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

Re: SSTC, xfmr gate drive oddity



Original poster: "Jan Wagner by way of Terry Fritz <twftesla-at-qwest-dot-net>" <jwagner-at-cc.hut.fi>

Hi Justin!

> > The mosfets should be clamping point Y to 13.6V and 0V in turn, 
> > right?
> 
> Yes. The upper MOSFET pulls it high, the lower MOSFET pulls it low.
> Or it's supposed to.
<snip>
> But I still don't like the circuit, it has never worked well for me.
<snip> 
> I think that scoping the drains at the same time would reveal
> everything.

It didn't reveal much. :( With the local decoupling, the voltages directly
measured between mosfet drains is 13.6V solid with only a few 10mV of
ripple during switching.


> To get around this problem, I switched to gate drive IC's such as
> MAX4420 or MAX4429 (TC4420/9's cheaper). 6A drive capability, 2500pF
> load, FAST rise/fall times. MAX628 dual inv/non-inverting, 3A drive.

Yeah, good idea! 
I'm starting to consider those too for my small SSTC too...  :)

But, it's just that the load of my intended "lots'a kW" SSTC is eight
IRFP460's, i.e. the transferred charge over one cycle, including negative
bias, will be Q_per_cycle = 2 * 8*150nC = 2.4uC. At 100kHz, average
current is 0.24A, i.e. pretty much. With inteded <100ns switching times 
I_peak >= 2.4uC / 100nS = 24A => any "low amperage" ;) driver IC would fry
_very_ fast... 

Well maybe using a bunch of maybe 4-6 driver ICs could help out here.
Gotta check the max free samples order count at Maxim... ;o) That might
resolve the cost issue.
Although, two SMD >30A peak mosfets are quite cheap and handle much more
current than any expensive driver, so getting this to work would be
beneficial. IMHO. If it just would work...


> > But scoping at Y gives 4V and 9V levels for the square wave signal.
> 
> > Very
> > very odd. 
> 
> Sounds similar to the problems I had, although I've gotten it to work
> OK before.

Good to hear! Or sorry to hear, really. 
Well at least the reason then isn't that it was just me again having
screwed up my really-simple circuit somehow, once again. ;o)


> I think I had to use different gate resistor values.
> Perhaps one of your MOSFET's is going bad...if you haven't already,
> try swapping out both MOSFET's at the same time and have a look. 

I don't have a gate resistor! hehe...  Just relying on CD4049 hex
inverter output resistance. Yup, bad design practice, I know...  :o)

The previous cascaded two NPN/PNP complem.emitter followers driving the
mosfets turned out to give only 3V<->9V voltages to the xfmr i.e. the
situation resembled this one slightly. "Less" voltage because of bipolars'
V_be=1.0V and mosfets V_t=1V. Mosfets heated up a lot.
 
CD4049 made the output/mosfet sources rail-to-rail again, but
for some reason it's rail-to-rail for very light loads only. Bespite 
mosfets running super cool even at 330nF xfmr coupled loads. That's the
puzzling part...

Swapping out both mosfets for know-good ones didn't improve matters. :o(


Darn, maybe the mosfets are plugged in the wrong way round??

Should the p-channel one be connected to Vcc and the n-channel to GND? And
not the other way round?

Next thing I'll try out is to add a series schottky to each mosfet source.
If this has anything to do with highly capacitive load and mosfet body
diodes...?? Dunno. Let's see.



> Also, (you probably know this already, but) if the circuit is drawing
> lots of current from somewhere, the supply voltage will drop and
> result in the low transformer primary voltage swing.

Yup. Local decoupling helps very much here.


> However, this
> doesn't explain why the voltage isn't going from GND up, or Vs down.

True... *sigh*


> Because the fact that it's riding somewhere in the middle (as if on a
> DC bias), this is most likely a timing problem.

Now that you mention it... that once was the case with totem pole drive
i.e. signal+inverted signal, where switching overlap of a few 10ns caused
a lot of mosfet heating. The output looked similar, i.e. not the full
voltage swing. Switched to emitter follower/complementary emitter
follower after that, so no delay/timing probs. At least I'd suppose there
aren't, because it is just one signal, not two.

But the odd thing is the mosfets run super cool in the current circuit.

Well I'll get going again and see if more gate resistance, swapping mosfet
places, and more, will help... 


many thanks!

cheers,
 - Jan

--
*************************************************
 high voltage at http://www.hut.fi/~jwagner/tesla