SSSG / SSTC - Chicago Power Electronics Tech. Conference

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "David Sharpe by way of Terry Fritz <twftesla-at-qwest-dot-net>" <sccr4us-at-erols-dot-com>

All

Just got back from Chicago PET.  Wednesday, Bert Hickman and I
traveled the exhibition hall looking at status of power electronics.
Two vendors (IXYS/Westcode, ABB) are promoting solid state
spark gap replacements, and third (ENERPRO) are considering
getting into this business.  Typical specifications on a bidirectional

blocking pulse power switch are:

Forward blocking voltage 15kV (up to 50kV is feasible)
Ipk                                   1.5kA (up to 150kA [single shot]
feasible)
Rep rate                            <400 per second

ABB has a catalog on pulse power devices, assemblies and applications.

Most use series pulse transformers for local isolated control power
generation and fiberoptic links for switch control ("firing").  Most
repetitively
operated assemblies are based on "hockey puck" geometry IGBT's,
bidirectional blocking units are two IGBT assemblies in inverse
parallel
or IGBT's with parallel diode stacks (enlarged version of Terry's
OLTC)..
Single shot, >50kA assemblies are usually based on thryistor or IGCT
(Integrated Gate Commutated Thryistor) technology.

Estimated cost for a switch is in range of $1500-2000 (US) for each
Megawatt of processed power.  The 15kV/1.5kA switch assembly
mentioned above was ~$15,000 (US) -  minus control units.
So the technology is still expensive, but only 5 years ago was
unimagined...

Illinois Capacitor has some IGBT snubber capacitors which were
polypropylene extended foil, very high di/dt - dv/dt rated and with
TAB connections for direct bolt down to IGBT's terminal (minimum
inductance).  These units could be used to make "Super-MMC"
cap assemblies that would be robust, bulletproof, and repairable.

I discussed with a colleague who now works for EUPEC out of
Boston about using IGBT's for SSTC, since this question has
risen from time to time.  The real issue is the tailing currents, and
how fast you can commutate the device off.  As an example, if
you used in a half-bridge, a 1200V/60A phase leg motor control
device is generally rated for less then 25kHz..

However, if you derate the current by approximate increase in
operation frequency, you may be able to successfully run at 100kHz,
issue is heat dissapation due to switching loss.  Conduction loss is
trival (about 2W per amp of current), switching loss at this
frequency will predominate. So your tank L/C  impedance would
have to be relatively high to limit conduction currents, or run
multiple devices in parallel, synchronously triggered using small
equalization inductors (toroids with 2-3 turns) to force current
sharing not unlike caps used by Terry in OLTC to each IGBT.
A hard 50% duty cycle is definitely a NO-NO, with tailing
currents you have to guarantee the switched off device is
blocking before commanding the opposing leg device to
turn on.  A half bridge topology is preferable to a full bridge;
less parts and simplier gate controls.

Another issue is designing a means in the circuit to absorb
reactive swings (i.e. kickback) when resonator is discharging
to air then contacts a grounded object.  An IGBT would
be much more robust in this application in my opinion versus
a Power MOSFET, since there is no blocking diode + high
speed schottky diode to deal with (and disable) the internal
blocking diode.  Less circuit parts in power path will generally
end up being more reliable in long run.

All in all was an excellent conference, came back with a list of
ideas useful to work (and hobby... :^) ).

Regards
Dave Sharpe, TCBOR
Chesterfield, VA. USA