Re: Comparison with resonant switching topology (was Re: My FET T.C.)

On Wed, 17 Nov 1999 11:45:14 -0700 Tesla List <tesla-at-pupman-dot-com> writes:
>Original Poster: "Marco Denicolai" <Marco.Denicolai-at-tellabs.fi> 
>To:   tesla-at-pupman-dot-com
>cc:    (bcc: Marco Denicolai/MARTIS)
>Subject:  "My FET T.C."
>>Original-original poster: Ken Herrick 
>>>3.  Driver-- 24, 85A, 500V power MOSFETs electrically in push-pull
>>>parallel, together driving the equivalent of 1 primary turn (in a
>>>patent-applied-for circuit).  Soon to be increased to 48, then--if 
>>>endurance holds out--to 72.  That will double, then treble the 
>>>length (and the input current for a given spark rate).  The primary 
>Just a little comparison with the switching topology I am using 
>(resonant load,
>I use 4 1200V 35A IGBT (oversized in terms of current). I switch 560 
>VDC into a
>15 (fifteen!) turns primary transformer, using a primary peak current 
>of only
>12A. At the secondary I get 5kV at 1A. All the switcher fits on a 
>100x160 mm
>(eurosized) board. The 4 IGBT use together the same 4 'C/W heatsink 
>and stay
>pretty cool all the time.
>>>Soon to be increased to 48, then--if my
>>>endurance holds out--to 72.
>You can convert easily to the resonant topology I am using by just 
>adding a
>series capacitor to the transformer primary and setting suitably 
>frequency and duty cycle. Then you'll save tens of MOSFETs and will 
>heat less
>your room!
>>Also FETs can be killed by over voltage. I tried this and managed
>>to fry quite a few FET's.  The problem is that when the FET's switch 
>>the magnetic field collapses a very high voltage is induced on the 
>>- it could easily exceed 500V.
>No problem with that. Never burned a single IGBT for overvoltage (or
>overcurrent). In my topology you switch on/off always at 0 current (so 
>there is
>no kickback from the transformer, or primary, and switching losses 
>Check my design at http://www.saunalahti.fi/dncmrc/ccps.htm
>From KCH to Marco Denicolai (& all):

Thanks for your comments!  I checked out your site & printed it out for
reference.  I might comment that, while the text came thru fine, the
drawings were chopped at the right (on my HP 1120C).  Perhaps if I'd
loaded HP's Print Smart utility & figured out how to use it, I'd have
avoided that.  But perhaps my cheapie ISP, Juno, may not allow for
printing the whole width; don't know.

I'll study the notion of a series-resonant drive some more (I'd briefly
considered using the scheme a long time ago) but our applications are
different:  You produce dc while I produce ac, from my MOSFETs.  One of
my goals was to utilize only off-line-rectified dc (of about 160 V in the
U.S.) and to then convert that immediately to whatever output-frequency
ac amplitude I would need to directly drive my t.c.  Also, a resonant
drive of an ac signal would add another frequency-pole into the circuit. 
As it is now, I have only one--the instantaneous self-resonant frequency
of the secondary--to contend with, and I manage that nicely.

As to all those MOSFETs I have to use, a) I want to minimize the voltage
each is exposed to and I do that by minimizing the primary turns (to 1,
resulting in the ultimate transformer!) and the supply voltage (to 160);
b) I'm willing to handle the resultant current, for a given power into
the t.c., by paralleling MOSFETs; and c) once I parallel enough to handle
the expected peak current, then I boost the voltage into the t.c. primary
by incorporating multiples of the paralleled groups.

To get the secondary cranked up to full voltage during 30-50 cycles of
each pulse-burst,  I need a lot of current into that 1 turn: about 700 A
with all MOSFETs in the circuit.  That comes out of capacitors storing
the 160 V from the power line.  I need plenty of Idsm for that since each
group of 12 MOSFETs handles it all, on half-cycles.

You mention voltage spikes:  I had briefly tried another topology in the
primary--still utilizing 1 turn--and quickly dropped it when I noted the
truly astronomical voltage spike between half-cycles.  Quite unbelievably
large--and narrow.  I don't get that with my current design.

Ken Herrick
Get the Internet just the way you want it.
Free software, free e-mail, and free Internet access for a month!
Try Juno Web: http://dl.www.juno-dot-com/dynoget/tagj.