I wrote:
Actually the inexpensive 3D printing processes use
ABS plastic,
which is quite good for production use. The extruded ABS doesn't
end up with exactly the same properties as molded ABS, but it's
close enough for many applications.
Mark Benson wrote:
The way it is placed by a 3D printer in spots or fine
layers would make it less durable, IMHO.
[as compared to molding]
Actual, it turns out the fused deposition process results in material
with strength, durability, and impact resistance that aren't that much
different than molding.
Feel free to prove me wrong, I've never had
first-hand experience
with these systems.
I think Stratsys has some white papers on the subject.
About PCBs:
Milling isn't the issue. What is the issue is
every one of those PCBs
has
to be laid out, spec'd and tested as a one-off.
That's quite a costly and
time consuming exercise.
The layout process can be easily automated. The arrangement of the key
matrix can be arbitrary since it can be remapped by a table in the
microcontroller firmware, so automated layout isn't nearly as hard as
the general case. One approach would be to generate XML files for input
to Eagle 6, then let it autoroute. In general autorouters aren't very
good, but for a simple problem like this they work fine.
The software that sets up the PCB layout can also generate the lookup
table for the firmware.
Testing has to be done, but that can be automated also, using a simple
cartesian robot driven by software using tables generated from the
layout. Could be the same robot that does the pick-and-place for the
keyswitches.
Relatively simple software could automate the entire process, such that
it would only need a minimally trained person to put the raw feedstock
into the machines, transfer subassemblies from one machine to another,
and assemble the finished electronics into the case. It's not a trivial
project to generate the required software, but it's not rocket science
either. No part of it requires any research or any development of new
algorithms.
Eric