Let me start by saying that I agree that desktop electronics manufacturing will be the "next big thing" in the maker movement, and I am a huge fan.
However, this article does not mention two very important issues. In order to make a circuit board of any interesting complexity (i.e. more than an Arduino and a few LEDs), you need more than 2 copper layers to route your traces on. And, you need VIAs (Vertical Interconnect Access), which are columns of copper that make electrical connections between the layers.
I have used a PCB mill and while it is very useful for making extremely simple circuits, it only lets you mill 2-layer boards, and it does not let you make VIAs (instead you have to manually drill a hole through the board and solder a pin into the hole, and this approach has a much much larger diameter than a real via).
In order to be interesting for "maker" goals such as Linux machines, IoT, and robotics, there needs to be a desktop fab solution that can at the very least make 4-layer boards, with VIAs.
"via" comes from latin, it means " a place to travel trough".
For example in Spain "autovia" means via for autos or "tranvia" means a vehicle that goes over rails. "Via pecuaria" means a road for animal transport.
You can do lots of useful things with just two layers. We prototype professional electronics in house with just two layers and modules(over a big area), then shrink the design, use more layers and get it done using a professional service.
I have worked in the electronics industry for over a decade and this is the first time I've ever heard the phrase "vertical interconnect access", so I suspect it's a backronym.
Looks like some others at Wikipedia are similarly surprised:
Commercial services for making PC boards are so competitive [1] that making them yourself is rarely worth the trouble. The standard photochemical process is quite good, better than any of the do-it-yourself processes. (In Shentzen, you can get 4-hour turnaround at no extra charge.) Machines for milling one-off boards have been around for decades, but they've never been used much.
Small-scale solder paste printing and pick and place is more useful. Those are hard to do by hand, expensive to get done for one-offs, and solder paste/place/reflow all have to be done in close sequence. Reflow ovens are easy; toaster ovens with special temperature controllers are often used.
Shouldn't the mill be able to precisely drill very small holes for you? Perhaps you could you use the pick and place machine to put small pins in the holes which would be soldered in place on both sides. You could conceivably use one 2-layer board and two 1-layer boards, each with alignment holes drilled in them, pins inserted and solder paste applied to make a 4-layer board. There would be a thin gap between the boards, but since everything is soldered together, would it matter?
This describes the one of the first places I worked at. It started as small contract manufacturing house before becoming a significant player in this space. However, one thing is clear, the amount of manual work involve does reduce the consistency of getting good results, ie the amount of solder paste squeeged onto PCB is not controlled, the placement of the components are also not well controlled etc.
However in the maker space, this option is great! Fantastic for getting a small production run out the door!
For decently capable(1) desktop manufacturing,"desktop" is a bit of a misnomer at this point. I've found that you need at least three desks, not to mention another desk for all of the test gear, and another for mechanical prototyping, and another for...
At this point, I have(2) an othermill and a pretty-near top-of-the-line LPKF S63 system + silkscreen and reflow capabilities, and am fairly impressed by both. I've been able to go from opening up a PCB editor to having a tested, functional board in under an hour, at least for simple breakouts.
The othermill is pretty flawed for PCB work, however. The spindle just isn't fast enough, cross layer alignment is inadequate, it doesn't have dust control for dealing with FR4 dust, etc. For good PCB routing, some sort of computer vision system for adjusting cut depths and picking up fiducials is also amazingly handy. That being said, we're really impressed with our othermill and do a lot of mechanical work on it - complex aluminum parts, engineering plastics, etc.
Cheap soldering for one-offs is pretty much solved at this point - laser cut a stencil on a decent laser cutter (which everyone should have, anyways), stencil paste, and reflow in a cheap Chinese oven with open source aftermarket firmware. You won't be able to do big BGAs or tiny (0201) passives, but who cares? If you have soldermask capabilities, either some photoresist method or laser cut kapton, soldering tricky components becomes much easier.
None of the additive (ink-based) systems have any hope of producing really useful boards - the electrical and thermal conductivity of real copper is pretty hard to beat, and doing anything controlled-impedance is pretty much out of the picture. With subtractive systems, you can just switch to 8+ oz copper or exotic substrates without changing the process materially...
Vias are tricky. LPKF makes two systems, one with conductive epoxy and the other with electrochemical copper, neither of which are quite satisfactory. Via rivets (or manually soldering wire) work for small boards, but are super tedious and not great for controlled impedance applications, either.
Being able to do decent multilayer boards is contingent on good vias, but in theory, the only other thing you need is a hot press for stacking things up?
Once you get into building complex multilayer boards, the time investment is just silly, even with perfectly calibrated processes, and you should just order from a board house and wait...
(1) I.e. boards better than DIY etching + manual soldering.
However, this article does not mention two very important issues. In order to make a circuit board of any interesting complexity (i.e. more than an Arduino and a few LEDs), you need more than 2 copper layers to route your traces on. And, you need VIAs (Vertical Interconnect Access), which are columns of copper that make electrical connections between the layers.
I have used a PCB mill and while it is very useful for making extremely simple circuits, it only lets you mill 2-layer boards, and it does not let you make VIAs (instead you have to manually drill a hole through the board and solder a pin into the hole, and this approach has a much much larger diameter than a real via).
In order to be interesting for "maker" goals such as Linux machines, IoT, and robotics, there needs to be a desktop fab solution that can at the very least make 4-layer boards, with VIAs.