube ends near the spool, you can skip this.)

Depending on the spools of filament you print with, you need:

  • 1x Nut (any of the three. I'm using the curved nut on one type of spool and the curved XL on another... I suspect another type I have will be fine with the flat one.... and I may have to use a combination of a curved on one side and a flat on the other for another type.)

Depending on the direction of feed you need to spool things off the spool you may need:

  • 1x Flat Nut (in addition to the above; the flat one will be just fine for everyone, since it's acting as a lock nut behind the other shaped nut, not in contact with the spool.)

IF you have access to metal shafts, such as those shipped by Prusa with the MMU v2, then you will need:

  • 1x Round Spring

Otherwise, you will need:

  • 1x Axle
  • 1x Clip
  • 1x Flat Spring

Now, for every part from the above, decide if you want single colour or multi-colour. Note that you can mix and match them. I've printed just the clutch in MM mode for example, the rest are all single.

Take a look at the "UniversalAutoRewinder_v1.01.stl" for reference. It has all the files in a single STL. Though Here is a list of what you need:

  • Axle
  • Clip
  • Clutch
  • Dial
  • Hub
  • NutCurved
  • NutStraight
  • SpringFlat
  • Stand
  • TubeHolder
    You just need one of each. Just gonna also let you know that slic3r will take a while to slice the spring so expect a few minutes of slicing time!

Thanks! It took a bit of time to slice like you said but I have it printing. Hoping this will solve my MMU2 issues.

Thanks guys, that's a small but extremely useful piece of engineering. "Small" is great thing here, I'm an engineer myself and have seen too many very big and absolutely useless pieces of engineering :)
Made one just now. I don't have MMU, but even for single-spool setup it is great 'cause rolls don't get tangled.

"Small" is of course a relative term!I It was hard to arrange all parts (except stand) on a single buildplate. The universal rewinder design also has a few "small" limitations that the parametric rewinder doesn't have. Still, I am also quite happy with the result!

I honestly think putting all the parts on one build platter is a bad idea. There are two radically different type of parts in this project, and they benefit from different slicing settings.

How exactly do you mean, the settings of which parts should be changed?

IMHO there are basically two groups of parts here: those with threads or other curves not on the XY plane, and everything else.

In group one I put:

  • Hub
  • Nuts
  • Axle (I think, I'm not using it, but I would put it in this group if I were)

This group of parts will go best with a thinner layer height and slower extrusion, as well as some other tweaks for better supporting overhangs and threads printed along the Z axis, I used the following settings for them, this is basically Prusa's shipped "0.15mm optimal", with a couple things changed:

  • avoid crossing perimeters (change from prusa)
  • additional perimeters if needed (change from prusa)
  • 15% infill in 3D Honeycomb (this is highly flexible... i just like the properties of this one.)
  • octagram spiral top/bottom fill. (this is purely aesthetic, it just looks awesome on the nuts and hub.)

The majority of that are my standard "things with threads" profile.

In group two I put:

  • Spring
  • Clutch
  • Dial
  • Stand

This group of parts benefits from a thick layer height for added strength, and can enjoy faster printing as a result. For them I used effectively prusa's "0.35mm Fast" "0.20mm Normal" profile with just a couple changes:

  • 10% infill in 3d honeycomb or gyroid
  • Archimedean Spiral top/bottom fill for round things (spring, clutch, dial)

The clip and tube holder don't really matter... they can go in either.

For my first complete set, I used 3 platters:

  1. stand + tube holder
  2. spring + clutch + dial
  3. hub + flat nut + curved nut

For the other four, I did a few platters that I'm working my way through:

  1. 2x stand (printed twice)
  2. 2x hub + 2x flat nut + 2x holder (printed twice... this is a long platter, overnight, adding the tube holder didn't appreciably change that fact)
  3. 4x spring + 4x clutch (they are nested)
  4. 4x dial (I printed it in a "clear" pla so the spring shows through. :)
  5. 2x curved nut + 2x curved nut xl (printed twice)

