MMU2S not ready for public use  

Page 5 / 6
  RSS
Anachronist
(@anachronist)
Trusted Member
Posted by: @jsw

...

Mounting the buffer vertically behind the printer. I printed a replacement 'base' which allows it to stand up, and two tube holders to direct the input to the spool rack above the printer.  This is done.

...

Switch to kill the power to the MMU2S so it can be used in the normal single-color mode when needed.  I printed the switch housing already and the switch is on order.

I considered a vertical buffer too, but with all the stories I read about how much trouble it is to load, I went with auto-rewinders instead.

The power switch is a cool idea. I'm tempted to do that myself just to have it... although, when I want to use a single color, I just do it through the MMU2S, selecting one of the 5 MMU2S filaments in the slicer (or by default, it uses filament #1). The slicer doesn't create a wipe tower if you're using just one filament. If your MMU2S is working fine, I can't see a justification to to operate the printer as a single-filament printer; just print with one of the 5 filaments.

My middle feed tube goes to an easy-load auto-rewinder that isn't covered by a drybox, so I can quick-change out a filament in that position if I need something special. I have a spool of PVA, and that's likely to stay in its bag rather than on a drybox rewinder. I'd take it out of the bag and install it on feed tube #3 any time I need it.

 

Posted : 23/05/2020 2:05 am
bob.s7 liked
jiral
(@jiral)
Eminent Member

Of all the mods I have never seen a reason to switch off the MMU in Single mode. In fact I consider quite some gain in comfort. You can have several filaments mounted and just choose which one to print with with no manual interaction. After the print its being unloaded again. At all its finnickiness, the MMU should not feel challenged by a single load and unload, unless something very fundamental is off or if you've got some especially nasty sort of material or a clogged nozzle.

Posted : 23/05/2020 2:25 pm
bob.s7
(@bob-s7)
Active Member

I am having really, really mixed luck with the auto-rewinders. It seems like they work great with small spools, but I can't get them to work consistently with full spools. How did you guys solve that problem?

Posted : 24/05/2020 9:44 pm
Anachronist
(@anachronist)
Trusted Member

@bob-s7 - my two rewinder drybox designs, which I mentioned about 8 posts back in this thread, work well with full spools. They are based on an integrated auto-rewinder hub, which I had to improve to get working correctly. What problems are you having?

The initial problem I had with rewinders is that they would pull the filament clear out of the MMU2S after a retraction, because the spring still had too much energy stored up. A rewinder for the MMU2S needs to pull back only about 1 turn, not 7! And it doesn't need to exert any real force; because the filament is being pushed back toward the rewinder by the MMU2S, the rewinder just needs to turn the spool, not pull filament. This problem of too much energy storage became evident when I replaced all my PTFE tubes with 3mm inside diameter tubing to reduce friction.

This post was modified 2 weeks ago by Anachronist
Posted : 24/05/2020 10:51 pm
jiral
(@jiral)
Eminent Member

I had issues as well, however, with the Universal Auto-Rewind spool ( https://www.thingiverse.com/thing:3338467) I managed to get a very reliable system by doing the following things:

  • Using the stiff spring and print it in (Prusament) PETG. The standard setting of the spring is 12-13 so far.
  • Relocating the spools from above the printer to below the printer.
  • Installing basically two sticks that guide the filament onto each spool, preventing dislocation besides the spools when rewinding. Compared to some guiding tube this has the advantage of less friction and more freedom for the filament in its movement, with less awkward bends right at the spool.

With the other 3 things in place the pretensioning isn't even strictly necessary but it enables a slightly cleaner start in the rewinding. Full 1kg spools work very reliable. I have not tried anything beyond that (2.3 kg spools or such), those extra heavy spools might be indeed too much.

I did not go for a zero friction system and only replaced the tubing between selection and extruder to a 2.5 mm ID one. A tiny bit of friction in the feeding tubes is in my view actually a good thing as it protects from pulling the filament out of the MMU idler.

 

This post was modified 2 weeks ago by jiral
Posted : 24/05/2020 11:01 pm
iracarmel
(@iracarmel)
Active Member

I must agree with the sentiment that the MMU2S is not ready for public use.  To me, this thing is a heap of junk.  Not only are there no clear manuals about how to set up this incredibly finicky pile of pre-trash, it's like voodoo magic when it does work. The number of times this thing has made me blind with rage is just amazing.  I honestly don't even know what to say about this thing in a constructive fashion.  This piece of shit makes me curse Prusa's name whenever I try and use my printer.  It's rubbish, absolute unadulterated rubbish.  I was so hopeful that the firmware update would make this thing usable.  No, that's a joke.  In fact it seems to be worse, if that's even possible.  😀   How many times have I had to take apart the hot end to fix jams or screw ups?  How many times have I had to try and fix a filament that won't load.  How should I set the tension on various screws to make it work?  I think it would be easier to count the number of successful prints.  3?  

