I have had a couple conversations in the last month about the future of 3D printers.
One potential application that has come up is the ability to create prototype (or small run ) circuit boards. The possibilities are lots of layers, layers of varying thickness, cooling tubes and potential when combined with a pick and place robot for embedded components, processors.
Is anyone providing this type of printer yet within the price range of small to medium sized companies?
Hi, Walter. I have a 3D printer here and some conductive plastic for it, as well (about 100 ohms per cm, if I recall.) I usually print at 0.22mm layer thicknesses, but could go .15mm or less perhaps. Or more. The nozzle is 0.4mm for extrusion. But while I can easily do conductive layers like this, you are talking 10k to 100k ohm ranges for anything reasonable. So it's not really useful for anything but high impedances. I've been playing a little with it for touch sensors and the like. There probably is more conductive plastic available. But I just don't have any right now and honestly have no idea if much more conductive is even possible in the system I have here.
Laying down conductive wire itself would also be possible and that could be buried into the plastic. I'd probably want a
2nd head just for the wire itself and I'd need a way to snip it. Could just be bare wires and those can be buried in hot plastic, easily.
Placing parts? I honestly haven't seen a pick and place robot running before, but it must be very precise so that's good.
Hmm. I wonder about using a thin layer of plastic on a copper board to use as "resist." Hmm. I think I'm going to try that out. I think it could work really well. That will be my next test, I think.
I am just at this point seeing what is possible. Probably more to the point getting an idea where this could go with equipment that a small company could afford. For the sake of an argument what could be done with a $100K capital cost.
Pick and place robots that are quite precise are available for small automated manufacturing machines. It has been a while but 20 years ago they were probably good to 0.1 mm or so. Board manufacturing accuracy is essential a 2D problem.
I'm not aware of any such product, but I've never looked. I work with inkjet, and here are some thoughts.
A pure deposition PCB (starting with nothing) is in the realm of possibility. The conductors would probably start as a slurry of low melting point metal, and then need to be welded in place with a hot laser. The insulators would probably be a UV curing polymer.
I don't know what the most conductive polymer that can be jetted is, but I imagine it's not going to work as a conductor.
A different approach might be dicing the foil on a clad board with a laser, then printing the insulating layers.
My guess is the commercial viability of such a machine is low, so this has a big DIY aspect to it. A small CNC laser etcher/cutter is in the $25K range. The better solution would use the same XYZ motion for all operations - cutting, printing, placing.
Obtaining the consumables might be the biggest challenge of all.
You can laser cut the substrate with a cheap laser cutter (~$10-20K) but it won't cut the foil (just reflects the laser energy). SRBP may be better than fibreglass (cutting glass sputters glass onto the lenses and causes them to go cloudy, which is awkward to clean off). Cutting a board with foil backing tends to end up in a charred mess. Laser cutters tend to have a specialised XYZ mechanism because the laser itself is heavy and lives at the bottom of the machine, and only mirrors are moved around.
You can also do CNC milling of a PCB, but this has constraints on feature size and the milling bits wear out fast - you try to minimise the amount of milling you need (ie most of the PCB remains copper-clad). It does have the benefit of also drilling the holes for you by just changing the bit. It also needs careful adjustment (particularly height adjustment) to get good results.
What looks promising at the moment is laser-etching the photoresist surface of a clad board and then wet-etching it (eg with etchant in a sponge, not a chemical tank).
None of these address the issues about aligning multilayers, which I suppose is one of the motivations for 3D printing. But I suspect that can be solved by computer vision in the CNC platform - pick and place machines already use this.
So there are ways, but I'm not aware of any polished solutions in this area. In particular, when you buy one of these machines be prepared for it to be supplied with terrible software and to have to reverse engineer it and write your own.
I forgot a promising approach. Print the conducting traces with an ink that can be plated onto electroless. Adhesion might be a problem, and it's a lot a work and not a one-and-done solution, but it has the benefit of possibly working.