Printing TPU with AMS

[NVNDO]

Hey everyone!

since yesterday I’m investigating this AMS/TPU drama because its driving me nuts not being able to print at least 95A TPU with my UNIT - Especially as slot Nr.1 works flawlessly, but Nr.2 always starts binding

Edit: I now know that all slots works, its just that normal Overture TPU dot NOT.
The HS version though works flawlessly!!

What I figured is, before finding a solution we might need to understand the actual problem here.

To mitigate friction being the problem I started to reduce the PTFE curves:

From 3…

To 2…

To just one:

Still the 2nd Slot would create a problem and having everything on sight, I noticed that meanwhile the feader-motor is pushing, swinging its arm back and forth…

…the AMS-motor does nothing.

Therefore the filament stays where it is…

So sure enough, considering the monstruos Spring being installed in the AMS:

No filament (especially TPU) is going to go past this. The filament starts to bend and curl as the spring is closing the gears too hard to let the filament pass through.

So that’s the reason its binding, but the culprit IMO is that the motor does not engage. If the damn AMS-motor would turn in sync with the feader-motor, the filament would be gripped by the gears and pushed through instead of being hold back.

Here some pictures:

My solution for now - looking for a weaker spring in order to let the feader-motor push the TPU past those gears of the AMS-motor without being hold back.

I’ll update as soon as I know more…

…3hours later:

I tried:

1. Using a smaller Spring in the AMS-motor (fail):
   

   
   
   
   
   
   
   
   
   Did not manage to push the filament through!!!

2. Using a bigger than prior, but still less strong spring in the AMS-motor (fail):
   

   
   
   
   
   
   Same as before. Some attempts got the filament slightly further but that was probably just randomness.

3. I changed the springs of the feader for weaker ones (fail):
   

   
   
   
   
   
   
   
   
   Same result!

I tried method 1+3, 2+3, just 3. No difference.

My verdict: Screw this…! XD
I’ll just buy HS TPU from overture or anything harder than that - that’s it for me with tinkering.

With best regards to everyone who may find a way.

PD: The paths are pretty wide IMO…

[PhillipM]

The path widths are the issue, along with the high tension - if you put teflon tubing in there to drop the clearance it will feed TPU with no problem - but trying to cut such thin tube and bond it for it to stay in position is nigh on impossible as soon as you want to do that with more than one feed.
The huge clearances and the square cross section make the part easy for them to make, but terrible for retraction - it’s also why some other harder filaments sometimes throw retraction errors. Really poor bit of design there, I’ve been working on fixing it for a while now.

It’s not and has never been a ‘feeding soft filament’ issue - it’s always been a retract issue right there in the AMS splitter.

[c5tiger]

I print a large tpu part I sell that would really be better with pva support so I need the ams. I am working on modifying the ams to use tpu and right now with a small sample size it works great with esun 95a. It switches back and forth between tpu and support with no issues 100% success on a small sample size, about 30 switches. I have a 100% fail rate (4 times) when I stop the print early and then press unload. It jams every time. What is it doing different when it unloads vs the normal retraction on a filament swap? Same thing happens with pp.

[FjolnirLabs]

After carefully reading many forums, I will be attempting to attack this disappointing setback of the AMS system upon getting my X1C.

My plan of attack, in case anyone wants to attempt it before I can, are these: 1) Modified paths piece for a more snug and straighter path 2) Run tighter and straighter PTFE tubes to avoid curves and potential catches 3) Register TPU as a PLA with TPU settings to bypass AMS blacklist 4) Wait for Bambu to fix this sad handicap.

I’ve printed standard Overture 95A TPU flawlessly on a stalk Ender 3 without needing to babysit it so I know its just a matter of tweaking the settings and adjusting the feed to account for the extra flexibility,

[5280]

Hmm, I’m surprised no one has mentioned this YT video with a claimed solution.

It seems quite similar to the spring swapping mods NVNDO performed. BTW - Thank-you for all the detailed photos!

[DzzD]

Probably because it is quite recent, but nice video.

I tried months ago to play/change springs without “full success”, it help but it is IMHO not a final solution.

PS: I still think that a good (maybe definitive) solution would be to use a dual direction extruder in replacement of the output one, but I did not have any time to work on this or any Bambulab stuff for a while now.

[5280]

@c5tiger

I am working on modifying the ams to use tpu and right now with a small sample size it works great with esun 95a. It switches back and forth between tpu and support with no issues 100% success on a small sample size, about 30 switches.

@c5tiger It seems you have had good success! Care to share want modifications you made that were successful?

[5280]

@DzzD

a good (maybe definitive) solution would be to use a dual direction extruder in replacement of the output one

Are you referring to the AMS feeder motor (vs. the spool feeder motors)? Based on your photo, it is both before and after the AMS feeder motor. Perhaps this motor is poorly synced with the spool feeder motors, and the printhead extruder motor - which I think you previously alluded to.

When you say “dual direction”, does that mean the current AMS feeder motor is only single direction (e.g., has a one-way bearing for retraction)?