Does the Prusa unit use resistance in the PTFE tube to tell if filament has passed the bondex gears?  I widened up the PTFE tube because it just seemed too tight even with nice big chamfers.  Then I noticed that the PTFE tube was crimped, so this means that the filament pulled on it hard enough to pull up so much that the locking plastic just bit into it.  So I assembled the heatsink, and then reamed the PTFE out with a 1.8mm drill.  There, fixed!  Oh no, next up the filament would just never stop going in during a load, and TONS of plastic would come out forever.  So I replaced the whole tube, and started over, being super careful, making sure the PTFE passed filament easily, but not too easily.  Oh joy, it actually loads again, but now the PTFE in the heatsink is too narrow, and it just won't load.  It's rubbish.  Just junk.   Is there a faq?  If so I can't find it on this miserable website which seems to be constantly redirecting me, and being generally annoying as hell.  

Internet rules right, don't post when you're angry, I can't help it.  I'm just about done with 3d printing because of this thing.  It has completely killed my interest.  I've been working on this thing for over a month.  WTF.

-Ira

Posted : 29/05/2020 7:38 pm
nikolai.r
(@nikolai-r)
Noble Member
Posted by: @iracarmel

 .... Is there a faq? ...

I'm always wandering how much time people are wasting on complains and how little on actual research.

There are tons of information about the MMU2 on the official website. Just to mention only some of them ...
https://help.prusa3d.com/en/search/?s=mmu2
https://help.prusa3d.com/en/guide/1-introduction_36832
https://help.prusa3d.com/en/article/mmu2s-setup-and-inspection_2233

We have also a tons of information in this forum and many people ready to help.  With your first post in this forum asking about your issue a month ago, you might could get a better understanding how MMU2 actually works and be happier.

Posted by: @iracarmel

.... Does the Prusa unit use resistance in the PTFE tube to tell if filament has passed the bondex gears? ...

The filament sensor is the trigger for filament loaded state.

Often linked posts:
Going small with MMU2
Real Multi Material
My prints on Instagram...
Posted : 29/05/2020 8:31 pm
jiral
(@jiral)
Eminent Member
Posted by: @iracarmel

Oh no, next up the filament would just never stop going in during a load, and TONS of plastic would come out forever. 

This does sound like you have an issue with the IR sensor. You can check in the Support menu -> Sensors if the IR sensor is triggered when the filament hits the extruder gear, if that does not work reliable it causes all sorts of very nasty havoc. I had to finetune the trigger arm of the sensor. What I did was I printed out modified trigger arms with incrementally elongated ticks. I think in the end I simply settled with a trigger arm that was merely 0.2 mm longer or so. Since then the IR sensor has worked absolutely flawlessly.

The triggering is also influenced by how strong you set the tension screw of the extruder gear, especially with softer filaments. They can be compressed a bit which reduces the amount the trigger arm moves, by a tiny bit. If your tolerances are to tight that can be enough that the IR sensor is not triggered.

 

I went through a valley of tears with the MMU for some time, but honestly I did not work that intensely on it, once in a while, until, apparently I was through with all possible errors and since then (knock on wood) it works just fine. Yes, it is wasteful but the prints are absolutely fantastic. Once you manage to dial the system in, it works. So it is not trash, but it is probably not fit to be sold like a ready to use product. Mind you, there are some people who had good experiences right out of the box, also with the MMU but you have to be ready to invest some time if it doesn't and you should enjoy trouble shooting and tinkering at least a bit.

Posted : 29/05/2020 8:48 pm
iracarmel liked
andreas
(@andreas-t10)
Eminent Member

My half a penny opinion on the matter. I can understand both sites. The one that says that this product is not ready for production and the other that argues that for a lot of people it work.  I have my mmu2s for almost a year now. I complained A LOT and almost axed (literally with a tomahawk) the damn thing. I did almost everything. From printing mods (ditched the buffer but didn't print any rewind spools -yet)  to using dribbling and skindipping methods. I had quite a few of issues but the most frequent one was/is the damn tips. I just let it rest for a couple of months just doing single filament prints and came back winter time. It turns out that the winter cold gave better tips an a few successful prints (no  intervention) . Yesterday i had a load fail that was due to the screw in the extruder that presses on the filament to be a little looser that needed and the sensor showed a 0. It was a nice suprise to see the cutter cutting through the filament and try again. this gives me hope for this nasty tips that refuse to load. There are more good days now than before but still a few bad ones.

