Prusa MMU2 Selector with magnet and opened FINDA canal

by emanuvele Mar 30, 2019
Download All Files

Thing Apps Enabled

Please Login to Comment

Would the magnet be strong enough to pull the ball bearing away from the sensor if the entire mmu sensor was mounted upside down, as I know it runs on a gravity assisted system. Otherwise I'd be looking at a spring of sorts. Tia

Maybe you want to include the following f3d and stl in your Thing. With default settings (speed/cooling) my PETG filament decided not to stick the tiny vertical mid section that holds the ball in place - resulting in FULLY opened ball channel... which for obvious reasons wouldn't work. In the attached files I've just made the vertical midsection larger, hence reducing the possibility of this type of error occurring.
Thanks for the post!

This may sound really silly, but where do you place the magnet? None of the pictures show the magnet in place.

This upgrade make my MMU2s work perfectly. I recommend. Give it a try. Thank you!

Is anyone using the latest version of this (E2) and flex filament? I'm having some issues and skeptical if it's any better than the stock selector for this. Obviously it's a huge improvement for harder filaments like PLA/PETG.

I printed the E2 version of this and tried to test feed some filament through it before hooking everything up just as a function check. I didn't work out so great. The sample filament I cut off to test with was from the inner part of a roll so it has more curl than filament from the middle or outer part of a roll. The problem is the open feed path under the ball is long enough for the curl to reform and change the filament direction enough to cause the feed to fail in most orientations. The only orientation that works is towards the magnet. If it the curl faces towards the window or towards the ball, it will fail to load. Note filament from the middle or outer part of a spool works OK. All samples were cleanly cut with flush cutters.

The root cause of the problem appears to be with the Prusa design. It appears this 7mm ball bearing size was chosen because the FINDA probe is 7.88mm wide at the threads and they just continued the hole down, resulting in a long free section of the feed path. I'm wondering if it wouldn't be better to narrow the bottom of the opening down for a 4mm bearing, which are readily available. This would reduce the "free" section of the feed path significantly and therefore the chance the filament would get stuck on feeds if it isn't straight enough.


Did you try to repeat your tests with the selector mounted in place, feeding the filament from the back of MMU2?
I think that when the filament passes through the gears and the pulley body, it is straightened and the path doesn't allow it to bend that much.

I just checked - the feed path does straighten it out. It seems marginal (high friction) but it does work in most cases. The only exception is if the angle of the filament cut compliments the curve - then it can get stuck on the attached face.

I modeled a smaller bearing and it seems 6mm may be a better size. A cold pull is about 2.45mm and the current feed path size is 2.5mm. To have the filament push the bearing up it needs to be 2x this size (or a little bigger). So I think an improvement could be made here but I'm not sure it's worth the effort unless it becomes an issue down the road. I spent a little while trying to narrow the bottom of the bearing channel in the fusion model but that breaks a bunch of the geometry which is based on Extrude12, etc etc...

I'm going to try this next. I was also thinking along the same lines - that the filament paths may straighten it out enough that that it doesn't happen in use.

@emanuvele Obviously an awesome design overall but perhaps one more mod can be incorporated into this. That is, a separate blade "cover". That way, you don't need to glue it or worry about it falling out. There are two different ones here: https://www.thingiverse.com/thing:3310038 and https://www.thingiverse.com/thing:3608229. I think the more flexibility this part can be printed with the better.

EDIT: Forgot to mention that @kmccon's great design also incorporates one: https://www.thingiverse.com/thing:3157550

Prusa MMU2 refined selector
De Luxe selector for prusa i3 mk3 MMU2s
by frimmel
PRUSA MMU2 Selector Improvements
by kmccon

Had a little time so I modified it myself. I took emanuvele's latest E2 design (thank you very much for the source) and modified the blade surface to include the 3 "pegs" in kmccon's design that the blade holder sits on top of. Use the blade holder there (also attached here for simplicity) to hold it in place.

If there's enough interest, I will post a proper remix.

Great! Just what I needed

I'm using your new design and it worked fine for me.

Comments deleted.

Hello, what is the face to put on the bed to get the best print to avoid any support and have the best bridges ?

Hello, the face to put on the bed is the one with three equal sized holes. Please see attached picture.

This is an improvement for loading default filament, but makes unloading filament impossible. Also flex filament won't work with this anymore.
The main issue with your part is unloading filament after a runout. When a filament spool ends, the FINDA probe will sense the runout , the extruder then tries to push back the filament. But there is no chamfer on the way back and the added force of the magnet will make the ball deflect the filament end on its way back. Its getting stuck 100% of the time. Spooljoin won't work anymore.
Also flex filament will get deflected too much on load and get stuck somewhere below the ball. Even manually its impossible to push flex through.

