I'm no expert on thermocouples, but doubt it's a very effective way to
run a fan. IIRC (NOT guaranteed, it's been 20 years since I seriously
used thermocouples!), you need MANY pairs of two thermocouples, one
"Cold" and one "Hot", wired in series, to generate any differential
voltage, and the voltages & currents, per-junction, are sort of low
(Maybe just use a heat pipe instead!?) - and the thermocouples would
have to be thermally connected to the CPU, getting in the way of the
fan's airflow, especially if you had "many" thermocouples, like 150.
(Could blow the incoming air past the "cool" thermocouple ends to cool
them further, I guess, but then I don't know if you'd get enough cooling
this way, at all.)
Think of a "star" configuration, with two leads coming in from the
side, then a continuous sequence of connection to wire type "a", runs to
a welded junction at the "Hot" central spot, welded there to wire type
"b", then out to the "Cold" outer ring, welded there to another wire of
type "a", then back to the hot spot, etc. etc. etc. so usually you see a
sort of "sun" zig-zagged around a bunsen burner in the center, with a
ring of junctions air-cooled at the outside. You could, admittedly, do
this in a rectangular matrix form, any & all heat siphoned away from the
CPU by the wires will decrease the efficiency of your thermocouples
though (Hmm, maybe use a water evaporative cooler for the "cold" plate,
and give it LOTS of thermal mass compared to the CPU plate? And you'd
want non-conductive electrically, but very conductive thermally,
connections to the "cold" plate.)
(I believe historically that thermocouples have been used to change
extreme heat differentials into power, but we're talking say 2000
thermocouple pairs to get 5 volts, at a small amperage, from a CPU, I'd
-GUESS-. Far less than 200 mA, I believe. I can be quite wrong,
though, I am not current on the field...)
Let's see: Altavista search gives:
http://www2.omega.com/techref/tctables/temper11.html, browsing those
tables I only see voltages, for temperature measurement purposes, in the
2-5 mV range (no currents given.) {as an example}
Other than that, I looked but didn't see anything with "thermocouple"
and one of { "efficient", "power generation", etc. }
I didn't see anything last I looked at
http://www.almaden.ibm.com/journal/sj/mit/sectione/starner.html, as a
distributed body-wide net of thermocouples might be practical (?over the
neck area?) but admittedly that's a different application...
Question: Would a Peltier device be more efficient than a fan, to cool
a CPU? These may be more efficient than 20 years ago's models, I am
curious but haven't looked yet... Wonder how those conduct heat while
no power is applied?!
Mark Willis
Steven Work wrote:
>
> How much power can be got from a thermocouple, heated by the cooking
> CPU to help cool it off?
>
> Steven Work
> Renaissance Labs
> 
>
> Mark Willis wrote:
> >
> > Hmm. On second thought: Couldn't you charge a large electrolytic with
> > the power from a solar array, then use that power to cool the CPU fan,
> > similar to the circuits at:
> > http://www2.xtdl.com/~bushbo/beam/oscillate.html,
> > http://www2.xtdl.com/~bushbo/beam/beam.gif {polarity errors on D1!},
> > http://www.geocities.com/ResearchTriangle/9879/solaroll.htm
> >
From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty
Archive created with babymail