My point is that even if there are a lot of people with success with, it there is a significant number with failures. And it shouldn't be a one to one comparison. Nobody count a product by the number of successes but the percentage of failures. Mmu2s has a lot of variables that can go wrong. A lot of pfte paths that can bend or collect something that causes friction (the worst for me was a slighlty deformed extruder pfte tube), ramming settings, cooling move settings, hot ambient temperature, cold ambient temperature, countries that are very warm others that are very cold,  finda random failures, sensor not sensing for one reason or another . Being a multi material device and not just multi color means that each of the different settings you have to play with you have to multiply them by each material you use.  The way mmu2s operates already adds a lot of time overhead even on to the simplest of prints. This overhead makes tinkering even more frustrating.

I didnt give up on mine yet, but im using the multi material less and less. It improved and im getting my hopes up with the latest firmware, dribbling (again thanks antimix) and getting a better grasp of fillament settings, but i have to agree that a product with so many disappointed customers and such a variety of errors cannot be considered a success. And prompting people to print mods to make a product work is not and should not be the goto solution. When you buy mmu2s you know that is going to take some work but for some of us it was a lot of work with minimal result. As a final note i tend to disagree with some that say that this kind of functionality normally it should cost 20k etc. There are dual heads out there, more expensive that mk3s with mmu2s but not that expensive. And is really a cheap excuse to say to someone that you should be happy that you car needs pushing every 5o km because a ferrari costs half a mil.

Posted : 29/05/2020 9:30 pm
ben_r_ and iracarmel liked
gnat
 gnat
(@gnat)
Prominent Member
Posted by: @iracarmel

How should I set the tension on various screws to make it work?

I agree with @nikolai-r that there is a lot of information out there about how to troubleshoot and fix most issues. It does take time and dedication to sift through angry posts (like yours) to find the answers that we've given. I also agree with @jiral that it sounds like a big part of your pain is your IR sensor. Personally I never found the settings menu helpful as you can't get to it when you need it the most. There is a thread in the Mod section about adding an LED to the sensor which makes life SOOO much easier.

I do agree with your point about the tension on the idlers though. There is nothing more frustrating that when you finally have gotten things tuned just right so that you can run through a few thousand tool changes with no issues, then you have to open an idler door and you are back to square one trying to get your tension dialed back in.

In regards to how pissed off you are at the thing, I was there. I spent a couple of months getting no where and was ready to give up on the whole thing, so I did. I switched to being mildly irritated by this pile of junk sitting in my office behind me, but overall just didn't think much about 3D printing for awhile. Then I needed to print something so I put the printer back together with the MMU disconnected. Finally I decided to try again and with some fiddling I got it working pretty reliably (maybe one or two interventions per print) and since have fine tuned my setup so that they only cause for interventions now is crappy filament breaking and causing problems. So try just stepping away for awhile and then return when your reserves of patience have refilled.

MMU tips and troubleshooting...
Posted : 29/05/2020 9:40 pm
jiral and iracarmel liked
andreas
(@andreas-t10)
Eminent Member

I really hate that 5 min edit rule. Its beyond silly. I  wanted to correct my last sentence but didnt make the time. I wanted to say that mk3s is cheaper than other printers but that doesnt afffect its quality. So being cheaper doesnt mean that someone should be content with constant issues.

Posted : 29/05/2020 9:42 pm
ben_r_ liked
iracarmel
(@iracarmel)
Active Member

@jiral

No, its not the IR sensor. I've tested it every time I've taken the hot end apart to deal with this.   Every time it jams up, I test the sensor.  Plus you can tell that it's working by the movement of the E axis as the filament reaches the bondex set.  It stops, does the heating, and then does the push down the the nozzle, and then never stops pushing filament.   I reflashed the whole thing, started from scratch (again), remade the PTFE, checked the paths on the MMU, and now it works as long as I don't try and change filaments.  Hooray.  I have zero faith in it.  

