"Don Foreman" wrote in message
...
On Sat, 25 Oct 2008 12:49:02 -0400, "Ed Huntress"
wrote:

But when Agie came up with sophisticated,
heavily researched power supplies during the '70s, the Japanese,
particularly, developed some work-arounds that kept them from violating
Agie's patents (more or less -- but that's another story g)

What was Agie's approach? Those patents are expired now.

Sodick's approach was common, although
they did it particularly well for the time.

What was Sodick's approach?

I'm hoping you just want a general description, and not design info, because
I've long since forgotten it and some of the details were secrets that I
never knew. But here's the general pictu

Let me first describe the EDM spark cycle as consisting of three stages. The
first is ionization and polarization of molecules of liquid dielectric. As
the molecules become polarized they start to flow; this flow, called
"streamers," starts heating the liquid channel according to the IR drop. The
second stage is low current flow via the streamers, within this "channel,"
that vaporizes the dielectric and then ionizes the resulting gas, forming
the low-resistance channel that allows high current flow in the third stage.
The third stage is an extremely high rate of current flow through the
ionized gas, originating in a low-impedance circuit of the power supply.
This is the spark that does the eroding of the workpiece.

Agie's breakthrough power supply design (it had a name -- Robo-something or
other), which came along in the mid-'70s, used an oscillator to produce a
string of pulses at the beginning of a spark cycle. These were fairly high
voltage, high impedance pulses, but the voltage was much lower (IIRC) than
that of other EDMs of the time. The pulses apparently were very quick and
effective at starting the ionization of the liquid and they could be
reliably controlled. The control is the key to producing a high duty cycle
and, thus, high cutting rates at relatively low rates of current flow in the
later stages of the cycle.

Sodick and others used an advanced version of earlier power supplies,
without the oscillation. (That part was an Agie patent.) They had a separate
circuit for each of the three stages: high voltage, high impedance; medium
voltage, medium impedance; and low voltage, low impedance. They used
transistors to switch each of the stages *off* at the appropriate time in
the cycle. Timing circuits for each stage were adjustable, although not all
of it was user-adjustable. Switching these circuits *on* was not actually
under direct electronic control; the physics of the process determined when
they turned on.

I think you can see in general how these circuits apply to the three stages
of the cycle. With this system you had pretty good control, although the
physics of the process limited what you could do electronically. Further
advances in power supply design, since the '70s, could be thought of as
efforts to "force" the physical dynamics to conform to desired timing,
voltage, and current flow, and to do it with the highest possible
predictability and reliability.

If you're interested in improving the RC relaxer circuits used in home-built
EDMs, there is an earlier stage of commercial development that holds more
promise and *could* be within the realm of a hobbyist who's knowledgeable
about electronics. Before these power supplies I've described came along,
the standard approach was an RC circuit with an electronic switch -- first
tubes, then switching transistors -- to turn off the high current flow. This
made EDM much faster, allowed finer finishes, and improved the process a
great deal over the earlier, plain RC designs.

To understand how and why they worked is another story. I'll just point out
the physics involved. The elemental RC power supply is a current source, a
series charging resistor, a capacitor, and a second series resistor, which
is the EDM channel itself, with the electrode and workpiece forming the
terminals, and the channel being the resistor material. The key to the
process is that the resistance of this "resistor" drops with changes in
voltage, due to the ionization of the channel. It starts out behaving like a
low-value capacitor, in other words, and then it becomes a resistor as
current starts to flow.

The storage capacitor is charged through the charging resistor until the
voltage on the capacitor is high enough to start the ionization process at
the electrode-workpiece interface. Then the cap discharges through the
channel, eroding the workpiece in the process. The whole process depends on
getting all the values -- charging resistance, capacitor size, and
electrode-workpiece gap -- in the proper balance for the process to proceed.

The limitation of this simple circuit results from the resistance of the
cutting channel immediately dropping to a very low value. If the value of
the charging resistor is too low (and you want it to be as low as practical,
because its value determines cutting speed -- lower value means faster
cutting), and if the channel resistance becomes too low from high current
flows, the spark doesn't stop; it becomes a continuous arc, and you've just
turned your EDM into a welding machine that wrecks the workpiece and the
electrode. This is the dreaded "arc" that was the bane of EDMs before
fancier transistor-controlled power supplies came along. A simple hobbyist
machine must be run at very low cutting rates to prevent this from
happening. That means that the charging resistor must have a higher value
than you might prefer, and it also places limits on the size of the storage
capacitor, because a big one will allow the EDM channel's resistance to drop
through the floor.

I hope you're following all this. I haven't tried to explain it for years.
g If you're with me, you can see how a switching transistor (or vacuum
tube) in the circuit can be used to great benefit if it switches off the
current flow under direct control, using something as simple as a 555 timing
circuit that's triggered by the high current pulse at the start of the cut.
It allows low charging-resistor values, larger storage capacitors, and high
current flow rates, and it keeps the sparks from turning into arcs.

In theory. In practice, it works pretty well, but not perfectly. The famous
old Elox EDMs (vacuum tubes) and the iconic Charmilles D20 (transistors) of
the 1960s were practical and effective machines that made EDM the practical
process that it is today.

Two more things: Notice that I'm not distinguishing between wirecut and
ram-type EDM power supplies. The differences are in the values of the
circuits more than in the concepts. Wire EDMs can be a little more brutal on
the power supply end because electrode wear is not a problem. You toss the
wire after it makes one pass.

The other thing to note is that EDM has developed from experience, not from
theory. The companies involved threw things at the wall until something
stuck, and then tried to figure out later why it worked. It's led to some
screwy theories that seem to contradict each other over time. The Swiss were
particularly good at buggering the story up. g The point being, don't get
too hung on the theories. And face the fact that EDM has been developed by
people who had well-equipped labs and lots of financial incentive, not by
hobbyists. It really doesn't lend itself to easy, low-cost experimentation.

Oh, I should mention that I haven't looked at the hobby designs for at least
five or more years. so someone may have hobby-level designs incorporating
transistors. If so, I wouldn't try to re-invent them, if my object was to
build a homemade EDM that worked and actually cut things.

Good luck, if that's what you have in mind.

--
Ed Huntress