Prusa MMU3 for MK4 - Complete Assembly and 3D Print
The Prusa MMU3 for the MK4 3D Printer finally arrived April 2024 and here is my step by step assembly guide with many tips and tricks where I show how to assemble the MMU3 (Idler, Selector, Pully Body, Electronics), upgrade the MK4's Nextruder for the MMU3, and configure the software and firmware for both. I show the new Prusa Buffer and Spools, but then showcase my own custom Multicolor System, and finally print a test model to demonstrate.
Prusa Links:
Prusa 3D Printers - www.prusa3d.com/
MMU3 Assembly Guide - help.prusa3d.com/guide/1-intr...
MK4 and MMU3 Firmware - help.prusa3d.com/downloads
MMU3 Printable Files - www.printables.com/model/5316...
Custom Parts Shown:
PC4M10 Rear Connection Custom Part:
www.printables.com/model/5424...
Custom Colored Label Plate:
www.printables.com/model/5425...
Custom Spool and Filament Multicolor System (1:11:25)
www.printables.com/model/5552...
Alternate Buffer by Martin:
www.printables.com/model/3081...
PINDA Strain Relief:
www.printables.com/model/1053...
3mm ID PTFE tubes throughout (the short ones in the MMU and behind it to the buffer/spools)
www.amazon.com/gp/product/B07...
My Printables Profile:
www.printables.com/@Real3DPrints
Basic Five Color MMU Test Model by Alloria:
www.printables.com/model/1568
Chapters
Intro: (0:00)
Box Contents: (1:00)
Self Printed Orange Parts: (1:32)
Why Upgrade: (2:05)
Where to Get New Parts: (3:49)
Idler Body Bearings: (4:54)
Idler Body Motor: (9:00)
Idler Body Alignment: (12:16)
Selector Magnets, Ball, Trapeze: (13:24)
Selector Tubes, PTFE Cover: (15:04)
Selector SuperPINDA, Cutter: (16:38)
Pulley Body Bearings: (19:38)
Pulley Body Custom PTFE tubes: (20:21)
Pulley Body Front PTFE Holder: (21:29)
Pulley Body Rear PTFE Holder: (22:26)
Pulley Body Pulleys, Motor: (23:30)
Pulley Body Alignment: (26:52)
Pulley Body Selector Install: (27:28)
Pulley Body Idler Body Install: (31:04)
Pully Body Idler Screw Tension: (32:01)
Required New Static Protection Parts: (32:59)
Pulley Body Control Board, PD Board: (34:05)
Pulley Body Cable Connections: (36:07)
Pulley Body Cable Management: (38:46)
Pulley Body Frame Holders, Label Plate: (40:05)
Attach Serial Sticker, Mount MMU3: (42:05)
Nextruder Disassembly: (42:48)
Nextruder Idler: (44:48)
Nextruder Swivel: (46:00)
Nextruder Tension Screw: (49:29)
Nextruder Gearbox Disassembly: (50:34)
Nextruder Main Plate: (51:05)
Nextruder Gearbox Assembly: (53:10)
Chamfered side faces Nextruder: (53:44)
Connect MMU3 Cable to MK4 Electronics Box: (56:29)
Update MK4 Firmware: (57:38)
Get MMU3 Firmware: (59:07)
Turn MMU3 On: (59:46)
Update MMU3 Firmware: (1:00:38)
Fix Unload Filament Message: (1:01:46)
GearBox Alignment: (1:02:40)
Filament Sensor Calibration: (1:03:37)
SuperFINDA Calibration: (1:04:26)
MMU3 New Buffer: (1:06:00)
Spoolholder Assembly: (1:07:43)
Preloading Filament: (1:09:09)
Custom Multicolor System Intro: (1:10:36)
Updating Prusaslicer Config: (1:10:52)
Custom Multicolor System Benefits: (1:11:25)
Fixing Loose Pulley Grub Screws: (1:12:54)
Preloading Filament Again: (1:14:05)
Printing Five Color Test Print: (1:14:43)
Conclusion: (1:16:01)
Thanks to your guidance building was easy and fun, great tutorial as an addition to the online setup guide. Great job!
Outstanding tutorial, Prusa owes you a ticket to the next rep-rap festival...the rest of us owe you a coffee, or some filament or something!!
