Ed Huntress wrote:
On Thu, 15 Mar 2012 13:21:37 -0700,
wrote:

Ed Huntress wrote:

(...)

I guess your clamping pattern assured that every cell
had an equal share of hot and cold sink coupling.

It was an even pattern, and any difference would be apparent at the
plates' edges, but it also counted on the grease. In any case, it
worked. We moved one set of cells all around on the plates and the
results were pretty consistent.

You didn't have variation in solder thickness, as our
friend did in his PWM MOSFETs, so that removed a
significant variable. Very cool, and warm.

(...)

Sounds like Science, too. One could bolt on thermocouples for instance.

I attached 1N914 diodes for my own initial tests. They're very linear
for temperature when back-biased within a certain voltage range --
MUCH better than thermistors. But the leads have to be short. Amplify
the results with a 714 op amp and drive a small meter. +/- 1 deg. F
over quite a long range.

That was in 1981, remember. d8-)

'Sounds like you had to 'level shift' and scale but you
didn't have to 'linearize'. -2 mV per degree?
Or is that the 'forward' and not 'reverse' tempco?

That's the story. They loved it. And they decided not to invest in
OTEC.g

Ouch! That's like deciding not to buy a car because all
you could test were Yugos.

g Their primary tests were with steam turbines using ammonia for a
working fluid.

(Gasp!) Though I am sure it was perfectly safe.

I love the TE concept but despair over their inefficiency.

I don't remember the values. I have a couple of them that I scrounged
from an old camping cooler. It killed the battery in my van one night
and I took it apart for revenge.

That fridge *needed* disassemblin'.

I suspect that Stirling motors would work the best in
this application, no? ('Sounds like a great purpose
for 'flare gas' at last.)

They'd likely be the most efficient energy coverter at those
temperatures. But you're dealing with very small temp. differentials.
It would take a *lot* of Stirlings to generate useful power.

In OTEC, Stirling 'ballast' would be a Good Thing, (at last).

--Winston