EDIT a couple days later: After a few more days working with these spools, and a much closer look at a failed spring, I've noticed that there is a rather subtle curve to the spring that I had not seen originally. Printing with 0.20mm layers seems to reduce the binding up of the spring if you have it set to collect around the shaft instead of around the rim of the hub... I wouldn't quite go down to the 0.15mm I use for threads, but 0.35 no longer seems appropriate for the spring, and since the clutch nests beautifully with it, I've moved it up to that resolution as well. The stand can stay at 0.35... heck I've considered swapping to a big fat 0.6mm nozzle to print another one. :)

So that is why both pre-sliced gcode and individual parts are provided! I like your way to efficiently print a whole array of rewinders. Personally, I myself would not print the clutch and dial at 0.35 mm layer height due to aesthetics and readability of the numbers. The added strength advantage only applies if layer bonding is optimal; if by applying this specific setting a breakdown is actually prevented then I would fix the apparent weakness in the design anyway.

Interestingly, I've just had my first spring failure... on a brand new mechanism no less! (Note that this one was printed at 0.20 on a solo platter with just the spring and the clutch as I wanted a really nice one in a specific color that I only had a small amount of left... a show room model to post a picture of if you will.) :(

I had just loaded up a brand new spool, had the tension set at about 7 or 8... same as the one next to it with the same 1kg spool loaded on it. I started to pre-tension the clutch by slowly spinning the spool in the unwind direction (filament still clipped to the spool). After 7 rotations the clutch still hadn't slipped, so I adjusted the tension up to full, then down to zero... I expected the clutch to slip and the spring to let out when I got down to 0. It didn't. I gave it a couple more rotations and PING the spring broke right at the outside edge.

Adding a couple hours later: After taking it apart and looking at the remains, I have a theory what happened: I think one of the tabs from the dial was on the WRONG side of the clutch arm and just wedged the spring in tighter when I dialed down to zero. In other words: assembly error.

And two days later now, after yet another spring breaking... this time I'm certain the above theory was not the case... I now wonder if the coarseness of the layers on the curve in the spring is causing it to bind up as it wraps around itself on the axle. Which I've just seen a comment from you saying it works better the opposite way, I'll convert all of mine around tomorrow and see if the change is noticeable. In the mean time I'm moving to a 0.20mm layer for the replacement spring, as the original one I have printed with that size seems to be more free flowing.

That file says it was sliced for a Mk3... from my past experience it's not going to print well on my Mk2.5. Could you post the STL and print settings for it? Or reslice for mk2.5 if you have the profile installed in Slic3r.

If you don't want to publicly post the stl, you can email it to me... my nickname at my nickname dot net. ;)

What about this one? The previous gcode would never have worked anyway, missed a critical design aspect for round-axle springs... :)

No, the springs in cabbey2.gcode are too tight. This one should be better.

huh... so I ended up with a spring that has some internal spiderweb of support looking bits in the middle... er, no, two springs in opposition? the whole thing is stiff... no flex at all and I'm leary to turn the central shaft as it will snap the springs off these supports in a heartbeat. Or should I do that?

Actually, the more I look at this, the more I think those supports were supposed to be temporary... I see a pretty simple way to clear them out so I'm going to try that.

Yeah, that came right out like it was intended to once I popped the outer ring loose.

hmm @VincentGroenhuis as beautiful as it is, I'm not sure that design will work. At any tension level below 10 it immediately slips and no matter how many times it rotates to unspool filament, once it's released it gets about 1/2 a rotation of back spin. Between 10 and 13 it gets about 1 rotation of backspin, and between 13 and 15 it gets barely over one... probably less than 1.5 revolutions if you manage to time it just right to release just before it slips. Above 15 it didn't want to slip at all and I didn't want to force it. This is all just with a spool mounted and spinning it by hand ... no filament feeding out. Compared to the normal spring, where most of mine are set around 7 and get a solid 3+ rotations of backspin, at least 2.5 if it's just slipped.

Thanks @cabbey for printing and testing! I also printed it myself and had the same experience. The 1.4mm blade is too thick, it would require a bigger diameter spring to get 3 backspins. So I now reverted back to 1.1mm, while my earlier tests in this size work quite fine, I still like to make the spring unbreakable under normal conditions. The idea is that the spring winding around the axle prevents the other spring to break at the axle. So I also adjusted the spiral geometry again for the 1.1mm blade size and added other reinforcements. Attached two new STL files, feel free to test. Printing it myself now as well.

