It’s been quite a while since I've done an update on the metal printing front, so I thought I'd do an update of where we are. In my last blog post I set out about choosing a low melting point metal which would have some unusual properties which would help with printability - mainly choosing a temperature which would minimise damage to our traditionally printed plastic components on to which our metal would be deposited, and also using a non-eutectic to attempt to minimise the effects of surface tension.
One of the main problems I previously had was solubility. Running molten metals were acting as solvents for my heated nozzle - resulting in the nozzle slowly dissolving during a print. At the end of my last post I'd just tried using anodising to create a strong oxide layer on the surface of an aluminium nozzle to protect it, and that the results were promising after little running - I've done hundreds of hours printing since and as far as I can tell no damage has been done and its still in its original condition. I'd anticipate that a stainless steel nozzle would also be useable as it also has a strong oxide layer.
Previously the plastic and metal were printed on separate machines. Anyway, I've heavily modified my X carriage to take one Bowden extruder (for the plastic) and one "standard" extruder for the metal such that I can (in theory) do one shot printing. The metal extruder is fairly standard, the only major modifications are the inclusion of an O ring to reduce leaking, and running the PEEK insulator right to the end of extruder to minimise the melt zone - the result is a slow extruder - I'm currently printing track at about 100mm/min - but hopefully one which we have the most control over.
I printed the above about a month or so ago. The plastic housing contains a female hole for supporting an ATMEL644P PDIP chip, as found in our Sanguino electronics. The metal tracks are housed within 0.7x0.25 rectangular channels. Surface tension would suggest that the metal would naturally want to take a circular cross section - however given the size of the track I'd be unable to get anywhere near this hence the rectangles. The component was inserted into the plastic and the metal track automatically deposited on top before being covered with more plastic. Importantly we can see that the plastic extruder isn't excessively melting the metal tracks when covered.
Since that print I've been battling a few bugs with the setup - namely reliably keeping the offset between my plastic and metal extruder - the sprung mounts flexed under the compression of the bowden cable - and getting somewhere near a reliable metal filament drive - It turns out my standard hobbed bolt I used to do the driving wasn't good enough - I think the problem was due increased wear of the bolt due to the higher stiffness/hardness of the filament and a lack of compliance in the filament reducing the contact area. Anyway creating a new stainless steel hobbed bolt seems to have improved things massively:
Here is a stab at the Arduino compatible Sanguino board (albeit simplified). It's pretty standard except we've removed the reset circuitry and alot of the pins. We still have 4 controllable pins, one for the LED and three spare for something fun in future. Once again the plastic was printed before dropping in pre-tinned components and finally printing the metal tracks. I have previously done some tests which show we need to have a radius on each corners of printed tracks, ideally at least 1.5mm, but for compactness I squared these off resulting in poorer quality but nevertheless its quite a big step forward from where we were a few years ago. Four extra tracks are required on a second layer to get the circuit fully working; I've done this manually for the time being. In addition I had to manually solder in 2/3 pins as the track had not connected properly, however I think I can correct this by extending sections of track beyond their required endpoints and utilising the bigger radii at corners that I've already mentioned. It's still a little blobby, but nevertheless here it is working running a simple blink program, although we can still reflash the chip to do something else with the spare pins: