Re: TESTED works, Flyback driver plans

["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 Justin,
          In view of your post, I've quickly reviewed some MOSFET 
data as well as my own memory. Here is what I've found:

On 16 Oct 2001, at 16:43, Tesla list wrote:

> There is about 150V p-p on the primary of the flyback. The IRF540 MOSFET is
> rated 100V. I knew something was up with that...
> 
> You are referring to the MOSFET "body" diode? I knew this was a diode that
> was inherent with the MOSFET silicon design, I did NOT know that it was a
> zener!

It is a zener, at least on some types of MOSFETs. There is at least 
one type of MOSFET which according to Motorola has no substrate diode 
at all (GEMFET). I have observed distinct drain-source clamping at 
the rated Vds on some logic level MOSFETs in the past. Some data I 
reviewed showed the diodes as zeners, some as ordinary diodes. The 
diodes are generally enhanced for energy recovery purposes. In view 
of your drain-source voltages hitting 150V, maybe it isn't a zener 
for your devices. Or maybe they are being pushed. Zener voltages are 
current-dependent, at least to a degree.

> So you think that is what's saving the MOSFET? You're correct in that it
> gets hotter than that hot place. I'm using about a 2" x 5" heatsink, fins
> spaced every 1/4" or so that are about 1/2" tall. I've got really good
> thermal conduction between the device and sink (no mica, lots of compound,
> tightly bolted in).

And luck probably, especially if Vds is rated for 100V.
 
> It will burn your fingers after about 30 seconds of continous power. Inside
> the enclosure, I've made room for a small DC fan, I'm looking around to
> yank on out of old equipment somewhere because they're so expensive.
> 
> So if I used a properly rated MOSFET (IRF740 or 840), it will run cool? My
> guess, I'm no EE. 
> 
> I'll put an IRF840 in it when I get home from class, to see what it does.
> It seems as if I've done this before, and performance decreased as well as
> MOSFET heating.

Not surprising. Rds(on) for the IRF540 is 0.07 Ohms or thereabouts 
and 0.8 Ohms or so for the IRF840. You can work out the voltage drop 
across the device based on peak transformer primary current 
Ipk = V.Ton/Lp and hence peak dissipation. Pushing the rated DC drain 
current of 8A through an IRF840 leaves a voltage across it in excess 
of 6V (48W) vs 0.6V or so (4.8W) for an IRF540 at the same current.

      I have to take all this stuff into consideration each time I 
implement a design with power devices. I chose bipolars for my two 
latest incarnations. There are pros and cons for every device in a 
particular application. For example, while Vce(sat) for a high 
current bipolar looks extremely attractive on the face of it, there 
are turn-off time issues, having to source a significant base current 
due to low device current gain at high collector currents etc. A 
partial solution for my case is to run the main switch driven by a 
lower power device as a Darlington pair, but then the voltage drop 
isn't quite so low anymore. Additionally, the lower power device has 
to have the same Vceo/Vcer rating as the main switching device. The 
upside of a Darlington configuration is sourcing the main switch base 
current through the output load and non-saturation of the main switch 
which promotes faster turn-off, and the downside is increased 
dissipation.

     In general, high working voltages and high current ratings in 
MOSFETs quickly push the price up. Engineering is truly the art of 
compromise.


Regards,
malcolm