OK, the 3a spring is on it's way to print now, after a small tweak to temperature and flow rates in the gcode to match the adjustments I've made in slicer for my hotend and preferred filament. :)

Yet another version! This seems to work perfect here. 2+ backspins is easy, 3+ even possible without breaking the spring.

ok, I have A9d printing now... the double spring solution made for a noticeable increase in slicing time... but still not too bad. Once this one is done I need to swap my printer down to 0.25mm nozzle for a few weeks for a project, so this will likely be the last version I can test for a while... though I may print a spring at 0.25 just for fun to see how they work. :)

This is an amazing design, thank you for publishing it. I only have limited space available to put the spools and this is the best solution I've seen. I've tried a number of different designs & I wasn't really happy with any of them.

I've only done one so far & I used the single colour gcode - everything printed perfectly!

Might take me a few days to do the rest, lol.

Glad to hear that! Please don't forget to post a picture once you have five in a row spinning around :)

What infill is everyone using while printing this out? I bought some inland PETG from MicroCenter and the spool keeps on unraveling from the spool. Can I use this on a non-MM Prusa i3 MK3 just to keep on the filament from tangling?

I used the gcode provided to do the mechanical pieces & then sliced the stand myself as I wanted the blank one.

For the stand, I'm using 30% infill. Probably overkill, but I want it to be strong.

Yes you can use this rewinder with non-MM printers as well, as many users do.
I usually use 10% infill but it should not matter much, as long as your perimeters and top layers are good.

Thank you! I will attempt to print this out later today.

Thanks for this Design.
It works just great. i allready made 3 / 5.
only thing missing is a wallmount. I will build one when i find time for it.

Parametric Auto-Rewind Spool Holder
Screw-down tabletop, wall, and ceiling mounted axle stands for MMU2 Auto Rewind Spool Holder
Hanging Holder for Parametric Auto-Rewind Spool Holder
MMU Spoolholder for Stuva / Besta

Just a note for anyone else hoping to upcycle the rods from the MMUv1 spool holders... they're not long enough to work on this otherwise brilliant bit of kit.

One could theoretically make a shorter hub using the customizer (or scale Z...) to make those aluminium rods fit!

What's the point of this? Why would you want that it rewinds? I don't get it tbh.

The Prusa i3 multi-material unit unloads about 50 cm of filament during a color change which needs to be managed somehow to avoid filament tangling. In single-material printers it may also be useful if print head movements or retractions might cause the same kind of issues.

Wow Vincent, truly amazing! You must be enjoying the art of designing things. En dat alles van de hogeschool waar ik ook gestudeerd heb!

I am trying to customize and fit 4 in a Samla 22L box. Thinking of mounting 3 to 4 of these on one rod and then hang them into the box, much like what I have now, see: https://www.thingiverse.com/thing:3172330. Has that already been done? Does it make sense to put multiple at one real? I am only familiar to Fusion 360. I think the easiest way is to use the spools as you designed them and design a "screw on" system to a single rod capable of holding multiple of your spools. That basically means replacing the stand to some kind of axis hanging in the Samla box. Would that work? Or is there an easier solution?

Prusa MMU2 close to wall, filament storage and guide
by Ruedli

Yes: https://www.thingiverse.com/thing:3268855
It should be compatible with the universal auto-rewinder, but you have to adjust the length of the threaded hub using the customizer.
(Hogeschool?! Universiteit heet dat ;) Maar bedankt!)

MMU V2 Autorewind Spoolholder for IKEA Samla BOX

Thanks will look into that and will post what I (auto) wind up with ;-)
(wat betreft de Hogeschool, in de tijd dat deze bijna gepensioneerde maker daar studeerde, heette onze technische universiteiten nog hogescholen, vandaar ;-) )

This is amazing! Thank you sir!

wow that was intense. Took just over 3min to slice that spring, 100% cpu on 12 threads, max temp 78deg... That was impressive.

excited to see how this thing work.