Thank you! I had fun making it.
thankyou so much for sharing. I didn't make an order on the mmu3 for my mk4 yet, but after watching your setup and explanations, it becomes more than likely to happen even sooner now. greetings from germany😎✌️🍀
The order queue time is about 9-10 weeks as of today so if you want one don't wait. You can get started now though and print the R4 parts from printables and plan out your colors. I would strongly suggest looking into building the spool and buffer custom system I show in the video too so when it arrives you can just build the MMU3 part and the nextruder mod.
just ordered mine today. leadtime to germany 6...7 weeks😊
Holy Crap... Mine just came...
Great.. I´m receiving mine today 🙂
Great 👍
Very good video. I ordered the mmu3 upgrade for the mk4. I already have the mmu2 (original relase version) but never assembled it due to the many problems back in the days. However, imho the basic approch bambulab uses is more interesting. For the basic principle of the MMU2/3 etc: - The MMU loads the filament to the printhead. - as soon as the feeder inside the printhead grabs the filament, the mmu disengage, both pulling the filament through the whole setup is done by the direct extruder - if the filament is unloaded, first the direct extruder unloads, than the MMU engage again and pulls the filament back (into the buffer or with spring loaded spoolholders back onto the spool. - Only one filament can be engaged at a time The AMS uses a different approach: - each filament has a seperate AMS-feeder, permanently engaged (PWM controlled DC-motors used, not stepper motors) - the active AMS-feeder permanently pushes the filament to the feeder in the printhead while printing - a "filament-sychronicer" is used to controll the AMS-feeder. Iritatingly bambulab calls it "filament buffer", but it "buffers" less than a cm. It's real function is quite interesting: A slider with integrated magnet, a spring and an hall-effect sensor. If the tip of filament can not move further (= it is pushed faster than the direct extruder takes it), the spring compresses and the slider moves. The hall-effect sensor detects this and the AMS feeder stopps. As soon as the spring decompresses due to the extruder using filament, the AMS feeder start again, pushing. This way both feeder can stay attached simultaniously. - Filament spool rotetes backwards, so a buffer is not needed (motors turning spools on AMS, springs on AMS light) The approach bambulab uses seems superior, the implementation could be better: bambulab uses cut&poo inside the printhead. Quite wastefull and slow. As every filament has its own AMS-feeder permanently attached, it would be nicer to disengage the printheads feeder and unload only the AMS-feeder. If the extruder gears are not attached, the filament can be unladed much faster, as it can pass the extruder gears without being squezed - so it wouldn't be deformed even if it is still to hot to be perfectly solid. Also the filament tip could be cut outside to have a perfect tip for the next reload. As the AMS-feeders are independend, the filament-tip-cut could be done while the next filament alsready is loaded and prints. Filament changing time could be reduced to 30 seconds or less (instead of 90 seconds). This would speed up the overall printing time for multicolour by a lot. (actually even some multicolor prints are faster on a mk3+mmu2 than on a bambulab, as the filament change with mmu is faster). Prusas approach was founded at a time where stepper motors and feeder were quite costly. Using 3 stepper drivers instead of 5 and a selector to use 5 filaments seems nice. Esp. only needing 3 quite stepper drivers. However, nowerdays the prices are much lower. Also you could use DC-motors instead of steppers for the AMS-feeder (as bambulab does). And the filament sychronicer is a gamechanger, as it allows the AMS feeder to be permanently attached to the filament. Actually, the Bambulab approach ist quite interesting even for single material. Use a strong "AMS-feeder" and a feeder-sychronicer, the extruder does not need to pull the filament from the spool. So you could use 10 or 20 kg spools instead of 1 or 2 kg ones without increasing the load the extruder has to carry...
