I now own the Prusa3D MMU2. The benefits, costs, and experience others have had is well documented. I am interested in rebuilding my large, home-designed delta machine to be multi-material, and don't want to overlook strategies I haven't considered.

My original implementation used an E3D Kraken as the hot-end, and handled the inevitable delta tilt by adding two additional degrees of freedom to the head to lower the selected nozzle to the bed. I've been through three generations of mechanisms, and I think the third will work.

But, I feel that I am not seeing obvious and better alternatives.

So, the question: Through what methods and mechanisms can a multi-material (different polymers, different temperatures) FDM printer operate, and are there available designs or examples of best practices for those methods?


3 Answers 3


Let's look at various methods:

Multiple Hotends

The oldest version and one of the best to print materials at vastly different print temperatures (like printing a cheaper PLA infill into a Polycarbonate shell - the print temperature difference is 60-100 °C) is to have 2 or more hotends. This way also avoids the need for purging towers. It does, however, limit the maximum size of the used printbed and few 2-printhead machines are cheap.


Using a bowden setup, a Y-coupler could be used to feed the filament from 2 extruders into one hotend. On the switching tool command, E0 would pull the filament back some couple millimeters beyond the coupler and then E1 would push forward back into the meltzone. One will need a purging tower/object.

Special, multi-entry hotend

Some Hotends had been concieved that have 2 or more ways into the meltzone and the multiple extruders push along them. They generally are quite complex and hard to clean, but they allow to seamlessly blend between two filaments of the same material and create pretty much a controlled fade by precisely directing how much of either side is used on any layer. For clean cuts, a purging tower is necessary.

Splicing filament

This is what the Palette 2 and the Prusa MMU do: they push pieces of filament into a feeder tube that then are consumed by the printer via its own extruder. If they melt the filaments together like in the PAlette, it's proper splicing, if they just line up the next filament piece without merging into a spliced filament it's more like instant color switching.

This method is good for multi-color prints or using materials that have the same or similar1 melting temperatures. It might or might not need a purge tower/object to get rid of the residue in the zones between the filaments.

This could btw also be done manually but should be avoided.

1 - or rather not too dissimilar, if the slicer is set up to do it right. By setting up the slicer cleverly, one can have the extruder retract the filament, then adjust the heat over the purge tower and then resume extruding in the purge object at the changed temperature. PLA/PVA from a Prusa MMU is known and advertised to be doable, PLA/ABS might be possible this way. For extreme dissimilarities like PLA/PC (60-100 °C) I have my doubts though.


All of these variants are basically viable, but some have benefits over others. Service is in this comparison meant as repairing a broken extruder, maintaining as the operations needed to keep it in printing order.

  • multiple fully independent hotends is among the easiest to services. It could be direct drive (good for flexible filaments) or bowden. It is however heavy and usually not an option for delta printers. It has a downside that you have to perfectly level two hotend nozzles to be exactly on the same height, putting it in the hard to maintain category.
    • multiple hotends on the same carrier is harder to service and maintain in comparison to multiple independent hotends as the components are very close together. Especially nozzle height adjustments can be more finicky.
  • Y-Coupler needs to be a bowden and has problem with materials that are very stringy. That makes it especially bad for flexible materials. Maintaining is like a normal hotend and servicing is almost the same.
  • Special hotends are hard to come by but could be available for direct drive, making them possible for flexible filaments. They are, as already noticed, very hard to service.
  • Splicing filament can be done with either direct drive or bowden setups. It is probaby the most convenient to use after setup and has the maintenance and serviceability of a single hotend and a fully separate machine. Their biggest downside is price and setup time needed.
  • $\begingroup$ Deltas are tough to use with multiple nozzles since the plane of the central carrier is not parallel to the bed over the whole x-y range. This is due to tiny manufacturing errors, wear, and other errors which are compensated through bed leveling. $\endgroup$
    – cmm
    Jun 1, 2019 at 15:45
  • $\begingroup$ @cmm Y coupler and splicing are the best choices for Delta. $\endgroup$
    – Trish
    Jun 1, 2019 at 15:54
  • $\begingroup$ I'm a fan of Y couplers for things like deltas where you're trying to hit minimal effector mass. My own coreXY build I swapped the 8mm rods for 8mm CF tubes, and I'll be using Y couplers on a bowden setup for multi-material printing. $\endgroup$
    – Nach0z
    Jun 1, 2019 at 15:57
  • $\begingroup$ The Prusa MMU2 prints materials with somewhat different temperatures, and heats or cools the heat on filament switches. $\endgroup$
    – cmm
    Jun 1, 2019 at 23:25
  • $\begingroup$ @cmm Similar temperature doesn't mean same. I don't expect it to have very good results with extremely dissimilar materials though. Think PET (200°C) to Polycarbonate (260°C) or even higher. $\endgroup$
    – Trish
    Jun 2, 2019 at 3:12

One of the easiest ways that I've seen, which I'm a fan of, is simply putting Y splitters on your Bowden tube and having multiple feeds to a single hot end. The main benefit is that you only need a single hot end, so you don't have to worry about extruder offset or alignment or anything like that, but you do have to worry about material blending somewhat. Basically you end up needing to build a "purge tower" next to your printed items that you use to transition from one material to another.

There's the Diamond hotend setup that basically moves the connections into the hotend itself, which reduces the size of your purge tower but increases the risk of burning if you're trying to print with materials with vastly different printing temps, like PLA and PETG.

You could also have swappable hotends but that requires you to be there to manually swap the print head twice per layer. Don't recommend.

Unfortunately there's only so many solutions to the multi material problem, either you put multiple materials through a single hotend, or you have multiple hotends. I'm a fan of the single hotend approach personally, especially on deltas where weight and space are at a premium and alignment becomes problematic.

  • $\begingroup$ Have you found a good way to use flex materials with Bowden extruders? $\endgroup$
    – cmm
    Jun 1, 2019 at 14:51
  • $\begingroup$ Not without ridiculous retraction settings. That said, you can also technically have your extruder mounted to the hotend, but still fed through a short 4-6 inch Bowden tube with Y connectors. It's kind of the worst of both worlds but a bowden that short shouldn't have substantial issues with flex filament. $\endgroup$
    – Nach0z
    Jun 1, 2019 at 15:55
  • 1
    $\begingroup$ @cmm: I'm using flex fine on the Ender 3 (bowden). See my question and answer here: 3dprinting.stackexchange.com/q/8512/11157 $\endgroup$ Jun 2, 2019 at 0:23

Another way to combine the simpler geometry of a single nozzle, and to get the reduced mass of a single extrusion tool would be to make it like a CNC machine with a tool changer. One material is printed, then the hot end, extruder, and feed tube are swapped out for another which is primed and ready with the next material.

Lots of mechanical precision problems exist for arranging for the nozzles to be in the very same place, plus or minus a small tolerance. This is worsened by the presence of filament bits and strings which seem to eventually pollute the workspace.

If that could be worked out, one could have a plethora of extrusion tools, nozzle sizes, materials, multi-material mixing chambers, and other complexities.

E3D was talking about such a printer, but I haven't seen a product... only an invitation to send money as a show of interest in such a printer.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .