Reg,
I saw your post in the EDM group as well as here. There is a good reason you have gotten no responses here, and that the responses you got in the EDM group were discouraging. The reason is that EDM is a poor alternative to other available means for creating boards. That does not mean you can't do it; I'm sure with enough effort, you can make it work.
And I understand the attraction -- when I first began contemplating building an EDM unit, I thought using it for PCB construction would be a great idea. However, after 1) actually building a working EDM unit, and seeing what it does and doesn't do well, and 2) learning how to make PCB's using toner transfer and etch with HCl-H2O2, I realized I did not want to mess with making PCB's using EDM.
Here is what I suspect you will find if you decide to go ahead:
1) The production of a board, especially with the geometry that you describe in the EDM forum, will be excruciatingly slow. Keep in mind that in EDM, fast = high amps = rough, smooth = low amps = slow. In other words, to get smooth traces, you are going to have to remove very, very, very tiny bits of metal with each spark. Don't forget that the "placement" of a spark cannot be controlled exactly, so you have to go slow and take tiny bites to get an acceptable result. Almost certainly you will need to trace each path at least twice, once to rough it out and once to smooth it.
2) As was pointed out in the EDM forum, the software needed to make this work will be extremely non-trivial. You must ensure that you maintain electrical continuity back to the ground at all times. (Note that the electrode itself may be positive or negative with respect to the ground, depending on your design, choice of electrodes, etc.)
3) Production of electrodes will be extremely non-trivial. Keep in mind that in EDM, the electrodes are gradually consumed. I say gradually -- exactly how fast depends on the material, polarity, use of pulse, etc. Keep in mind also that the electrode will burn a path larger than itself. To achieve 10 mil spacing between traces, therefore, you will need an electrode that is perhaps 6 mil (.006") in diameter. Producing such an electrode in graphite (the best choice) will be extremely challenging outside of a specialized production environment. (No, pencil leads will not do -- they are not pure graphite.) You could instead use copper wire of the appropriate size; expect it to wear very quickly, and to be extremely challenging to keep it sufficiently straight (rigid). You may think you could get around the problems by using a pointed electrode -- the point will be small enough, and the rest will provide rigidity. But after just a short time, the point will have eroded away, and now suddenly your 10 mil spacing has become 15 mil, and the 10 mil trace in between has disappeared.
Compare all of the above with two readily available alternatives:
1) Milling a PCB using a cone-shaped bit -- I have never done this, but those who do seem to feel that it works great. Google here or elsewhere for details. Cost will be considerably less than you would be investing in the EDM route, speed will be greater, and results will be more consistent. And you can have it drill any through holes as well.
2) Chemical etching of a board produced using either toner transfer (TT) or UV-sensitive coatings. Both methods are described at length on this forum. Using TT, I routinely produce boards with 10/10 geometry, which, IIRC, will handle the board you have described on the EDM forum. I have produced tests of 8/8 and 6/6 and even 5/5; based on those tests, I would expect that I could produce an 8/8 board with reasonable confidence. Though my 6/6 and 5/5 tests were successful, they were pretty marginal. I have not used UV transfer, but those who do seem to get even better results than TT. Note that the costs for either of these methods is astonishingly low. Toner transfer costs a few cents for a laser printout; UV artwork costs maybe a few cents more if you use inkjet on transparency. (If you need to buy the laser/inkjet printer, there are some that work well that are inexpensive, and you need a printer anyway, right?) Bare copper clad is extremely inexpensive on eBay; coated costs more, but not unreasonable. Transferring TT is best done with a laminator, but can be done successfully with an iron. (Some people have reported good results even with a $30 laminator!) UV transfer requires a suitable UV lightbox, which can be made from parts available at Home Depot. Etching can be done with a number of chemicals; I favor hydrogen peroxide - hydrochloric acid, which costs pennies per use.
So, bottom line: By the time you get an EDM production working for even one prototype, you could have made 1000 successful boards with one of these other methods -- and they would probably be better quality than the EDM one -- and you would have spent a whole lot less money!
Your choice, of course. Sometimes discouraging words are like waving a red flag in front of a bull, at least for me. But there are so many other things I want to spend time on, with much better chances of success ...
Final word: the place that EDM shines is in machining hard metals, where ordinary tooling would be ineffective. Two classic examples: burning out a die for press production, and burning out a broken tap. (Wire EDM has some additional -- but also specialized -- uses, but we're talking sinker for PCB production.) Copper isn't a hard metal; it can be removed quickly and easily, either chemically or mechanically. You can use EDM to remove it, but you can also use a slegehammer to kill a gnat. Either of the latter will work, but the results may be messy!
--- In Homebrew_PCBs@yahoogroups.com, Reginald Beardsley <pulaskite@...> wrote:
Show quoted textHide quoted text
>
> Hi,
>
> google turned up a discussion here and in EDMHomebuilders about EDM fabrication of PCBs 6-7 years ago, but surprisingly little else.
>
> I'm particularly looking for information about pulse voltage & current profiles and repetition rates required to allow me to design a power unit.
>
> Can anyone point me to such data?
>
> The concept is have a single CNC XYZ table w/ interchangeable heads for plastic extrusion, drilling, etching, solder paste application, pick & place, etc.
>
> Have Fun!
> Reg
>