That is a great comparison analysis of the motion. I did not fully know how the AMS worked. I think the main two weaknesses of the MMU approach are passive higher friction spool beds, and less control of filament position after a retraction, but both of these are not about the MMU itself but the path behind it, and can be both be solved with a custom solution. The AMS solves them now. The MMU lets go of it, unlike the AMS, which holds onto it. Like me, the AMS also went with a center spool exact fit design to solve the spool friction issue. It seems that the AMS saw these two issues and solved them. These two weaknesses for the Prusa design force a requirement that the extruder must pull all that filament through a long path and around a buffer wheel, and that after a retraction, the tube diameter and buffer design need to balance lowering static friction vs holding onto the filament to prevent a slippage past the pulley gears in the MMU. While the AMS as you state solved this filament control problem by using more motors that stay engaged, Prusa stuck to using one motor that disconnects from the path, and a buffer system with a loop. I haven't seen anyone point it out, but I think that Prusa also intended for the tight bends in the buffer loop to 'lock' the filament in place after a retraction. That in my mind is something the buffer is probably doing, but it is not ideal and is done better before the filament gets to the buffer. Prusa also seems to be struggling with playing with tube diameters, going from 2 to now 2.5, perhaps to balance this friction. (why not 3mm?) I eliminated moving parts and sharp bends in the buffer, and in the filament path, but still force a gentle S curve filament path shape behind the MMU with my arches to lower friction and control the position after a retraction. I don't say this in the video when I describe my custom setup because it goes beyond the scope, but it is really important. That 'S curve hill pattern' right behind the MMU is what is pushing the filament against the tube walls and keeping it in place. I even am able to use 3mm ID tubing to reduce it further. I feel the spool friction issue for the MMU is best solved by the setup I demo here with exact fit center ball bearing spools, so heavier spools could be used, as well as cardboard. With the lower friction benefit there, I have no problem using rough matte filament in the rest of the system. I do also have a more simple mod that replaces the stock spool wheels in the stock spoolholders with bearing wheels as another spool friction solution on Printables too. Interestingly with the Prusa design more colors could be added without adding more motors by elongating the MMU like the Enraged Rabbit Carrot Feeder (12 colors) or the MX Research (9 colors), an interesting MMU remix on Printables with custom firmware as well on github. I wish Prusa had this as another MMU option, as more colors with no added motors could be seen as a good thing. Adding an optional but more advanced spoolholder to the Prusa shop with adjustable width center bearings and metal supports would also go a long way. There is also the issue of filament waste/cost for the AMS vs the MMU, which is an unrelated benefit of using a buffer system, not to mention the cleanup of cut bits on the AMS. Overall as the system ships I think the spools are the main weakness, but that is easily solved by using ones with center ball bearings. And one could swap in 3mm ID tubes with the stock buffer and lay down a path with a curve in it behind the MMU to prevent slippage.
@@real3dprintsYes, for scaling up to more filaments using only 3 motors, the MMU2/3 design (ERCF is also based on it) seems nice. However, the force needed just for filament feeding can be be aplied with a quite cheap dc-motor controled by a simple double-h-bridge. So you can calculate less than 15€ per colour. Also only two I/O pins (one with PWM controlled) are needed per motor. One other advantage of Bambulabs approch: As the filament transport while printing is done by the seperate AMS feeder, the extruder feeder needs less torque. So the motor can be less powerful - and therefore less weight. Actually I had some issues in the past with friction on the filament transport parth, leading to inconsistent extrusion. I also solved it with 3mm inside diameter bowden. A totaly different aspect: The old mk3 (not mk3s(+)) filament sensor was quite nice, but unreliable with high glossy filament. It's like using an optical mouse on a glas surface. However, if an AMS-Feeder is passed before, the gears leave marks on the filament - and detection of the optical sensor gets reliable again. Like with an optical mouse on glas if you scratch the glas with sanding paper. So with AMS-style filament transport you could use the mk3-sensor again, which detects movement, not only filament presence. Actualy back in the days of the mk3 without s I actually used a feeder without motor in the filamentpath to "prescretch" the filament before it got to the sensor ;-)
the rear PTFE tube holder was a nightmare. AFTER I managed to get it done (without losing screws and nuts, but I came close many times), I saw a comment "flip the 3x30 over and push it to hold the nut up." I'm like....dammit!!! That's a good idea! I wish I thought of that.
Agree! Here is my suggestion for that frustrating step which I also added to the online assembly comments - This step can be frustrating because you are not only trying to keep the hex nut in place but also oriented, and the pressure of trying to screw in the m3x10 keeps pushing the nut out when you try to screw in the m3x10. Instead try this: screw the nut onto the end of the m3x30 but just barely with one turn into the threads, just enough so it doesn't fall off. Put this into the hole, press it in so it is in position, and keep pressure on the m3x30 holding it there. Now flip the entire assembly over. Screw in the m3x30, all the while keeping pressure on the m3X30 to keep the nut in place.
