Re: SSTC, xfmr gate drive oddity

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

Original poster: "Malcolm Watts by way of Terry Fritz <twftesla-at-qwest-dot-net>" <m.j.watts-at-massey.ac.nz>

Hi Jan,

On 6 Jun 2002, at 7:09, Tesla list wrote:

> Original poster: "Jan Wagner by way of Terry Fritz <twftesla-at-qwest-dot-net>"
<jwagner-at-cc.hut.fi>
> 
> 
> > Original poster: "Malcolm Watts by way of Terry Fritz
> <twftesla-at-qwest-dot-net>" <m.j.watts-at-massey.ac.nz>
> > If you are really set on doing this with FETs, you will need to 
> > ensure that they cannot turn on together. You can approach this by 
> > using a 2-zener series string between the gates and driving the 
> > zeners at their junction.
> 
> Like this?
               Vdd
>               |     Vdd
>               R     |
>               | ||--+ source
>     +--+--|>|---||     p-channel
>     |  |   Z1   ||--+ drain
>     |  ^            |
>     |  - Z2         |
>     |  |            |
> ----+  |            |
>     |  |            |
>     |  - Z3         |
>     |  V            |
>     |  |   Z4   ||--+ drain
>     +--+--|>|---||     n channel
>               | ||--+ source
>               R     |
>               |    GND
               GND


almost - see amendments to above. Z4 is the wrong way round. Also, I 
forgot to mention that a couple of resistors would be needed. Z2 and 
Z3 are unnecessary (Z3 is also the wrong way round). The idea is that 
by the time one or other zener conduction voltage is hit, the other 
will have been turned off. The input swing should of course be rail-
rail or near as. The thing is basically a current amplifier. 
Personally though, I'd go with bipolars. Much simpler. More below:

> (hope it doesn't get garbled up... use a fixed font to view!)
> 
> Hmm the zeners there are probably all the wrong way round...? Or are they?
> Also, what v_br voltage should the zeners have? around 5V?
> 
> > Bipolars will not exhibit this problem in 
> > the original arrangement with their bases tied together.
> 
> I tried looking for fast, high current, low voltage, low V_ce_sat & low
> V_be bipolars. 
> 
> There are many "selection guide" lists around, but all the more commonly
> available bipolar transistors for this task seem to have microsecond range
> rise&fall times, i.e. much too slow to drive fets? Also, horrible V_ce_sat
> voltages...
> T.ex. the NPN BD909 seems otherwise nice and common but V_ce_sat~=3V and
> for BD910 (NPN) V_ce_sat~=6V.
> 
> Maybe someone who has already built a high peak current gate drive
> with bipolars could help me out and suggest some non-exotic
> bipolars that would be suitable?
> 
> I.e. ~24A peak >40V NPN+PNP transistors that could switch at <=300kHz
> "cleanly" with <100ns rise and fall times?
> 
> Though I'm starting to doubt this is possible at all... ?

I admit that's rather a lot of current. It is peak however so the DC 
collector/emitter ratings don't really count. I have successfully 
driven a 40A MOSFET from the ICL7667 which is supposed to be only 
capable of sourcing 1.5A. That of course is a MOSFET-based driver. 
But I have also driven MOSFETS directly from a TL494 with additional 
transistors to act as active pull-downs. And most SMPS I've examined 
also use bipolars as gate drive elements, usually via a transformer 
to provide good coupling in conjunction with high-side isolation. 
Those transistors are not at all big. Don't forget also that some 
series gate resistance is desirable to quench the tendency for 
MOSFETs to oscillate all by themselves and these also limit the gate 
charge currents.

Regards,
malcolm