I've attached a picture comparing your parts filament path with the original one and I'd say it would be better to just retain the original path below the magnet, basically adding plastic between the magnet and the ball

Hi DonKarnage, thank you for pointing it out.
I've now published a new E2 version of the part, in which I tried to fix this issue, reducing the plastic cut between the magnet and the finda ball, and restoring a smooth reverse path for the filament, maintaining at the same time an open window for letting the debris go out.
I've attached a picture that shows the new filament path.
However I've not printed this new part yet, so I cannot guarantee that it works correctly.

Has anyone tested out this version? I've just about given up on my MMU2S and have completely bypassed it for now, but I'm willing to rebuild it using some of these great parts.

it worked great for me. My new MMU2 (out of box) with this and a few other mods has now printed a number of multicolor parts without any problems.

Thanks for the report! Next time mine jams I will mod it. I also intend to install universaljoint's underslung control panel here (https://www.thingiverse.com/thing:3211711) at the same time. I think these are among the top mods of the MMU2S here.

MMU2 underslung control panel

Awesome! I highly recommend this along with https://www.thingiverse.com/thing:3541897. I had my first multi-color print last night that required ZERO intervention due to MMU2S load/unload problems. When I first attempted the same print a few weeks ago I had to intervene at EVERY color change. This was due to the fact that I was using 3 different brands of PETG to get the colors that I wanted. (I was trying to duplicate a company logo and match colors as close as possible.) I probably could have spent a lot of time tuning the temps/settings for each individual filament, but with these 2 mods i didn't have to.

Prusa MK2.5S/MK3S+MMU2S + M10x1 PTFE QuickConnect Passthrough

How many perimeters did you print? Also, is there a benefit to printing at .1 in lieu of .15 layer height?

I printed it with standard 2 perimeters. I don't think there would be significant benefits using less than .20 mm layer height.

Just a tip for people disassembling their old selector for parts, I got out the square nuts with a dental Pick, and a chopstick helped me remove the tubes.

This worked absolutely perfectly. I couldn't recommend this method more.

This is inspired greatness! Solved all my problems with my selector, would highly recommend.

I remixed this so it can use the stock magnet provided with the mmu2s/mk3s upgrade kit, https://www.thingiverse.com/thing:3688458

Prusa MMU2 Selector with magnet and opened FINDA canal for stock rectangular magnet

I printed your version, very smart to use the magnet that it came with. Thank you!

Comments deleted.

WARNING!! Take special care about the spring screws for the PTFE holder. If you inserted them too much into the selector, the screw will hit the FINDA. It happen to me... =(. Sadly I damaged my FINDA, no more MMU2 for a while until a new sensor arrive...

Thanks for the upgraded selector, it works really well. And thanks for including your fusion 360 file, I was able to change the magnet hole for the 6x1.5mm (1.38mm effective) magnet I had.
If anyone needs it, I included the modified STL file.

I saw on your picture that you use spring loaded screws to hold down the ptfe coupler on the body of the selector. What springs did you find for this? I am assuming this helps when the filament hits the wall to the sides of the hole? Seems to be the only issue I am still running into

Hi Sindarius, the purpose of the springs is not to avoid filament hitting the cover, but to help when MMU fails to load the filament into the extruder. The distance traveled by filament when loading is not very reliable, and sometimes it stops before the extruder gears and doesn't reach the inside, so the extruder prints "air". Using the springs and calibrating the loading distance a bit longer help preventing this. The springs avoid pushing the filament too hard into the extruder gears. When the filament reaches the extruder gears and MMU still feeds it, the springs absorb the extra distance.
I used two steel spring taken from standard ballpoint pens (4mm diameter). You could also buy one of this:

A 4mm diameter should fit the screws.

Thanks for the follow up! I see what you’re getting at and it makes sense to me. Thank you

What surface should be down when printing?

The one towards the PTFE tube holder. See attached picture.

Perfect, thanks. That was my guess as well.

Does this also fit on MMU2S?

Yes, I have the MMU2S and I highly recommend this mod. I now have over 40 hours of printing since I installed it, and have not had a single failure. All of my prints have finished without requiring user attention.

On average, I have been getting a FINDA problem causing an MMU2 error about once every 200 filament loads. I printed and installed this, with the magnet. I have run 3 different MMU2 prints for a total of over 20 hours of printing with no problems at all. Thanks!

Just finished my MMU2 and I've been getting jams every 4-5 filament changes. Going to see if this fixes the issue!

This is just brilliant. Fixed the issues I was having with filament getting hung up at the selector, and being able to see the pinda ball is great for troubleshooting loading issues. I haven't tried adding a magnet yet because it doesn't seem to need it. Thanks for posting!

Looks good. I’ll have to give this a try.