Re: [CAD_CAM_EDM_DRO] Re: EDM home brew units ?

Ed Huntress described an older commercial system that he was very
familiar with
a couple of years ago. I understand how this works, and have a fair
idea how one
would build it. Knowing just a little about other EDMs, I think
paralleled FETs can
be used as a high current source, perhaps something like 100 A per
transistor, since
the on-time is going to be low duty-cycle. (That 100 A is the peak
current, not the
average. The average per transistor may be something like 5-10 A.)

Here's what he wrote :

Re: building an edm
From:
"Ed Huntress" <huntres23@...>
Date:
Mon, 26 May 2003 17:57:36 GMT
Newsgroups:
rec.crafts.metalworking

"steamer" <steamer@...> wrote in message
news:F6rAa.16980$JX2.1052802@......

> Spehro Pefhany <speff@...> wrote:
> : control stuff? I've done it with a paper towel, a stencil and a bench
> : supply.
> --Gotta know: what voltage, amperage did it take??

A lot of different ones.

Here's the basic concept of Sodick's power supply ca. 1980. It was designed
to get around the patents of Charmille's Robofil power supply, which used a
more complex (and slightly more effective) approach.

The simplest, standard circuit (there were about four separate ones, IIRC,
for different cutting conditions) was actually three power supplies in
parallel. The first was a high-voltage, high-impedance circuit intended to
polarize a plasma channel. It was about 300 V, but I don't remember the
impedance.

In parallel to that was a 90 V, medium-impedance circuit that initiated the
plasma and got some current flowing.

In parallel to that was a high-amperage circuit that could deliver something
like 50 average amps, at around 10 - 12 volts.

The first circuit prepared the way for the second, and the second for the
third. When the third circuit fired, the plasma channel opened up and
carried an extremely dense current, which melts the little glob of metal
that each spark produces.

The sequence was repeated at about 4 kHz. At the end of each spark, the
circuit has to shut the current flow down, HARD, in order to prevent a
self-perpetuating arc. Once an arc starts, current will flow even from the
high-impedance circuits, and the workpiece will develop a weld spot and be
wrecked.

All of this has to occur with a minimum of inductive reaction. If that
sounds easy, remember that you have peak amperages of around 500 A flowing
during the discharge, if you're running at a 10% duty cycle (roughly average
in those days). Current density within the plasma channel is immense and
uncontrolled reactions ring like a bell.

Controlling the initiation of each spark is a mean-voltage sensing circuit
and a current-sensing circuit that actuates the servo and that overrides the
shut-off circuitry to prevent contact-welding. If you get contact, the
electrode often will weld to the work and blow the power supply to hell.

This is an extremely simple system by today's standards, although it was
very effective on the ordinary run of work, with ordinary projected-area
EDMing. When you have pointy electrode projections or when you're trying to
do contouring and have minute projected areas, this circuit slows down to a
crawl. Still, it's many times faster than an RC circuit under the same
conditions.

--
Ed Huntress
(remove "3" from email address for email reply)

________________________________
I hope Ed doesn't mind me quoting him, and that this makes as much sense to
others as it did to me. Of course, I'm pretty good with power
electronic circuits,
so to me this is almost as good as a circuit diagram.

Jon