@@real3dprints Yeah I saw you did that in the video.
As good as this video is, it should be required watching for anyone considering which direction to go: Bambu/CrealityK/AnyCubicKobra3/Anker vs PRUSA/Voron/LDO/RatRig/etc. It should immediately become clear which direction to go
Thanks for sharing! I expect the mmu4 requires less space Unfortunately, this model is too big for my small table
There might be a way to put the buffer and spools on a shelf above the printer.
I've always used filament dryers as buffers for the mmu2. I have 2 brand new sunlu s4 filadryers for the mmu3. Those will be my buffers. It gets messy, but I've never had anything get tangled.
Does it buffer around the spool or in the air. If in the air what is the open distance gap?
@@real3dprints it just kind of falls to the side of the dryer. But the amount of filament isn't enough to be a problem. At least not with the dual slot units I've used in the past. Using a quadruple slot is new. I've used it for 2 color printing on the same side so it wasn't any different than a dual slot. But I haven't used all 4 slots for a multi-color print yet. I still don't anticipate any problems.
The video is perfect for shortening my waiting time. My MMU3 for the MK4 should arrive this week. Does it only come with black filament? I just want to add a few orange accents. I always have Prusa Orange around anyway. I just like the color
Yes it comes only in black Prusa PETG. I ordered some Prusa Orange PETG (part) and PLA (multicolor label plate) so I could do it like that. You might check out the different colors folks did here - www.printables.com/model/542588-mmu3-label-plate-with-custom-colors-mmu/comments
I am so glad that I canceled my MMU3 and ordered the XL.
I see folks printing one model with 500 plus tool changes on the MMU and don't get it, as the tool changes really add up. IMO the MMU is great for models with a low number of tool changes, or for small models where you can print three or more at once to lower the added tool change penalty time. I can't afford the XL so ended up getting several MMUs instead, but most importantly I don't print big things.
Does your XL run ok? I want that printer so bad
Very nice Video... I realize I don't have the space for the mmu... Right now at least.
thanks for the video, it's very interesting. can you tell me what kind of cloth that is
I got it it a a 3D Printing convention from a booth vendor - i cant remember the brand - one side is a heat quilt
So I have the R3 electronics cover. I intend to use it.....until I can reprint it with color numbers (MMU based) on my new MMU3 and then replace it with the R4 version. I suspect it shouldn't pose a problem for a couple prints.
I bet it is fine without the new part and they only added that part in reaction to some specific government ESD spec.
@@real3dprints So I installed it last night. It doesn't fit correctly. It's good enough for now. I suspect it will fit correctly after I print the new part with the colored numbers.
Can someone explain why ERCF exists when this much nicer solution should be adopted for Voron?
Just read the Prusa blog on MMU3 development. Having to replace the main plate of the nextruder every 80000 color changes is an absolute no go and in my opinion speaks for a faulty design of the extruder
Don't know why, though. Without MMU3 it works just fine without the MJF part. Great that Prusa got MMU working, but it just takes to much space compared to Bambu's approach. They need to come with a different design I think.
@@marcel151 In their block they wrote that the original petg main plate was "strong enough" for many years of manual filament changes. With the mmu3 there will be like up to 1000 or more changes so the main plate would not be strong enough to work correctly. In their block they also showed a multi color print of a little cat with ~500-960 color changes. So "just" 80 of those little cats and you need to change the extruder main plate? In my opinion that sounds really sad
@@Skaten7 Sure, I read that bloG post. Seems that they'll have to redesign the nextruder in the MK4S. ;-) I still have problems with twisting filament, a lot of people say because of these part that you need to change with MMU3 build.
lol I've never been afraid to tighten the shit of the grubscrews. I didn't have a problem in that area with the MMU2.
I hear ya! One of my MMUs I have I got secondhand and those screws are so tight I couldn't move them and feared stripping the tool hole so didn't try. Luckily they were in the correct position. A couple people suggested using blue Loctite to reduce slippage, verses overtightening them and risk. At this point I am paralyzed with fear of messing them up to the point where I drastically under tightened them.