Heh, you are not the first one! Further down in the comments there are a few more uses who reported excessive CPU loading. But your PC looks fast if it slices in 3 minutes!

Wow, third design in such a short time. And it gets better with each iteration. Love especially the adjustable slipping mechanism. This was a bit hit and miss on the older designs. And it is nice to have a single design, that fits many different spools.

That said, I'm a big fan of the Rubbermaid container solutions to protect the filament not only from moisture but from dust as well. One drawback in using these containers is their limited width which limits the length of the axle to 95 mm. In this new design the length of your axle has increased to 128 mm. Don't think I could shave off 33 mm and still have enough width to work with standard width spools. Both the spring with adjustable slipping mechanism and the nut need more space than the corresponding parts in your second design. Thus your second design is probably the better choice for Rubbermaid like container solutions. And version 2 is probably faster to switch spools.

But I'll definitely give your new design a try to see how it works.



We can see what is possible. What about scaling the spring, dial and clutch to 50% height? I'm curious also, so printing it now myself!
The straight nut has a 2 mm backplate which is the same as in the parametric design, so that does not change. Although you still need a spool centering thing to avoid balance issues.

Apart from that, every design has its own philosophy so the latest one is not necessarily the best. The 2nd also has a bigger spring by default and many useful adjustable parameters.

My office is under reconstruction and all printers are stored in boxes, so I can't do any prints currently

Took some measuments on the STLs and V3 could fit with a 95 mm axle the majority of my spools which are 52 mm in width.

Would be nice if you could make an axle with 95 mm length and a matching hub. With 5 mm on each side for the stand and roughly 15 mm for clutch/dial/spring (please check with your printed version), the hub should be about 70 mm in length.

Would give it a try as soon as my printers are set up again.



Hi Pete, the parametric version allows to create threaded hubs and suitable axles in any length, so you can first try to remix them yourself :) If you get stuck somewhere then let me know and I'll see how I can help.

Thank you. Wasn't aware you added hubThreaded to the SCAD file.

The slightly different shape of the generated threaded hub is just cosmetic?

The generated axle seems to be not compatible with your v3 stand. I couldn't find any parameter in the scad file to change the shape. But can remesh your v3 STL.

V1.11 of the parametric auto-rewinder now offers the ability to create vertical rims on the axle anti-rotation block to match the grooves in the frame. The different shape of the hub is indeed mostly cosmetic, but it also offers better grip when adjusting the dial and makes disassembly easier.

Nice. Your post overlapped with my reply above. So scaling down would be an alternative as well though cutouts for bearings won't fit perfectly on the hub then.

The hub backplate thickness is also parametric, so you can make it fit perfectly. In my case the 1.5 mm thickness got rounded down to 1 mm in OpenSCAD's customizer, so the editor has to be used instead.

Can somebody post a video of the clutch and rewinding action? :)
I don't have a use (yet) for these, but I'd love to see them working

The original auto-rewinder has a video compilation: https://www.youtube.com/watch?v=1Cki6mBVA8s
Maybe I'll make a video of the new one someday, but i have other plans first ;)

Fantastic job! I especially appreciate how you have managed to do more with the spoolholder using what looks like less material when compared to your last design for fixed spool IDs.

One question - do you foresee any trouble with the funtionality/ fit if I were to print this with a 0.6mm nozzle instead of the default 0.4mm nozzle?

Thanks! I think it should work fine when printed with a 0.6 mm nozzle, although I am not 100% sure. Just try it out and let me know if there are any issues.

Okay, I did a first try with 0.6mm nozzle at 0.35mm layer height. The result was not ideal: threads were tight, and the result looked a little messy. A 0.2mm layer height will probably work, but I'll go back to 0.4 mm nozzle/ 0.2 mm layer height as default.

Great design! I found that I have to use it in the same orientation as the pictures with the spring on the right and the spring unwinding for the clutch to slip correctly. If I flip it over it binds up.

Did you try to rotate the dial back to a low value after flipping it over? I myself found out that it works fine after the adjustment.

I did, but to be honest I didn't experiment too much after I saw how well it worked the other way. Do you think orientation affects the required settings?

