Re: rochester cooling problems

[Tom Droege <tdroege@veriomail.com>]

Merle,

First, thanks for offering help.

The design runs the two cooling heads in series, just to prevent the 
problem you suggest.  This results in the downstream cooling head getting 
warmer water than the upstream one.  One solves this problem by just 
running a high enough flow rate so that the temperature rise is not 
significant.  In this case, with a cooling load estimated to be 25 W per 
head, this requires about 10 gallons per hour to keep the difference in 
cooling temperature to the two heads to a few degrees F.

The TEC sits on a cooling plate where there is a meander milled into the 
back side.  The thickness of the plate is 1/16" in the meander to limit the 
temperature drop through the plate.  The large area of the meander insures 
good thermal contact between the cooling fluid and the aluminum cooling 
plate.  There is an aluminum block on top of the TEC which contains the 
thermometer, then a grease joint to the back of the CCD.  The TEC is glued 
to the aluminum plate with silver filled epoxy, there is also a glue joint 
between the aluminum block containing the thermometer and the TEC.  In the 
early days, these glue joints sometimes failed.  This would result in poor 
thermal contact.  Now I use better glue technique.  Now they really stick 
and I have to destroy the TEC to get it off to replace it.

The design is such that forces on the TEC do not change significantly with 
temperature changes, etc.  It is free to move in the stack direction, being 
constrained only by the pressure supplied by the deformation of a thin 
printed circuit board with provides a few ounces per square inch 
compression.  There is nothing leaning against the stack to provide a shear 
force.

There are a number of systems working of identical design where the 
expected cooling is achieved.  This points to a failure of the TEC and not 
a design problem.  While there are a number of places where a failure could 
occur, experience points to a failed TEC.  These units are very 
fragile.  They can stand some compression, but fail easily in tension or 
shear.  The mount is designed to provide a proper mount.  It is mounted in 
mild compression.  I have seen the units fail from shock, and tension as 
when the thermal grease joint is pulled apart.

As I have pointed out in previous e-mails, the TECs are easy to break in 
tension.  Just a few pounds of force required to pull the CCD off the 
aluminum block will do it.  I have learned to gently work the grease joint 
back and forth to gradually break it free.  The also break from 
shock.  Just dropping a camera head a few inches on to a concreate floor 
has done it.  I measure the health of a TEC by measuring the current at a 
set voltage.  Good ones draw a high and constant current.  Bad one carry 
much less current which generally decreases over time.

I just looked at the data sheets, and the solder used to assemble the 
junctions melts at 135 C.  One might try to heal the TEC by heating it up 
to 130 C or so.  But this would probably damage the electrolytic 
capacitors.  Most of the other parts would stand it.  So not much can be 
done other than to take the unit apart and replace the TEC.  Looking at the 
data sheet, they claim the device will stand hundreds of pound of force in 
tension and shear.  Don't believe it.  No doubt such tests were made by 
some uniform force machine.  In practice you bump them and the force is far 
from equal.  So the weak overstressed device fails.  Sigh!  I note that the 
data sheet recommends mounting in compression of several hundred 
psi.  Possibly they might fail less easily if mounted in higher 
compression.  It is really too late to design a different loading scheme 
now, since the design is complete, and most units continue operating once 
installed.

Tom Droege

At 04:08 PM 1/10/02 -0800, you wrote:
>Michael,
>I have been reading the TASS e-mail for two years.  I think I can help 
>here.  I detect either a hydraulics problem or a heat transfer problem.
>
>I take it you are cooling two CDs with one pump.  Apply the conservation 
>of mass.  With one source of flow and pressure you must evenly split 
>(divide) the flow between the two coolers.  That means the same diameters, 
>lengths, and fittings on the two parallel lines.  The pump output is also 
>divided evenly.  If you have a design that requires a certain flow rate, 
>the "rule of thumb" is to choose a pump design where the design point is 
>75% of the pump rating.
>
>The heat transfer problem may be inadequate contact between the CD and the 
>cooler.
>
>I would recommend that you tackle one problem at a time, eliminating each 
>one until the real culprit is found.
>
>Merle Jephson-King