I was re-reading Adrian's blog article about etching PCBs with electrolysis and a movable electrode. The main problem seemed to be that a small electrode etches a hole much bigger than itself. I read that spark erosion can produce very accurate shapes [en.wikipedia.org]. How about replacing the electrolyte with oil, moving the electrode very close to the work and apply a large voltage. The electrode will wear down but we can just feed it as a wire from a slightly modified version of the extruder.

Daft idea or worth a try? I am willing to have a go if people think it has a chance of working.

interesting idea. we've basically pushed all the pcb stuff to v2.0. its definitely something worth experimenting with though!

The wiki page suggests this is done under parafin. I have also seen parafin recommended as a lubricant when machining aluminium. I have always wondered why it does not ignite in the presence of heat and sparks? I have some parafin, maybe I should try an experiment outside before I put it in HydraRaptor which is largely made of wood!

Building a solder extruder is my next big project after I have Tommelise printing acceptably.

I think the electrical conductivity of solder is about 10 times less than copper. Presumably thermal conductivity is similar. For a lot of simple low power circuits this will not matter, but may be a problem for high power, high frequency or low noise circuits.

No biggie. Just use ten times more solder in the traces. :-)

Not quite the same when it comes to high frequencies. The extra distances will add to the inductance and there is also the skin effect.

Extra distances? I was talking about increasing the cross-sections of the traces. Tell me more!

Also, how high are high frequencies?

I am thinking of hundreds of MHz and surface mount components as that is where most things are at these days.

Digital signals have harmonics way above their fundamental. In fact it is the edge rate that important, so if you run some fast logic at a low clock speed you will still get high frequency signal components. When running power to, and decoupling high speed circuits, short lead length is crutial to keep the inductance down.

By making the tracks ten times thicker the signal will have to travel deeper to make use of the extra thickness. Due to the skin effect, [en.wikipedia.org], high frequency signals only travel close to the surface.

I don't think we use anything like those kinds of frequencies on the controller boards that we are currently making. The PIC that I'm using has a 20 MHX external clock that could be hyped up to 48 MHz. Nowhere near hundress, though. The rest of the traces rarely get above the low KHz.

Mind, what you've described would rather limit the usefullness of the print your own boards thing using eutectic metals like solder.

Yes, I don't think it would be a problem with RepRap control boards, or any through hole components for that matter, but my machine for example has 100Mbit Ethernet so that will involve frequencies well above 100MHz.