I’m thinking of building a new printer and would love to be able to make 2 color prints or dissolvable supports. Does anyone have any experience with this. I see 3 paths.
-
Two hot ends Pros: Most robust No retraction or filament waste Cons: Alignment issues. Reduced travel Dragging nozzle
-
Dual filament hot end(taichi style) Pros: Easy to mount No offsets Cons: Jamming? Long retraction
-
Y splitter Pros: easy to mount No offset Compatible with any hot end Cons jamming Super long retraction
Multi extruder printing is very hard and no one has done it well for hobbyist hardware so far.
The issue is technical, but not too hard to understand. One of the main reason hobby machines are cheap is because these are Open Loop linear actuation systems. This is opposed to Closed Loop systems found in more industrial machines. Open Loop calibrates close to a known location using some form of end stop. Computationally, this location is set as “0, 0” Home and all motions are relative to this Home. There is no additional feedback in the motion system. In a closed loop system there is an additional set of sensors that confirms if each linear motion system produced the desired motion.
The difficulty with dual extruders is that the Home location isn’t actually repeatably accurate. It doesn’t matter if Home is slightly different every time because every movement is relative to this location. You never see the slightly different location because they can not be compared.
If you haven’t encountered this issue before, you are likely thinking it is easy to solve, but let me help crush some assumptions. Virtual endstops have a large range of triggering and are nowhere near absolute position or repeatable. Physical switches are slightly better but not even remotely close to absolute position reliable. Their leverage and metal spring contacts never actuate exactly the same and every pairing of contacts will produce slightly different electrical properties. In addition, every stepper motor is slightly different and microstep positions are not entirely equal in the application of force. These uncertainties all add up to a system where it is
nearlyimpossible to know exactly where the actuator is located. This does not matter when you only have the one print head. The second you need to align two heads, this is a gigantic problem. The only real effective solution is to use a closed loop linear motion system and add a shared gate that both actuators on the same axis can home to. This involves a sensor like an optical encoder on each print head that passes a stationary interrupter that triggers the encoder. This will get at least one axis aligned. You still need to account for the nozzle’s offset relative to each other and this is another far more complicated than it first seems X-Y-Z measurement issue.These are hard problems to solve and no one has solved them well. The evidence is easy to measure. Many of us have wanted IDEX (independent dual extrusion) machines. Until I see people posting functional print projects with integrated multi material and dissolvable supports regularly, this is a pipe dream and deep dead end rabbit hole. I went down this one once already. Don’t get fooled by anyone posting about some IDEX thing where they are showing off IDEX. The thing to watch for is complex problem solving where IDEX is not even worth mentioning because it is a mundane part of the design process. You don’t see IDEX prints. Don’t be fooled into thinking you are alone in this interest. Its absence is most telling about its effective application in practice.
Have you looked into Trinamic’s fancy stepper drivers that have feedback for step counting and skipped steps and other things? They might be suitable because industrial optical encoders are like $800 and I’ve never seen one small enough to go on a nema 17 stepper.
Also, what tolerances are you trying to hold here? My ender 3 repeats within a thousandth of an inch. I haven’t upgraded the limit switches, but I got a new main board with trinamic drivers.
Given the physical properties of FDM, I don’t think chasing tolerances lower than that will actually improve the quality of prints at all, and surely one thou is close enough for what you’re trying to do.
As I mentioned, virtual end stops are nowhere near accurate. This is the trinamic system. It is using over current detection as an end stop.
This is a precision verses accuracy type of problem. If you haven’t worked through these kinds of issues and ideas before it can be a struggle. Accuracy and precision are very very different things. All of our printers are precision; none of them are accurate.
Optical encoders are simple and cheap devices. You are likely limited to the shiternet’s echo chambers when it comes to searching for information. There are a half dozen optical encoders built into most cheap inkjet printers. The system needed to locate each head on a printer is cheap, the tricky part is the software and designing the second head to be adjustable to match Z offset with changes like nozzles.
The main point is not tolerances; it is alignment. Alignment requires accuracy. Printers are cheap and simple because they are only precision machines. This is a fundamental design thing. If you want to build an accurate design, this needs to be engineered into every aspect of the machine from the beginning. This affects every aspect from hardware to software.
None of that applies to option 2 or 3 though.
Dual filament hotends haven’t managed to solve the oozing issue. The multi material units can’t handle filaments with vastly different temperature ranges or material composition due to clogging. Also the temperature control is poor when changed by a large amount because of how PID works.
These systems are really only useful for multi color printing. I really wish this was not the case. I would love to design stuff with dissolvable supports, and integrated TPU for seals and tool grips.
The real key tech for functional prints is IDEX. Sorry I assumed this was more obvious that it really is to others. I tend to hone in on things like this and ignore the rest. I built a half finished IDEX prototype at one point, but the software was beyond my skill level at the time.