Yes, for sure! The main reason is that when the spring is wound around the axle, there is an imbalance in the forces acting on the outer ring which then gets deformed. So that case needs to be dialed differently, otherwise it may bind up as you say.

Really good design. Fun to watch my PC beg for mercy slicing it. See the attached pic. :)

5:46 min slicing time on my system (spring only, 0.2mm), reaching 87 deg max. Didn't get a warning message though :)

i have the same CPU/GPU... but OC with 5 GHz, oversized watercooling and replaced the intel toothpaste with liquid Metal...
97% CPU.... for a while... but only 72 °C :-)
i found a new perfect prime95 :-)

Hehe. I hope no CPUs are actually destroyed by this thing..!

I didn't see the MMU version of the Stand_notext_1.0.STL stand, but I wanted the accent color without the text. I accomplished this in Slic3r PE by opening the settings for Stand_notext_1.0.STL and adding a generic slab. The slab is 1mm thick, with an initial Z height of 3.5mm. I believe that is the same, or close to the MMU text version.

Not sure if it is worth making the STL since you can already accomplish the same thing.

Nice that you take care for original details and found a way to include it yourself! Anyway, multicolor blank stand it is now included in v1.01.

Great model, thanks for sharing!

When I try to slice the model in Slic3r 1.42.0-Alpha2 nearly every model is reported to contain a lot of errors (that are automatically fixed). I usually never get errors, so I suspect that something is wrong with the models.

Therefore I printed the g-code you posted on my Prusa i3 Mk3, it worked perfectly. Before printing, I had no clue how much filament was going to be used and how much time the print would need. It would be nice if you could add that information on your website.


Should all be fixed in v1.01

I just checked and I can confirm: No more errors!

Could you please click on the error message and post the specific details you see? Like how many faulty faces, edges etc. Thanks!

I imported all the files from UniversalAutoRewinder_Parts_v1.0.zip and only these files don't have errors: Both springs, nutcurved and the clip. All the others have problems with the edges and some (like the hub) additionally have reversed facets and backward edges. The hub has a total 4636 errors.

You are completely right and I was already investigating into this issue. There are two causes. The first one was easy to find: the parts consist of multiple bodies for multi-material purposes and saving it as a single STL keeps internal faces. I can easily fix it by merging the bodies prior to single-color STL output. The second cause was much harder to resolve and eventually I found the culprit: a very tiny difference in the outputted coordinates of the same vertex (in an unrelated testing object) due to a rounding error:

vertex -3.289694e+03 -6.162976e-30 3.765357e+03
vertex -3.289694e+03 9.016132e-312 3.765357e+03

You see that one y value is 0.000000000000000000000000000006162976 and the other is 0.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000009016132, for the same vertex!

Now I know this I can simply translate the part so that no coordinate is zero anymore, and generate proper STLs. Will do this soon!

This is great, thanks for taking care of it so fast! This is definately one of the most perfectly engineered pieces I have seen here!

I will check the new files once you reworked them and report back.

oh yay! Looking forward to printing this. There's a "parts" and a "parts mmu" version, is the former the single-filament version?

Comments deleted.

Really nice work!! Would you consider uploading the frame without the text?

Great! The other parts are already printing and they are coming out really nice! I will keep the RaM in the dial :)

The Stand_notext_1.0.STL file is now available for download!

Dat is snel! Bedankt!

Just printed this and want to say that it is a beautiful design! Your attention to the little details is very impressive. It may seem strange that I am gushing over a spool holder, but your thought and engineering show in every little 'click in' connection. Well done!

I never get used to hearing such nice words :)
Keep 'em rolling! Which specific detail did you like most?

Ok, I'll fanboy your design. The barely there indent on the axel that keeps the clip from sliding. The small indents between the hub and clutch 'petals' that ensure solid seating, one into the other. The tongue and grove joint between the axel and stand. You don't need it but it makes for a better design. The great tolerances between all your parts. I sliced and printed the UniversalAutoRewinder_v1.0.stl, popped it off the plate and put it together. Standard Prusa Mk3, cheap PLA, beautiful result.