I’m just trying to get a sense of what the actual problem(s) is. It doesn’t seem that the exact problem is yet known, so finding a solution is a bit hit and miss. For example, the YT video claims to have a solution, but fails to state what the problem is and how repeatable the success rate is. You and NVNDO have provided the best insight so far!

[5280]

@FjolnirLabs

My plan of attack, in case anyone wants to attempt it before I can, are these: 1) Modified paths piece for a more snug and straighter path 2) Run tighter and straighter PTFE tubes to avoid curves and potential catches 3) Register TPU as a PLA with TPU settings to bypass AMS blacklist 4) Wait for Bambu to fix this sad handicap.

Nice plan of attack. Please share your findings! Based on this thread and the YT video in this link, my plan of attack is similar:

1. Replace existing PTFE tubing with Capricorn tubing (thanks DzzD!) for tighter diameter and reduced friction.
2. Shorten tubing path length for less friction (thanks NVNDO!).
3. Redesign the filament hub (“merger” section) with round, narrower profile so the filament cannot kink as much. Also narrow the input paths to the hub (“merger” section) using a round instead of square profile (thanks PhillipM!).
4. Remove springs on the spool feeder motor gear lever (thanks YT video!).
5. Replace AMS feeder motor lever spring with small springs (thanks NVNDO & YT video!).
6. Redesign the filament merger so the spring tension can be changed without disassembly of the AMS.
7. Redesign the AMS feeder motor lever so that it does not close completely. This, combined with a less forceful spring, should make it easier to feed filament past the AMS feeder motor gears. As I haven’t taken the AMS apart yet, I’m not sure how much the feeder gears close, but my guess is all the way against each other. If so, they shouldn’t need to and a stop (perhaps adjustable) should allow the filament to spread the gears with less force.
8. Use the OrcaSlicer to bypass any blacklist problems.
9. Hope AMS Rev 2 is compatible with TPU.
10. If none of the preceding work and/or are not reliable, use Overture HS TPU (thanks NVND)!).

It may be a while before I can perform any mods as I have other workflow problems (non-Bambu) to resolve first, but I hope to eventually make the AMS TPU compatible.

[DzzD]

From my tests (and I did a loooot of a year ago), the only problem I have encountered was that when the flexible filament is pushed out of the AMS because the output block is not designed to works well with non solid filament (this can be mitigated by playing with the tensioning spring but it is not reliable), my plan is to pull it from outside using a third party extruder.

I have studied a bit the electronic and the way the AMS works (at least the output block) last year too, and it does not appear to be that difficult, the output block is pretty simple, but simple does not mean fast to make and unfortunatly I have no time free to work on anything since a cupple of monthes, but if an easier solution emerge that would be great.

[5280]

@DzzD, thank-you for the insight! I just noticed you had also posted more info in your other thread.

I ordered Capricorn tubing today as a start.

[c5tiger]

My modification only works if you have multiple ams units. One of the main problems is the filament kinking in the internal ams hub on retraction. I designed and printed a part to block all but one slot so flexible filaments don’t have any alternate paths. I have one ams that will print any single filament and the other ams for any 4 ams compatible filaments. Takes about 10-15 minutes to install and a little less to remove to go back to using all 4 slots.

Picture of cad drawing

[5280]

@c5tiger Clever solution - Thank-you for explaining your mods! Too bad it effectively renders 3/4 of the AMS useless. However, this feedback is very helpful. I’m think I am finally understanding the problems better.

Essentially, the shape of the hub (“merger” section) is not critical when extruding as the filament is in tension and the filament takes the shortest path. However, in retraction, the filament is in compression and any resistance will cause it to kink, particularly in the hub. Tighter tolerances in the hub and a round cross section instead of square may help reduce the kinking when retracting.

I’m curious about the behavior of the three motors when extruding and retracting. For example, do all three motors (print head extruder, AMS feeder, individual spool feeder) rotate together in sync for both extruding and retraction? It seems they should. If not, that could also be a problem source. Perhaps the individual spool motors only feed to the AMS feeder motor. I have to pay attention more next time I print something to the behavior of the individual spool motors. Regardless, I hope that motor synchronization is not a problem source as that would be more difficult to resolve.

Anyway, thanks to everyone who has contributed thus far and taking time to share photos.

[just4memike]

Anyone use the Smart Materials INNOVATEFIL TPU HARDNESS+ or Smart Materials INNOVATEFIL Carbon Fiber TPU? Matter Hackers sells the CF TPU.

Their TDS’s don’t have a Shore rating but by there names I would expect an extended additional level of stiffness to the filament.

[just4memike]

Ok, the Smart Materials INNOVATEFIL Carbon Fiber TPU I ordered arrived from Matter Hackers overnight, and it’s definitely compatible with the AMS (Shore D - 67). I didn’t stick it in the trouble slot on the AMS, but it’s so stiff, I can’t imagine it being a problem. It feels a lot like a slightly flexible PLA CF, however, it’s still a TPU. I printed an 80mm long x 5mm tall, and 20mm wide test coupon and it can bend over on itself fully. However, it is stiff, so for those of you looking for gummy soft TPUs, this isn’t it. It definitely took some tuning, but that only took 5 short prints. Unfortunately, it’s a uber slow printing filament, I set the flow to 4.5mm3, but it should probably be around the 3.5mm3. I’ll adjust it down if needed though.

