An incomplete list of MMU2 failure modes I've seen  

Re: An incomplete list of MMU2 failure modes I've seen

After doing a Bowden length calibration. Hit middle button to save length. Then hit reset button on MMU2 as per manual, to exit service mode.

On several occasions after doing the above calibration, 1st filament fails to be pushed from MMU2 unit on a new print. Mk3 registers a filament load failure and prompts user to unload filament (that isnt there). After user presses button to remove fantom filament, mk3 extruder again tries load filament but only moves bondtech gear enough length as if filament was on top of mk3 extruder not the length required to move from MMU2 unit. I just end up resetting mk3 to clear this. After reset mk3/MMU2 functions as it should.
(No mechanical filament issue at MMU2 when this occurs, it's firmware related. After hard reset, filament is pushed to bondtech)

Re: An incomplete list of MMU2 failure modes I've seen

The filament is loaded properly by the MMU2, but the Bondtech gears don't engage with it. I think this is caused by a slightly misshapen tip on the filament, but this failure isn't detected at all. The printer will continue on its merry way not printing the filament. Often this clears on its own after a few missed layers, leaving a visible gap in the print. It might be preventable by having the MMU and Bondtech both turning at the same time for a short distance on load, allowing the MMU to push the filament a little further and ensure the filament is loaded enough for the Bondtech to engage.

Ive had alot of issues with MMU2. Most my fault, either finding proper tensions for idler body or bondtech gears, or using filament that I haven't found the correct ramming settings for. (Inland PLA). But this is a problem that still occurs even with prusa filament. This is the reason I have yet to complete a sheep without a missing color on a layer.
It would seem that along with your idea of having both MMU2 and bondtech gears moving at once for a short period to help force the loading of filament that will not necessarily be a perfect shape...
The logic for filament sensor needs to be tweaked to catch this condition. Currently it doesn't and u end up with a missing layer of one color on a layer. Seems the filament sensor is failing to detect a lack of movement. Prior to this problem I have never received the warning sensor performance low, disable filament sensor.... So I assume the sensor is seeing movement of filament.... Is the filament movement detection logic on during filament changes of MMU2?

If u have multiple filaments colors. Potentially 5 colors. I've been loading the darkest colors in 1st filament position. Is the detection logic the filament sensor uses to see moving filament dynamic ??? Say u happen to use a dark color as your 1st color in a mmu print. Will it then have trouble detecting movement on lighter colors because the 1st color it saw was dark???

Re: An incomplete list of MMU2 failure modes I've seen

The logic for filament sensor needs to be tweaked to catch this condition. Currently it doesn't and u end up with a missing layer of one color on a layer. Seems the filament sensor is failing to detect a lack of movement. Prior to this problem I have never received the warning sensor performance low, disable filament sensor.... So I assume the sensor is seeing movement of filament.... Is the filament movement detection logic on during filament changes of MMU2?

It's my understanding that the optical filament sensor in the extruder is disabled with the MMU because the optical sensor has proven to be unreliable.

Which is too bad, because using the FINDA in the MMU in conjunction with the optical sensor in the extruder would (in theory) trap more of these failure modes.

Supposedly the new firmware improves the reliability of the optical sensor. If this is true, perhaps a future firmware revision will re-enable it with the MMU and make the system as a whole more fault tolerant.

Re: An incomplete list of MMU2 failure modes I've seen

It's my understanding that the optical filament sensor in the extruder is disabled with the MMU because the optical sensor has proven to be unreliable.

If you are correct the mk3 has zero chance to catch this condition. As long as finda on mmu2 registers filament present, mk3 thinks it has filament to print with. (Needs a patch right away)
Let me know if u make a GitHub issue about this. I'll post my experience as well.

Re: An incomplete list of MMU2 failure modes I've seen

I have completely the same issues as you all have.
Contacted Prusa about this but did not get any response yet. The filament sensor is disabled which is kinda problematic for me. If I have to help the mmu each 1-200 changes ok, not really nice but better than a crashed print.
Point 2 from the original post with the missing layers is the worst and most common issues with the mmu which causes me to stop most of my prints as they would not yome out nice.

Prusa please enable the filament sensor or at least place an option in the settings to enable it!

And thumps up on the sync motion of mmu + extruder to prevent that issue from appearing.

Re: An incomplete list of MMU2 failure modes I've seen

Final 3.4.0
https://github.com/prusa3d/Prusa-Firmware/releases/tag/v3.4.0

Detailed description of changes (compared to 3.4.0-RC2):
Multi Material unload improved:
Filament unload now contains ramming sequence. This ensures that tip of unloaded filament has no significant "strings" and thus overal reliability of printer in multimaterial mode is improved (stringing during unload could sometimes cause false FINDa triggering or jamming).

Gonna install now and try sheep again 😀

Re: An incomplete list of MMU2 failure modes I've seen

I caught one of these in the act last night, paused the print, and opened the extruder up to see what happened.

It turns out my theory was wrong. When I took things apart I found that the filament had been jammed by a huge hairball of filament strings clogging the little PTFE tube above the filament sensor. This had obviously been building up for some time, and I'm guessing that sometimes the incoming filament pushed it all out of the way and loaded, and other times it didn't. That explains why the print will work for a while, then miss a few layers, then start working again. I think this hairball had been collecting around the optical sensor, where it doesn't have any place to go and it's hard to find and remove.

Also, I had a new one:

Re: An incomplete list of MMU2 failure modes I've seen

I still wish they'd use the filament sensor in mk3 when mmu is in use, to catch the lack of movement of filament into mk3's extruder. If the cause is bondtech gears not engaging or strings blocking the filament path the out come is the same.....missing layers...