Also RE the person with the snide comment about wasting time with complaints.  I can tell you my ratio of searching for answers to time complaining.   10 minutes writing a complaint after 60~ hours of working on this thing gives a complaints to work (reading trouble shooting, surfing around etc) of 4.62x10-5 to 1.  Also blame the person having problems much?  Please don't reply again.  Am I complaining now?  I'm sure my ratio is getting worse by the second.  ._____.

I'm so good at taking apart hot ends now. I can do it in about 15 minutes flat (including heat up and cool down times.)  I honestly don't think that's the way it should be.  If you think the MMU is some kind of proof of Scottsmanship in the 3d printer world, I'm pitying you.  Why else would you defend a thing that doesn't reliably work?  Would you be happy with a car that popped out of gear 30-50% of the time when you came to a stop sign and then required 15 minutes spent fiddling around with it?  I certainly wouldn't appreciate it.   

The whole idea of crowd sourcing a product by using consumers as beta testers seems pretty crappy to me.  This is what the MMU is, it is a crowd sourced thing that doesn't yet work very well. 

For over  $1000 the i3 Mk3S MMU2S is a failure as a product.  

Eventually I will get this to work, but I won't recommend it ever.

Does the Mk3S make beautiful prints if I treat it as a single filament machine?  Yes, when it works, it looks amazing.  My first layer setup is measurably correct at exactly .2mm for a .2mm 9 square test on all 9 squares.   Adhesion is amazing.  

Design wise, I think it has serious flaws.  The PTFE tube seems overly complex, and is source of problems.  The channel guides for the filament in the MMU only have centering cones in one dimension.  

I was describing the problems of this machine to my wife, and I explained it as similar to a one sensor ice maker, but not being precise enough to work in such a manner.  Ice makers in refrigerators use a single temp sensor to operate by measuring how much the temp changes when water is added to figure out indirectly how much water has been added.  The filament sensor is like this, a single sensor, but it is  binary, not a range value.  It detects filament at a single point.  Wouldn't it have been better to put the IR sensor to reading a slotted wheel that was driven by the passage of filament? You could detect exactly how much filament went by, you could tell if it was jammed.  You could know exactly how much the bondex gears were slipping.  If you wanted more precision you could add more slots.  If you had a 1cm idler that was driven by passage of filament and 50mm slotted wheel with 30~ slots it would render .5mm filament passage resolution.  It would allow for much smarter programming of fail safe.  This is impossible right now.  You can detect that the filament entered the bondex set, but if there is a problem below that, you can't tell. Without analogue data about filament passage the whole thing is a crap shoot.  Without detecting actual passage of filament you depend on a whole host of filament related things being perfect or close to perfect.  This is not a robust design, and it shows.  This is why there so many search results about the MMU not working.  Systematically speaking, adding the MMU adds at least 7 filament transition points that can go wrong.  Murphy's law says, what can go wrong, will, and in this case it does. Considering that the MMU itself will grind the filament and cause a problem down the line is huge.  If it had filament motion you could write start code that tried to move 2 mm, and if it worked, you could move more.  I'm digressing. 

Meh.  I'm already regretting this post.  

-Ira

Posted : 29/05/2020 11:05 pm
gnat
 gnat
(@gnat)
Prominent Member
Posted by: @iracarmel

@jiral

No, its not the IR sensor. I've tested it every time I've taken the hot end apart to deal with this.   Every time it jams up, I test the sensor. 

When we talking about the IR sensor we aren't talking about the sensor itself being bad or having problems (though I have seen a few reports of that). It's getting it to line up correctly with the flag on the arm of the idler door. If you look closely at the design of the sensor and flag, there really isn't much room for error in the alignment.

The endless filament issue you describe means that it doesn't think there is filament at the extruder which is caused by the IR sensor not tripping correctly. Even though it tripped and stopped the filament in the initial load, if your flag is just at the edge of the sensor boundary even just a little vibration (e.g. the extruder motor stopping and restarting) can move the idler arm just enough that the flag is no longer tripping the IR sensor any more. It could be as simple as adjusting the tension on the idler door, or it may be (like me) you need to print a new idler door. Sometimes it turns out that the door wasn't printed very well by Prusa and simply reprinting the factory STL fixes it. Others (myself) had to get a modified version with a larger flag that we could sand down for a custom fit.

I've lost track of how many times I've read about people that followed the IR alignment procedure in the manual, but still had problems once things started moving around. My recommendation is to disconnect the MMU entirely and print a test print (a 20x20x20mm cube always worked for me, but the more jerkiness of the extruder the better) with the filament runout option enabled. If you can't get through the whole print without it thinking that you ran out of filament erroneously, then the IR sensor needs some tweaking still.

