On Sat, 20 Mar 2010 19:06:24 -0500, John Fields
wrote:

On Sat, 20 Mar 2010 13:28:20 -0700, John Larkin
wrote:

On Sat, 20 Mar 2010 05:58:22 -0500, John Fields
wrote:

On Fri, 19 Mar 2010 18:17:58 -0700, John Larkin
wrote:


How about this...

ftp://jjlarkin.lmi.net/Amp.jpg

---
Not bad...

JF

We exploded a coffee mug full of fets before we found some that
survived.

---
Hmm...

From:



You wrote:

"I *calculate* in advance if things will work, and
almost exactly how they will work."

Funny, I would have thought that "almost exactly" would have been a lot
fewer than a coffee mug full...

JF

We did a bunch of safe-operating-area characterization of the best
fets (dynamic Tj measurements and tests to destruction) to develop a
thermal model of them. The amp has a microprocessor that measures
everything (namely all relevant voltages and current and heatsink
temperature) in real time, 2000 times a second, and dynamic-model
simulates actual junction temperatures. We shut down if simulated Tj
gets too high. This allows us to push the fets a lot more
intelligently than any simple current limit could do... especially
since we usually run into inductive loads.

So I let the uP do the safety calculations for me, in real time.

These are (were) all 300-watt rated fets. Note the variation in
silicon area.

ftp://jjlarkin.lmi.net/ExFets.jpg

John