Semi-Flexible Filament and the MMU2S (A guide and mostly-success story)  

Semi-Flexible Filament and the MMU2S (A guide and mostly-success story)

When I placed an order for the MMU (2s), I knew what I was getting into.  Alongside nice looking prints of multicolor PLA models there were horror stories of bad feeding, skipped layers, and general unreliability that made it clear the MMU was anything but plug-and-play.  And since I knew from the beginning that I wanted to print in TPU, I knew that 'stock' probably wasn't going to cut it.

Rather surprisingly though, I don't see much information here or elsewhere on the subject.  So I set out to get my flex on.

And about two months later I've reached a nice stable plateau.  Sure, I still get failed prints, and I'm not using the really flexible stuff either (just 95a), but as there's very little information on printing flexible filament with the MMU2s I thought I'd share my experiences (and maybe learn some tricks that'll bump the success rate up further).

The journey started with printing TPU on a regular, non-modded MK3s (my first Prusa and third printer overall).  I'd heard lots of talk about how difficult TPU was to work with but I didn't see this firsthand, so I guess the MK3s was just perfectly designed for it.  The default printer profile was a good starting point, but since it was referring to a different material with a different hardness (98A as opposed to 95A, which all of my hatchbox or noname brand TPU is), some adjustment was in order:

• The extrusion multiplier of 1.2 was too high; rather consistently the material would build up, the nozzle would jam (usually around the same layer height), and the rest of the print would be air.  Reducing this to 1.1 fixed that problem.
• The regular flex profile prints at 240 degrees.  I found that most of my filament was more comfortable in the 225-235 range, and that 235 first/225 other resulted in the least amount of exacto-knifing later.
• The K value (set in the custom start g-code section) was set to 10.  I found 125 to be a bit closer to reality (but since the MMU has a bowden tube I guess this adjustment is moot).

(Incidentally, I've never seen reasonably priced brown flex filament.  The only brown I've seen has been ~80/key, and seems like people only make that color for prosthetics.  I wonder why most people don't make brown...)

With these settings, and with ColorPrint, I was able to make cool colorful tpu objects.  I even increased the max volumetric speed as things were going pretty damn good.  But the MMU, despite the hefty price tag of a bowden tube and more general limitations, promised me coloring options that ColorPrint couldn't muster.  Eventually I cracked open the kit I ordered and put it on.

The MMU proved frustrating (and that little blade drew first blood--I've since removed it as it never seemed to help anyway).  I was still able to print my TPU in single filament mode, but actually using the MMU features never seemed to work.  I eventually figured out what the pain points were and how to fix them, and some required new parts:

• Resistance kills MMU + Flex.  Make sure there's as little resistance as possible.
  • I didn't use auto-rewind spool holders.  I actually used the default MMU trays to good effect, and eventually moved to an elevated closet rod.
  • No filament buffer.  TPU doesn't automatically curve and tangle like hard plastic round on a spool.  When using the trays I unspooled a few feet and let it sit on my workbench, which worked well for the colorprint style prints that only changed over 1-2 times per print.  When using the closet rod, well, it just hangs a bit but doesn't get tangled.
  • No ptfe tubes held in with screws.  There are several designs on thingiverse that use fittings and they're all a huge quality of life improvement.
    • https://www.thingiverse.com/thing:3233579 (the one I used, they're probably all good)
  • No overly long or crappy ptfe tubes.  You don't need three feet of ptfe in my opinion, and if your tubes are scratched or bent you'll need to chuck them.  Some people have reported the tubes provided by Prusa have too much resistance; I didn't notice anything wrong with my set personally.
    • You do need SOME resistance or the filament will fall out of the MMU.  I cut my tubes to about 6 inches.
• The FINDA probe can be finicky
  • Try using an alternate MMU2 Selector.  I didn't actually have too many issues here, but every little bit helps.
    • https://www.thingiverse.com/thing:3688458 (There are lots of these, it's hard to choose)
• The filament sensor cannot handle flexible filaments at all
  • Use a larger tower.  It looks funny, and I had to do some sanding and adjusting before I got it fitting just right, but it now detects the filament every time.  This DOES make maintenance harder, and if your filament gets wrapped around the gears it's an even bigger pita than usual.
    • https://www.thingiverse.com/thing:3554066
• Failures at the hot-end are usually fatal.
  • Disassemble the hot-end and chamfer the ptfe tube that sits just below the bondtech gears.
  • Put a little collet clip in the collet that's on your heat sink (didn't have a problem here myself, saw threads from others that suggested it couldn't hurt and might help).