I without a doubt experienced the strings causing jams with inland pla

Re: An incomplete list of MMU2 failure modes I've seen

I still wish they'd use the filament sensor in mk3 when mmu is in use, to catch the lack of movement of filament into mk3's extruder. If the cause is bondtech gears not engaging or strings blocking the filament path the out come is the same.....missing layers...

Absolutely. It's preferable for the firmware to detect an abnormal condition and pause so the operator can clear it.

I've found that if the printer pauses when there's an MMU problem, I can very often fix the issue and continue printing without a visible defect. Of course it would be even better for the printer to clear the problem itself or not have a problem at all, but baby steps...

Re: An incomplete list of MMU2 failure modes I've seen

Created a ticket on Github, please add your comments to the ticket so Prusa will hopefully work on that as soon as possible as we cannot use the MMU as it is supposed!

https://github.com/prusa3d/Prusa-Firmware/issues/1163

Re: An incomplete list of MMU2 failure modes I've seen

Created a ticket on Github, please add your comments to the ticket so Prusa will hopefully work on that as soon as possible as we cannot use the MMU as it is supposed!

https://github.com/prusa3d/Prusa-Firmware/issues/1163

ok replied
mylife4aiurr on github

Re: An incomplete list of MMU2 failure modes I've seen

Also replied. I cant use my MMU in current state, too many problems and errors 🙁

Re: An incomplete list of MMU2 failure modes I've seen

PavelSindler from firmware team responded to GitHub.
Thank you for these information. We will look on it closer

Re: An incomplete list of MMU2 failure modes I've seen

Very helpful video on MMU2 trouble shooting by Chris Warkocki:

Re: An incomplete list of MMU2 failure modes I've seen

Very helpful video on MMU2 trouble shooting by Chris Warkocki:

Somebody needs to buy him a copy of iMovie. As much as I wanted to see his troubleshooting tips, I didn't want to sit through almost two hours of video.

Re: An incomplete list of MMU2 failure modes I've seen

Yea I agree on length. Most of the good stuff is at the beginning. Basically video trumps a manual.

Bondtech gear alignment and filament resistance (yes yes). But the tips and tweaking, seeing him do it, his technique. (mine needed adjustment 🙄 )

The manual says put a piece of filament in the selector and adjust finda until it hits filament and then back out finda probe with 90 degree turn or something similar.
vs
Use filament with a blob on the end and keep running it back and forth checking for resistance and finda clearance. Lock in place when no resistance. To see him explain what to feel for and demonstrate the process was miles better. (mine needed adjustment 🙄 )
Turns out a big problem I was having was my finda was so low. Clean new filament passed fine. The manual isnt clear enough here - how will they know the size of a blob the user will have. So once the filament retraction happened with the blob on the end, it would get caught on the finda.

Ramming - Showing us what a good tip is suppose to look like. And the factors that influence this. This video will help miles for users trying to tweak filaments Prusa hasnt tested.

Yes it was long, but lots of good info.

Re: An incomplete list of MMU2 failure modes I've seen

The filament unloads but the FINDA doesn't register the unloading. This may be because of a bit of debris in the path, a filament thread that keeps the ball from dropping, or the ball just gets stuck for unknown reasons. The printer will ask the operator to clear the problem. Usually I will recut the tip of the filament and remove any debris I find, to try to keep it from happening again right away.

I think my pinda was to low, so on the retraction it would get caught instead of unloading + Tweaking ramming for better tips >> hope following the video tips helps this issue.

The filament is loaded properly by the MMU2, but the Bondtech gears don't engage with it. I think this is caused by a slightly misshapen tip on the filament, but this failure isn't detected at all. The printer will continue on its merry way not printing the filament. Often this clears on its own after a few missed layers, leaving a visible gap in the print. It might be preventable by having the MMU and Bondtech both turning at the same time for a short distance on load, allowing the MMU to push the filament a little further and ensure the filament is loaded enough for the Bondtech to engage.

Bowden length calibration, correct bondtech alignment as demonstated in video, and better formed tips >> hoping these tips clear this issue up for me. But clearly the mk3/mmu2 should be able to catch this. They are expecting the user to have a finely tuned machine instead of actively checking for filament in extruder

Re: An incomplete list of MMU2 failure modes I've seen

Bowden length calibration, correct bondtech alignment as demonstated in video, and better formed tips >> hoping these tips clear this issue up for me. But clearly the mk3/mmu2 should be able to catch this. They are expecting the user to have a finely tuned machine instead of actively checking for filament in extruder

The bowden length calibration helps but won't eliminate this problem. The root cause is that the FINDA senses the end of the filament at a different point than where the bondtech grabs the filament.

A well-shaped filament tip should be a stubby cone. The FINDA senses the end of the filament at basically the very tip of the cone, but the bondtech engages a little further up, maybe where the cone is 75% or so of the filament diameter. If the cone is stubby, there's not much distance between those two points (maybe a millimeter or so). But if the cone winds up a little longer, you might get 2-3 millimeters between those points. So a bowden length that's correct for a properly shaped filament tip might occasionally fail to engage because the printer can't precisely control the shape of the filament.

Re: An incomplete list of MMU2 failure modes I've seen

Knock on wood....
I think I turned a corner with my MMU2.
10 hour Print with Inland PLA....no issues

Not a single jam @ extruder
Not a single load or unload issue at MMU2
I think the video really helped me tune my MMU2, along with a better profile for Inland pla