Glorification is sweet. Any critics/improvement points as well? I know that it is sometimes hard to turn the dial back from a high position, so might have to look into it.

Am I missing something about the tube holder? Is it meant to rotate freely through about 200 degrees in the upright position? If not, I can't figure out how to lock it in an upright position.

Turning the dial back was 'stiff' but not something that I would consider a problem.

It is meant for mmu2 and reverse bowden setups. Just leave it out if none applies to your setup.

Got that, in process of adding the MMU2 to my mk3. I've read all about the filament snarls and what people have done to mitigate them. Your image shows the tube holder upright. I can do nothing to lock it in that position. Is it meant to float?

Yes, I leave it floating. This makes it easier to swap spools and move the stand back and forth if needed. My tubes are around 300mm long with the other end attached to the MMU2 so the tube holder position is pretty well defined anyway.

Thanks. I didn't realize you intended a continuous tube from spool to MMU2. I was thinking of the solutions that have a short piece at the spool and a longer piece into the MMU2. That allows for expansion during retracts. But I guess you don't need that if your spool is retracting itself.

Correct, the tube must be continuous. Otherwise the rewinder won't work properly as the spring torque must be smaller than the tube friction force in order not to pull the whole filament out of the MMU2. And expansion during retracts is a feature of other spool holders.

What mat'l do you recommend printing this in?

I myself use PLA, this is also the easiest material for MMU prints. I also have a few PETG spring to test out if it works better than PLA in the long term, but at the moment I don't know yet. The clutch also has small springs, no idea if PLA will hold on the long term but we know that you can always adjust the setting if needed.

Hi, upgrading my older version to this version, i tried to use the SpringFlat_v1.0_PETG.gcode file that is on the link but it does not appear to home correctly on my mk2.5 and almost prints the spring off the platform, I am not great at reading gcode files, can you check/update?

The Y axis crashes into the back after of printer after the purge line, this does not happen on other prints

Well, I cannot reproduce your issue. The given gcode works fine on both my MK2S and MK3MM (single material mode). I installed the MK2.5 printer configuration in Slic3r PE 1.41.2 and compared the MK2 and MK2.5 versions of the outputted gcode. The only differences in the start section are a few additional "M73" commands (build percentage) in the MK2.5 version and an extra "M204" command (acceleration limit) in the MK2 version. None of these can explain the difference in behaviour.

One possibility is that it is a printer glitch. I occasionally experience weird behaviour (unexpected print head movements) in the startup/priming sequence which in my case does no harm and is gone in the next print. So I'd suggest to reboot your printer and try again. Do you have a single-material or MK2.5 MM?

I reset and tried it a 2nd time and it did the same thing, but I just let if finish as it was not to bad, I have a mk2.5 with mmu2. the multi-colour "full bed" gcode works fine (though i had to re-slice as my filament settings are different)

i will try it again if I feel i need the PET instead of the PLA spring.

as there are not many colour swaps I may even try to do the whole thing in Multi-colour PET as I have a load spare

Oh, a mk2.5 with MMU in single-material mode. I installed the MK2.5 MMU profile and compared the gcode output. I notice that the G21 and G90 commands occur before printing the purge line while the published gcode have the G21 and G90 commands after making the purge line. My theory is that your printer was somehow in a G20 (inches) and/or G91 (relative coordinates) state, resulting in an incorrect purge line. I experienced the same odd behaviour once or twice with my MK3. The fix is to put the G21 and G90 commands before the purge line commands and it should be solved.
The MK2.5 MMU single-material profile also outputs Tx and Tc commands, to show the filament selection dialog and move the filament to gears/nozzle. I'm not sure how non-MMU printers handle this command, if it is ignored then I could include it in the "universal" single-material gcode.

Gcode improved in v1.01, you could try it out if you like. I added G21 and G90 commands before starting the purge line.

Super impressive design, Vincent. Please forgive me if one day you see a remix without the branding. I'm not in a hurry to print these since the hubs I made from your first 2 designs are still working really well, but I'll get around to it eventually.

Thanks! No worries, the spring guard also has the RaM logo. :)

Will the printed axles from last parametric spool holder design work with this one ? Thanks !

Yes they will! (if long enough of course)