Over the summer Richrap came up with a nice way to do multicoloured printing by joining filament. At the time I figured that automating it would be pretty cool so I did a few quick welding experiments but between one thing and another I haven’t looked at it since. Anyway I decided to take some photos/video and pop it up here in case anyone finds it useful. The two main approaches I tried were with lasers and resistors.
Filament Fuser – resistance welding:
The first thing to do is to heat the PLA in order to weld it. For this I soldered a bunch of SMD resistors in parallel onto some stripboard (18 x 120 Ohms giving 6.6 Ohms and almost 22W @ 12 V). I covered the lot with kapton tape and then taped some thin aluminium strip on to conduct the heat out of the resistors. This was then in contact with an aluminium tube which had a PTFE liner (same as Adrian’s recent RepRap hotend designs) within which the PLA sections to be joined meet. This is the heater and I connected it to my RAMPS setup and set the temperature to 160C for welding as per Richrap’s blog.
The second part of the problem is to cool down the weld so as to move the filament thru …. and continue printing if this was in some sort of automatic setup. For this I attached a heatsink to another Al strip (both strips shaped a little concave to give greater contact with Al tube). Since I want the heatsink in contact most of the time and the heater only during heating I attached both to a relay which is wired in parallel with the heater. This means that the heatsink is not in contact during heating. (Note I also tried nichrome wire wrapped around the Al tube but it took much longer to cool because the wire was always in contact with the Al and so you are waiting for filament, Al and the wire to cool… also the contact with the heatsink was poor due to the wire.)
It works pretty well and is repeatable with decent quality but at the moment takes over a minute to get to 160 C and then back down to 60C (glass transition for PLA…. maybe I could move things at a higher temp?). Also, I have to push the filaments together a little by hand – a feeder system could do that easily and more reliably! Photos below. Note: the ‘housing’ is a bit of a messy hack job and leaves lots of room for improvement!
SMD heater, heatsink and relay
SMD heater on left with heatsink on right. You can see the aluminium strips with the concave shapes (ignore aluminium shims behind stripboard!!!)
Al tube on PTFE liner. Theres a small section of larger PTFE under the Al to stop it slipping down and separate it from more Al which is inserted into the wood. Another idea is to have a longer Al piece but drill holes in it to prevent heat conducting away from the melt zone, both work.
The crude spring loading is for returning the solenoid/heatsink to the cooling position.
Theres room for improvement but results aren’t bad. The join is a little ‘fatter’ than the rest of the filament because the PTFE tube has a 2 mm id and the filament is normally 1.75 mm od. So as I push the filaments together to apply pressure to aid the weld…. the join expands to fill the PTFE tube id. Since the liner is PTFE it usually slips out easily. Usually the 1.75 mm hotends have 2 mm id going down to the hotend (to allow variation on filament od I guess) so it should go through the extruder ok… but I’ve not tested that yet. One thing to note is that the PTFE can expand but if its snugly fitting in the Al tube that probably shouldn’t happen.
Filament Fuser – laser joining:
I figured this would be a little quicker as you are directly heating the filament so theres no extra thermal mass to heat/cool like SMDs and Aluminium. I’m using 3x 200 mW laser diodes at 808 nm running at ~ 100 mW outputs to (over)ensure I avoid pushing them beyond 200 mW and frying them. I used this circuit from instructables.com (recommended by the folks at laserpointerforums.com) but with a high value resistor instead of the zener.
I borrowed the PTFE liner idea from the resistance version above as this keeps the filaments aligned! Results have varied greatly and this is mostly due to the fact that not having laser goggles for 808 nm I put everything into a box and can’t really get in there to push the filaments together, though more power would likely help too. So its a bit random and as such I haven’t been able to get reliable weld time info. I’ve tried spring loading the filaments but not very neatly. Maybe I’ll have another look at this in the coming weeks….