Re: Sheppard-Talyor circuit up on my website/ low Z construction technique

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

Original poster: "rob by way of Terry Fritz <twftesla-at-qwest-dot-net>" <rob-at-pythonemproject-dot-com>

Tesla list wrote:
> 
> Original poster: "Justin Hays by way of Terry Fritz <twftesla-at-qwest-dot-net>"
<pyrotrons2000-at-yahoo-dot-com>
> 
> Hi Rob and All,
> 
> > I am finding that the FETs dissipate little wattage, but the FREDs
> > get extremely hot.
> 
> D26 and D25 are definately in the current path, they will see the
> same amount current the MOSFET's are. I suppose D24 is getting pretty
> hot? It will take the full current flowing in tranformer TX4's
> primary.
> 
> Ummm...IRG4PH50U is a 24A 1200V IGBT, good up to no more than 40kHz
> (http://www.irf-dot-com/product-info/igbt/navigbt.pdf). No way this will
> work at 240kHz but maybe you're only using those part #'s on the
> schematic for simulation.

Hi Justin,  actually I'm not using IGBT's at all.  I must have goofed
and put up the wrong schematic. I will fix it. I'm using IRFPG50's. 
They definitely switch at 240khz, as I am now getting corona off of very
low input voltage.  I have a big 10A variac and isolation transformer
coming next week so I can slowly ramp up the voltage to some comfortable
level.

> 
> So as I look closer, it seems that both transistors are simply in
> parallel, being used in a common buck converter topology...the output
> of which goes straight into the primary of the step-up transformer.
> This differs from the normal buck topology - where the output is
> rectified and filtered.
> 
> But is this really the way the circuit works? Again, I can't see
> waveforms and all, and would like to have confirmation that that is
> how the circuit generally works (or should work). Otherwise I might
> be misinforming...

Its a type of buck-boost topology.  The diodes charge up C at t=0, and
then when the FETs turn on, they reverse bias the diodes, and dump -Vc
into the load.
 
> 
> So if this really is the way things work, then both gate signals
> would be in phase, and the transistors would turn on and off
> simultaneously. This leaves the option of paralleling the transistors
> directly (drains go to 160uH L opposite 170VDC power,,,sources go to
> GND)? This would get rid of (expensive, hot) diodes in the main
> current path?!
> 
> A freewheeling diode COULD NOT be used after this modification,
> because it would short out the negative voltage (coming from L16)
> that is supposed to go into TX4's primary. But then again it
> shouldn't be needed...?
> 
> In the current schematic, if both gate signals were 180 degrees out
> of phase, the transistors would give a DC short through the 160uH
> inductor.....which would only let up at very small intervals where
> there is dead time. That's why I figure they're in phase and both
> devices turn on simultaneously.
> 
> The gate voltage to these paralleled devices (or a single device)
> could be provided by only one secondary winding on the gate isolation
> transformer...that would be really nice too. No worrying about one
> transistor turning on before the other.
> 
> As mentioned in a different post (Dan's I think), you will need a low
> impedance driver to nail the high-capacitance gates to 10 or 12V.
> With these large gate capacitances, rise/falltimes will be a
> significant total of the switching waveform. Best to keep things as
> fast as possible without ringing L's and C's.
> 
> But anyway Rob, you're making me want to try this!
> 
> Also real quick here, if L16's core saturates it will draw gigaamps.
> I don't know if you have yet, but I'd put little current-sense
> resistors EVERYWHERE and have a scope on as many of them as I could
> at one point in time. (watch grounding of course...)

I'm using a very large core.  I can always gap it if need be.  But sims
show the biggest current spikes flowing thru the diodes when the MOSFETs
are turning off.  I found that by shaping the gate drive I can modulate
this current pulse.  Its a kind of cross-conduction, somewhat like what
you see in half-bridge supplies, but I don't think its as severe, since
the diodes are super fast.

Rob.

> 
> Take care,
> 
> Justin Hays
> KC5PNP
> Email: justin-at-hvguy-dot-com
> Website: www.hvguy-dot-com

-- 
-----------------------------
The Numeric Python EM Project

www.pythonemproject-dot-com