@@real3dprints yeah I don't go to town on them until I'm convinced they're properly aligned.
IMO the 2.5 tubes with a single bezzle on one size is either a mistake or the instructions are wrong (just as the instructions were wrong on the mmu2s as well as it had the same issue). The problem is I found it was always quite difficult to load my mmu in the first place. Then I had a line get stuck completely. I couldn't fix it. In fact a year later I still couldn't fix it. I had to replace the tube. When I was about to put it back together I'm like.......wait......why is the bezzle sticking out? Shouldn't it be sticking in? So I flipped all of them from out to in. INSTANTLY I found that it made a HUGE difference. Suddenly loading the mmu was super easy. So last night when I started building my MMU3 for my mk4, I I installed them backwards for that very reason. At some point I may go in with a drill bit and add the bevel on the other side, but for now the bevel is only on the side facing away from the extruder. I'm not making that mistake again. Someone mentioned their MMU2S (one of the last ones shipped before the MMU3 was announced) included dual sided tubes. I don't know if that was true and prusa stopped doing it, or what, but it seems like a good idea to me. Why not make it dual sided? I do have the drill bit. So I might do it. The downside is I have to uninstall the selector again. sigh.....
If they are going to bevel one side but not the other I feel there should be some sort of explanation as to why. That narrow tube does force the filament to be positioned better, but getting it loaded is harder as you say. I just used 3mm instead so there is no need to figure out the bezel as it effectively now the same width as a bevel on both sides. If you have the drill bit to bevel it yea do it, but remove them first so you don't accidently bevel the surrounding plastic too.
@@real3dprints right. Someone said it's for when the filament runs out. MAkes sense. If the finda isn't triggering then it's not in those tubes. But I've ran my MMU2 like that for a while and that was never an issue. It was always stringy issues and stuff like that.
Why dont you use rewind spool holders like in unoriginal box?
I used a custom spool holder with center bearings to eliminate all friction caused by spools pulling. This makes it easier for the filament to get pulled by the mmu and extruder gears. This allows me to use filament that is more rough like Matt filament and cardboard spools which don’t rotate as well on the stock spool holders rollers. The ones that it comes with are not rewind hence the buffer is needed.
I see I'm not the only one that got 2 mini spools. I thought it was kind of funny.
I wish they could let you choose orange or black as the PETG color (and get matching label plates), or even better have an option to not even order the official buffer, spools, or black filament and have an MMU3 light version that is much cheaper (as I didn't need/use any of those items).
@@real3dprints right. I haven't even decided if I'm going to build the buffer or not. I haven't printed the parts. I'm using purple for the mk4 (which is also purple) and I'll use the included black for the mk3. But honestly I wish I could have it in orange like the mk3 is.......hmmm I could always print it in clear (I have clear PETG lol)
Nice Video. But please note that if you assemble your MMU3 with this video instead of the official instructions you are unfortunately not eligible to eat the gummy bears.
I never get to eat those bears - my kid says they are bad for me and then takes them to school to give away.
Have you already printed something without problems? 😂
Yes once i fixed the grub screw for the pulley idler (which I didnt screw in enough) the quick test model printed fine (100 tool changes). See the end of the video where I show that.
@@real3dprintsI just laugh. At work I have two MK3s with MMU2s and it is practically impossible to print anything without the attention of the MMU. It causes too many different problems. I hoped that MMU3 would be better and maybe I would buy one for my MK4, but after buying Bambu Lab A1 mini with AMS Lite, I might have to say goodbye to the Prusa brand altogether. Prusa used to be the best, now it is overrated and not worth the price.
I’m so confused, they make the customer 3D print the entire unit and supply a labyrinthine gamut of parts to install, and then supply injection molded spool rollers that look like they came out of a Cracker Jack box. 🫤
The new spool holder wheels are a special kind of plastic different from the rest of the spool holder. They also give you divider panels for the buffer too. But I agree, although they work ok and are quick to set up, I think the spools are the first thing to replace with something custom with bearings and a more solid footing. Prusa should sell an upgraded spool holder solution, but I think they are keeping costs down on this particular spool holder solution. There are quite a few custom spool holder designs out there too one could try.