Usually when my prints fail or miss layers it's because the filament made it to the gears (triggering the filament sensor) but wasn't making it out of the nozzle (I thought the feed-retract-feed step was meant to catch those circumstances but I'm not too clear on the process).  I really wish I could see inside the hot-end to figure out what exactly is happening, but lacking superpowers I instead have to use deductive reasoning.  And I've deduced the problem is almost certainly bad tips causing the filament to get stuck.  That's mostly solved with slicer settings.

Prusa says they'll support flexible filament on the MMU one of these days and that they plan to do it with just software / firmware updates.  I'm not willing to wait so I made my own MMU compatible profiles.  They were mostly built with trial and error, and there are a few more complicated settings that would probably help if I knew their effects better.  Still, these are what worked for me.  At the very least I've identified settings that had a big impact for me, which hopefully means less searching for someone else later:

• Printer Settings
  • Extra Loading Distance is set to -25!  A thread on these forums pointed out that basic math dictated it should be -15 (which is where I keep it at for petg), and I eventually settled on settings closer to -5 for TPU.  This value being too large (in the absolute value sense) resulted in a nasty, gap filled wipe tower which on longer prints more often than not ended up failing and taking the print down with it in one way or another.  Values closer to zero often resulted in blobs on the wipe tower, which was a more manageable problem for me, and there's a sweet spot in there somewhere.
• Print Settings
  • I kept my .1mm layer height profile despite this not being an official MMU setting.  This may have resulted in more issues for me, but I think the main fixes are still valid.
  • Wipe tower Maximal Bridging Distance is set to 10mm.  For TPU, which doesn't really 'bridge' per se, I found that large values were bad, and much like the extra loading distance, eventually resulted in a messed up wipe tower that was just begging to screw up prints.  Not that I like wasting filament on fill, but 1mm or less was about as high as I could go and not suffer.
• Filament Settings (aka Da Big One)
  First I copied my regular TPU profile and changed the reqs so that Prusa Slicer would permit it on the MMU.  Most of my work here was focused on getting good tips--whenever the thing started air printing the tip I pulled out was flat and ugly...
  • I knocked down the extrusion multiplier to 1.05; this unfortunately resulted in a few more obvious seam lines where before there were none, but zits and blobbing were reduced, meaning fewer random failures.
  • I was less willing to deviate from a nozzle temp of 225 for non-first layers.  Seemed to result in more tips failing to feed.
  • The toolchange parameters are where the magic happens.  Unfortunately, I couldn't find much information on what practical effect the settings had and the presets--which were all for hard filaments--mostly differed only in the number and speed of cooling moves.  This is the area that I could use practical information on; I can't find jack on ramming settings, which pops up a scary yet enticing dialog when clicked.
    • I started with the default settings stored by all of the non-MMU2 profiles.  Notably they have four cooling moves.
    • TPU doesn't like to be loaded fast.  I used 1.5/9; this was a marked reliability improvement over the default of 3/28, though I think once you get below a certain threshold the actual number doesn't matter too much.
    • TPU DOES like to be unloaded fast.  I kept the unload speeds at 100/90, and any attempts to lower them resulted in a noticeable increase in failure rates.
    • I'm not convinced it helps, but I added a 2 second delay after unloading since that setting was clearly designed for flexible filaments.
    • I normally use retraction with my flex filament as it really helps prevent oozing.  But disabling retraction also improves reliability.  In this case, once I had other settings dialed in, I re-enabled it because the streaks the oozing tpu left ruined my designs.  But starting out I overrode this value and it helped.

That's my story, and I am pleased with the results, though I really want to crack open that ramming settings option and learn how exactly that's supposed to work.