[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Triggering Stacked IGBTs



Original poster: "Steve Ward" <steve.ward@xxxxxxxxx>

Steve,

What is the nature of the load?  I was recently thinking of such an
IGBT array, but the load was a parallel tuned LC (like a VTTC primary
circuit, replace the tube with IGBT array).  In this case, the dv/dt
upon turn off is relatively small, because you basically have to
charge up the cap across the primary.  This of course allows much more
slack in the IGBT drive timing.

It may be possible to simply use a SISG array along with pulse
transformers (which are easy to isolate to many kV).  This would
normally use the pulse transformer to turn on the switch, but if
something went wrong, the sidacs should kick the IGBTs on.

Anyway, the design constraints are gonna depend a lot on what your
load is going to be.  Unfortunately, i dont yet have experience in the
"seriesed switch arrays" yet.  Havent really needed them.  Though i
did build a marx-like inverter that produced around 1600V from a
400VDC input at 150khz (not sure if this is of any use to you).

Steve Ward

On 2/16/07, Tesla list <tesla@xxxxxxxxxx> wrote:
Original poster: "S&JY" <youngs@xxxxxxxxx>


I am looking for ideas on practical and not overly expensive ways to
trigger a stack of IGBTs.  For example 10 1,200 volt IGBTs in series
to take the place of a spark gap.

The SISG (SIDACs) scheme is good, but I want more control.  I would
like to be able to externally vary pulse rep rate, pulse width and
the DC voltage to the stack independently.  I suppose I am after a
poor-man's variant of a klystron modulator such as used in particle
accelerators.

One method is to use fiber optic coupling to photo detectors
controlling each IGBT.  I have read these are susceptible to
electrical noise.  Other methods use pulse transformers, e.g.
modulated several mHz inputs and rectify the outputs to control the
gate.  Or use pairs of pulse transformers to trigger a flip flop on
and off to control the IGBT gate.  Of course, whatever scheme is used
must trigger all the gates simultaneously.  And the trigger circuits
on the IGBT side must derive their power from the main DC power to
the IGBT stack (to avoid the need for isolated power supplies).

Can some of you steer me to a source of ideas?  Better yet, can any
of you report your practical experience triggering series connected IGBTs?

Thanks,

--Steve Y.