Printing TPU with AMS

I never tried to print TPU with the AMS as it seems lot of people had trouble trying this but… I was wondering if it would not be possbile to fine tune setttings to enable this without too much troubles.

I would really love to make part made of both solid plastic and TPU without using the clipping parts method.

So does any of you succeed in reliably printing TPU with the AMS ?

One thing I’m thinking of would be to replace the original white PTFE tubes with the blue Capricorn tubes, it has helped me in the past with bowden & TPU, to use this type of PTFE tube which are more slippery and have an inner hole a litte more tight.

1 Like

I don’t use AMS because I think the Capricorn tubing increases rather than decreases the problem with soft TPU. With a direct extruder, the TPU is pulled and stretched. With the AMS, the extruder is in the AMS to push the TPU to the direct extruder of the print head, and that is the problem. The moment I push 30 cm of soft TPU all the way through the tube, the TPU is compressed and clogs the tube before the TPU can reach and be pulled by the printhead extruder.

Now imagine the compression and retraction with each change. What a strain on the AMS motors and the hose along with the connections. A hose with even less internal play doesn’t make it any better :wink:


I can assure that I get better result & reliability with Capricorn tube on a bowden printer while printing TPU. But well inded there is not so much retractation & pull back in comparaison of AMS :roll_eyes:

Note that Capricorn are not only more tight it is also more slippery like greased, but inded not by greased, just the material is more slippery.

the TPU is compressed and clogs the tube before the TPU can reach and be pulled by the printhead extruder.

This is why I talk about Capricorn, TPU have less space to be compressed

EDIT : on my other printer the bowden tube is as long as the one from the AMS and TPU print well


Bambu specifically say to not use TPU in the AMS. You should not try it. you’re going to screw up the AMS, its too soft. That’s why so many people are printing the Y splitter. To switch to the single feed for things like TPU.

1 Like

Does not seems true for all kinf of TPU, so with better PTFE tubes… I really want to give it a try :rofl:, I already have filament that breaked inside the AMS, so I know hos to disassemble/reassezmble it, it is not that hard.

1 Like

Nothing is stopping you. Just lie about the type of filament you put in the AMS (and if you want to use Bambu TPU maybe remove the rfid tag),

Well, I managed to make multimaterial prints with TPU using the AMS !

Using ESun TPU95A Green (as a side note this translucide green color is ugly)

Some notes :

  • I reduced PTFE length as exlpained here
  • I found that TPU can stuck if the extruder push too much on it and it is rewinded, then it often get stuck at the output of the AMS, a good idea is to use a very low “Max Volumetric Speed” for the TPU
1 Like

Even if it works, it is not reliable and often stuck, I will try with TPU shore 98A, anyone already try this shore ?

Here is one of the issue that can happen when printing TPU using AMS.

Note that I did not unassemble it to remove the filament, when this happen, you can remove the filament without unassemble anything, just pull hard the filament on both side to make it more “flat”, and then remove it by pulling it from the spool side. I unassembled it without removing the filament to understand what was hapening and where it was happening.

When this happen, it become impossible the filament can be pushed out by the motor of the AMS, or even pulled back, and it stay stuck, by releasing the pressure of the spring a little it is a little better as the gear slip rather than bending the filament but as it then eat the filament this is not ideal either.

Despite what the picture is showing I suppose that the bending happen between the four hall sensors and the motor gear, not after, there is not enough space after the gear to enable the filament to bend like this.

Not sure but I think that at some point the spool feader go faster or push filament more hard than the AMS ouput motor does, and then it bend the filament in the space just after the hall sensors.

I wonder if that removing the motor to let the gear turn freely and to keep the speed meter active can works if the PTFE length is keep short enough, but I am tired to unassemble this thing :slight_smile:

If you plan to do so, be aware that you will need some M2 screews, some are shorters than the original design was targetting for and will become loose

Here are the new ones (wich go all the way trough) I used side by side with the old ones used by Bambulab


1 Like

Any updates on that ?