MMU tips and troubleshooting...
Posted : 29/05/2020 11:34 pm
jiral liked
nikolai.r
(@nikolai-r)
Noble Member
Posted by: @iracarmel

....

Also RE the person with the snide comment about wasting time with complaints.  I can tell you my ratio of searching for answers to time complaining.   10 minutes writing a complaint after 60~ hours of working on this thing gives a complaints to work (reading trouble shooting, surfing around etc) of 4.62x10-5 to 1.  Also blame the person having problems much?  Please don't reply again.  Am I complaining now?  I'm sure my ratio is getting worse by the second.  ._____.

....

No worries, I fully understand you're here to vent your anger and not asking for help or any information. Just FYI, this is from user to user forum trying to help each other.

Often linked posts:
Going small with MMU2
Real Multi Material
My prints on Instagram...
Posted : 30/05/2020 12:01 am
iracarmel
(@iracarmel)
Active Member

@gnat

Thanks, that helped.  

Posted : 30/05/2020 12:20 am
iracarmel
(@iracarmel)
Active Member

Gnat was correct.  The answer is that the IR sensor works fine when you test it at a lower tension, then you set the screw to the recommended depth, wham doesn't work, or works in an incredibly flaky fashion. 

Reminds me of an old database problem that would go away as soon as we turned on logging to find it, turn off logging to increase speed, wham came right back.  Of course it was some kind of hardware readiness issue.  Turning on logging to find it resulted in slowness that made the problem go away.

-Ira and the WOPR are playing post 90's internet forum posting.   Strange game.

Posted : 30/05/2020 1:14 am
jiral
(@jiral)
Eminent Member

^^ I would really recommend printing out the extruder door anew or possibly try extending the tick or latch, or whatever you want to call it, in a way that it is just a bit too long so that it always triggers. Then like gnat said, you sand it down stepwise until you get a very reliable on/off feedback from the senser, with hard and soft filament alike. You do that once and then all your IR sensor issues and hickups should be a thing of the past and the sensor should be robust against all sorts of different tension settings of the screw.

Posted : 30/05/2020 7:46 am
ben_r_
 benr
(@benr)
Trusted Member

I just got my MMU2S added to one of my i3 MK3S printers and I do think I'd have to agree that especially when compared to how easy it is to get the i3 MK3S printers up and running the MMU2S is a whole lot more finicky. And yes there is a lot of support and mods (which really shouldn't be needed) for it, I have found they only seem to help/fix the most common of issues and thankfully I haven't yet had any of the most common issues with my MMU2S. My issue with the globs in the wipe tower (LINK) doesn't seem to be a solved one nor very common. So all the hours of research I have done so far have been of little help.

My point is that Prusa has made a name for themselves as being the most reliable and easy to get up and running with lots of support and I have found that to be true for the i3 MK3S printers. I just don't see that same claim reflected in the MMU2S and neither to a LOT of others who have bought one. That says a lot right there. I feel more like those who bought any MMU models from Prusa are the ones doing field testing for them with all these mods, add-ons and work arounds while they work on a better functioning unit. But maybe thats the way it's been all along with their products? I'm not sure as I only got in the i3 MK3S printers.

If at first you don't succeed, redefine success!...
Posted : 04/06/2020 4:15 pm
gnat
 gnat
(@gnat)
Prominent Member

Ugh. Had it working perfectly. Then I had a jam due to the issue with the heatbreak around heavy retraction prints and now I can't get the tension on the idler door dialed back in. Tips are stringy again and now it's jamming due to heat creep from the extruder motor because I apparently have the tension too tight.

I'm finally giving up on Prusa's extruder design and have ordered Bondtech's extruder. That will take care of the heat creep issue and I think I'm going to give a normal e3d heatbreak a try too. So here's looking forward to a few months of fiddling again 🙄 

MMU tips and troubleshooting...
Posted : 04/06/2020 4:37 pm
nikolai.r
(@nikolai-r)
Noble Member

@gnat

I think this is usual summer time. Have at home now 82-86F in the afternoon and most of my printers (prusa/non-prusa) are struggling with certain PLA prints. This is the price we are paying by having quiet printers with not so powerful cooling fans. My solution to this problem is either to cool down the room or place a big fan pointing to the printer. That helps.

Often linked posts:
Going small with MMU2
Real Multi Material
My prints on Instagram...
Posted : 04/06/2020 5:00 pm
Page 5 / 6
Share:

Please Login or Register