I wanted a TPU that the AMS can handle to do some multi-material printing so I’ll play around with that, but I do think this should work out. PETG/TPU or ASA/TPU and PLA/TPU (to a lesser extent) work well together. It should be useful to mix the properties together on some projects.

Some things to know about it. It’s not soft in any way, but it is flexible. Think flexibility like a car’s fiber reinforced radiator hose (but a little stiffer and no soft touch). If modeled correctly, it can make very rigid parts without insane infill or walls. Equally, if modeled correctly, if can make relatively flexible parts.

I wouldn’t use it for gaskets like a normal TPU. It really feels like the texture from the CF would be too lossy for a tight seal.

The last thing I would think is notable is its cost. At $40 for half a kg, it’s not cheap.

@BambuLabWikiTeam

You can disable this feature here and it won’t alert you anymore.

Unfortunately, this didn’t work to allow me to print from the AMS. Tried it in Orca and Bambu Studio, it simply displayed the following error and looped back to the “Send” print dialogue box.

I was able to clear the filament info from the AMS memory slot in the slicer and then the AMS Blacklist override would work (shown below). But if you tell it there’s TPU in there it will error out. That “Blacklist Check” override might need little more tweaking.

[Hank]

@just4memike

Thanks for the tip - Shore D68 wow. I’ve been thinking about opening a post to see if anyone else knows of a harder TPU than the D58 from Extrudr. This week I also reprinted a container lid - someone dropped the PETG, now it’s made of hard TPU and you can throw the lid against a wall. Making two label boards that are now made of ASA instead of TPU (UV exposure) TPU would be much easier.

INNOVATEFIL - interesting, they are from spain. I’ll have to take a closer look at what they’re doing.

AMS TPU embedding - well, they have to know that themself. 2 officially alternatives are announced, one of which may will hold the shipment and one of which is already overdue (Co Print). Those who may hold shipment also exclude TPU on there AMS (but the printer is focusing massively on TPU, finally bringing some printing speeds by TPU - because Bambu also becomes a snail with TPU prints)

And finaly even two more manufacturers may soon announce AMS too - so there will be a lot to choose from without any hack… So TPU integration will probably become more of a take-it-or-die issue. So I don’t have to be interested in whether Bambulab will soon integrate the TPU into the AMS or not. They get it or they get lapped, so there isn’t much more to say about it. And if than people want to spend their time defending it like others defending other who have been lapped, then they should feel free to spend there time doing so

This has nothing to do with the TPU issue, more how to spend his own time: I was also shocked this week that the Two Trees SK1 has now found its way onto the shelves of official resellers in the EU zone. It just proves, you will never win alone - the others are simply too many…and simply deal with the problems better since it there competence - e.g. Sell ​​printers off a shelf and to look the customer in the eye when problems arise. So it’s about whoever wants to remain a partner has to deliver, announce this over time, serious without barking in a collaboration and with open communication.

[just4memike]

@Hank

bringing some printing speeds by TPU - because Bambu also becomes a snail with TPU prints

This is mostly from the TPU’s formulation. BBLs 95A HF is on par with any fast printing TPU. Some TPUs just can’t flow. Hopefully, those days of slow TPUs are mostly behind us, because many filament makers are providing a high flow variant that matches the times. Unfortunately, Innovatefil isn’t one of the high flow variants.

@Hank

two more manufacturers may soon announce AMS too - so there will be a lot to choose from without any hack

Yeah, looking forward to competition pushing this idea further. Creality and Phrozen have the budgets to really push things along. My problem is, the AMS is still mostly for show and trinket type printing (and of course, it’s a huge convenience as a filament hopper). Takes a lot of hacks to do the interesting stuff with it, and that’s a pain, because everything is so locked down.

As far as BBL introducing an AMS that is fully compatible with TPU, that’s gonna take a serious effort. Pushing and pulling the wet noodle consistently through 1-2 feet of tube is going to take a lot of engineering. Hall sensors and extra drive motors might help, but they would need to be implemented very clever.

[Hank]

Well, the next shopping date is probably next Black Friday. Everything else is already set in stone for me.

Let’s see what happens until then… For spring shopping, only Flashfroge has delivered and 2024 is already almoste gone.

[BambuLabWikiTeam]

Thanks just4memike for your feedback.
We will pass on your request about the TPU printing limitation to the relevant team for future upgrade consideration.

[Hank]

@BambuLabWikiTeam

just so we don’t misunderstand each other. Every printer, no matter what brand, which can’t change materials (not just color). Regardless of whether it`s an FDM or an SLA printer will be may in the year 2025 just scrap metal. SLS is at 2999 USD, becomes more user-friendly too.

I know you can do it and you will manage it. Playing around, you definitely won’t have time for that anymore.
Sorry for the clear words - but being nice never helped anyone. It is simply the current development. An extreme growth market also brings very quickly a lot of innovations too

How we BUILT a 3D printer that can print ANYTHING!