Nop, I did not had any free time to work on this more, but still in my todo

I would love to collaborate with you on this. I am almost ready to switch to Capricorn tubes and give the experiment a try with 4 different color TPUs at the same time. I’m am trying to use 100a or 65d TPUs to start with and then try to dial down to 98a if everything goes well. I keep telling my self this has to be possible some how and we only need to do the work. Also, not sure if Varioshore TPU from Colorfabb would be ideal. It is a 92a TPU but colorfabb claims that it is as easy to print as PLA because they use an additive that makes the TPU expand when heated and makes it soft. So I wonder if when the TPU is cold it is stiff enough to bypass any of the tube, ams, and filament buffer issues that can cause the TPU to jam.

1 Like

Hehe :slight_smile: I have tried to put a filament to fridge to see if it help, just to know, spoiler alert, it does not really works and filament pulled from the spool get back to room temperature very quickly. But to be true this was just for the love of science, never thought that it may be a real solution, but I just wanted to know.

Anyway you are welcome to post all your ideas here or PM me, I am pretty sure that the main problem does come from between AMS spool feeder and AMS output motor.

My main idea from the start is to make some kind of buffer between those two motors (spools feeders motor and AMS output motor)

EDIT : I already have a first design that I will test this WE if I get enought free time.

1 Like

is it possible to slow down AMS speeds when TPU is used

I think that the problem when using TPU is too fast feed speeds and the wire is pushed too strongly into the hot end. therefore the TPU twists and gets stuck.

can we adjust the AMS operation? as far as I know we don’t.

You probably right, and this is probably something like this which is happening, the output motor pushing too hard and twisting the filament (I have posted a picture of the internal of the ouput motor unit somewhere upper in this thread that show this problem)

No there is in my knowledge no access to the AMS controller software, but we have access to the data bus and to some other parts and connectors on the AMS board. I have not that many time to work on this, but I work on it a little regulary each week, actually, I am looking for a solution that would not imply too much works or cost, I dont want to re-build a full AMS :).

The next try I will do, will probably to use an external extruder at the output of the AMS, if I can not get it working with a more simple solution.

In my case, with TPU, I never had any problem other than in the AMS motor ouput unit, so I am concentrated on this unit and how can it be modified.


Below is my last attempt, the idea is to enable some kind of buffer between AMS spool motors and AMS output motor, spoiler alert “this does not correct the problem” but I think that in some way it help as it enable to better handle a wrong synchronisation between different motors (spools and output motors), I have to try such buffer after the ouput of the AMS to avoid TPU filament (or any other filament) to be pushed with force when it have no more space in the PTFE tube, this little system couppled with a spring or elastic is intended to let some free space/simulate an extendable PTFE tube (to avoid permanent deformation of the filament).

Some news,

I just ordered that on Aliexpress :slight_smile: :

I am convinced that the AMS output motor is not strong enought, it is audible that as soon as the spool feeder motor stop, the AMS output motor is forcing a lot too much and filament is moving too slowly IMO (I have tested with two different AMS and three diffrent AMS output motors units), it is even more noticable with TPU filaments (also spring tensioning of this extruder is too strong for TPU, this may also explain the problem).

When unplugging the tachometer of the AMS output motor everything works better as the spool feeder motor does not stop and both motors are working together, but then problem is that the AMS get crazy and indefinitly load & unload the filament.

There are many ways to improve the printer, but you can use a new material developed by our company Moisture-Activated Filament to make the part you want, made of both solid and soft plastic. Printing with this material is easy, just print it as PLA.

I wont call that a solution in any case but at least it is interesting to know you make filament of blended PLA and TPU.

That’s odd - it’s bunching up so you want to push on it harder? Ah, that is the final output motor?

Thanks for posting the picture of the filament jam. I’ve never had one with TPU, but at least I can see where it happened. It seems like the spring-mechanism on the back of the printer perhaps wasn’t taking the TPU as fast as it could have? That or maybe a filament sensor didn’t trigger?

@ marko.pietikainen Yeah, that’s a good point, on the MMU feeding slower could help, and it’s reasonable here, given what he’s seeing.

I did some short print, I think a squeegee, with probably 15 changes and it worked fine, but even just for material handling I’d like to see switching work.

The lack of TPU support is really annoying. I want to print a hybrid PETG/TPU part (mixed layers) and without the AMS I don’t see how it’s possible.

I can’t practically change the filament manually throughout the entire print. Are there any other approaches I might not have considered?