Dwayne Reid wrote:

> The only thing that I found non-intuitive was that TE coolers do NOT
> like PWM power - I always found better efficiency if the PWM was
> smoothed to DC first.

  This relates back to the thread on LED brightness.  The difference is
that the series chain of Peltier modules achieves a higher voltage drop
and runs at significant power making it really worthwhile to use SMPS
techniques.

  The equivalent curcuit for a Peltier device is a "back EMF"
proportional in significant degree to the temperature differential, in
series with an internal resistance.  Power dissipated in the EMF
equivalent does useful work (cooling) whilst power disssipated in the
internal resistance is deleterious to performance (it actually heats the
"cold" side!).

  The significant point is that the "back EMF" is very much a constant.
PWM without an inductor and commutation diode, preferably a capacitor
also, simply biases the internal resistance much harder for a fraction
of the time.  Unfortunately, the power dissipated by this resistance
varies as the square of the current, whilst power achieving useful work
in the "back EMF" (situation is identical in a motor, BTW) varies
directly with the current.

  IOW, delivering twice the current for half the time generates twice
as much waste heat.  In a motor, this is a minor nuisance but can
perhaps be designed around.  In a Peltier cooler however, it is
absolutely antithetical to the primary goal.

  The other aspect of PWM, indeed as I am given to understand, frequent
and repeated cycling of the power, is that the thermal response of these
devices is instant and their thermal mass quite small.  You will thus
produce a thermal alternation and through expansion, a mechanical
oscillation which is stressing the structure of the device.

  Not a Good Thing, so I am told.  It is recommended to drive them with
a power supply which ramps up the drive current relatively slowly and
keeps it constant.  Raw DC from a battery charger is not advisable.  A
well-designed SMPS, more to the point.
--
  Cheers,
        Paul B.