Re: "Flyback" Driver - Why are my MOSFETS hot? (fwd)

[Steven Roys <sroys@xxxxxxxx>]

---------- Forwarded message ----------
Date: Mon, 5 Jul 2004 11:12:32 +1200
From: Malcolm Watts <m.j.watts@xxxxxxxxxxxx>
To: High Voltage list <hvlist@xxxxxxxxxx>
Subject: Re: "Flyback" Driver - Why are my MOSFETS hot? (fwd)

On 2 Jul 2004, at 22:30, High Voltage list wrote:

> Original poster: Steven Roys <sroys@xxxxxxxx>
>
>
>
> ---------- Forwarded message ----------
> Date: Sat, 03 Jul 2004 10:17:47 +0930
> From: Matthew Smith <matt@xxxxxxxxxx>
> To: High Voltage list <hvlist@xxxxxxxxxx>
> Subject: Re: "Flyback" Driver - Why are my MOSFETS hot? (fwd)
>
> Malcolm Watts replies to Matthew's Message - Matthew's reply
> interspersed:
>
> > Is the core gapped? The average current doesn't give a good idea of
> > the peak currents. What is the duty cycle of the driving waveform?
> > Do you have a circuit you can post? Better still, an oscillogram of
> > the waveforms across the MOSFETs?
> I sketched a schematic, but found that my scanner is currently hors de
> combat - too much furniture to move to get to the cables!
>
> Basically, 555 drives totem pole via resistor, both powered by common
> 5V supply.
>   Totem pole drives MOSFET gates via parallel resistor and diode
>   (diode points
> back towards totem pole).  MOSFETS have their own 24V supply sharing
> common ground with the 5V stuff.
>
> <http://www.mss.cx/tmp/scopeshot1.jpg> shows the circuit running with
> the 24V supply turned off.  Waveform is taken between totem pole and
> MOSFET gate resistor.  Lower trace (straight line) is taken from a
> resistor divider (n/2) connected between ground and MOSFET common
> drain line.

Shouldn't that be "source"?

> <http://www.mss.cx/tmp/scopeshot2.jpg> shows same hook-up, but with
> 24V turned on.
>
> Images are reduced to about 30kb for painless viewing ;-)
>
> > Some quick and dirty calcs:
> >
> > ASSUMPTION: duty cycle is 50%
> Yep - I've tweaked it so that it is :-)
> >
> > Ipk = V*Ton/Lp = 24*25e-6/2e-4 = 3A   which sounds OK ASSUMING the
> > core is gapped.
> It is - there are mica (?) shims 0.3mm thick.
>
> > How many turns on the primary?
> 26 - as near as I can see.
>
>  > What is the effective pole area (Ae) of the core?
> 150mm sq, based on the legs on which the windings sit and deducting
> the cutaway that accommodates the tie bolts.
>
> > From that information I can calculate the core flux and
> > determine whether B (flux density) is being kept to a reasonable
> > level. You should be able to do this all with a single MOSFET and I
> > speak from experience.

>From the data provided, The core appears to be experiencing 150mT
which is fine for any 3Cx material so the answer lies elsewhere. The
first thing I would look at is significantly increasing the applied
gate voltage and I'd recommend 12V minimum. Also, just use a single
MOSFET so that the gate capacitance and hence switching time is
reduced to a minimum. I looked at the oscillograms with interest
(BTW, the old HP scope was one of my favourites for many years). The
drive waveform suggests that your MOSFETs are being turned on
sporadically when they should be off.  Do you have a Vds waveform to
post? Also, the source current waveform appears to be ramping in the
wrong direction. Something appears funny there.?
     I do not recommend using source current-sharing resistors as
they introduce NFB into the circuit. For fast transitioning
waveforms, these shouldn't be necessary for multiple MOSFETs
(different to bipolars). I'd run with a single MOSFET (e.g. IRF840).

Malcolm