A low cost temperature test chamber

[Martin Czech]

A low cost temperature test chamber.
(or another boring 15 minutes with m.c. bla bla)

Why do I need this?

First of all to examine thermal behaviour of my circuits.  All this
tempco stuff, bias currents, offset drift, leakage , etc.

Then you might want to avoid MARGINAL STABLE DESIGNS.  Even if you work
very carefull, this may happen.  A cicuit that works only sometimes is
worse then a circuit that doesn't work at all. It will make you crazy.
To check for this you can change the circuit parameters and/or supply
and/or temperature voltage within reasonable limits.  If the circuit
works under all conditions, you can be pretty shure that it is stable.


Design constraints:
No hazardous voltages should be in reach of my fingers, so I dropped the 
idea of 230V heaters.


You need:

-a 12V (axial) blower
-a PVC sealed fuse box with clear top
-some wire wound resistors
-power trafo and rectifier
-plastic foam sheets for isolation

You know, back in 1983 I had no idea about electronics (I was 17), and
followed the ttl circuits I found. Thus I thought it would be good to get
this cheap 8V 10A Transformer. Not very usefull today, I don't use ttl
any more and power consumption has dropped very much. So no real need
for such a 5V supply.

Heater:

This idle transformer comes handy now for heating my soldered combination
of severall wire wound resistors (about 1 Ohm) out of the junk pile. Looks
like a strange apparatus, some ceramic rods with 5mm gap. This draws 8A,
ie. about 64W heater power.  The resistor array will simply stand in
front of the blower, and all this is put into a corner of the PVC fuse box,
on a little piece of carton to avoid burning the PVC under all conditions.

Box:

The PVC fuse box is about 25x18x10 cm if I remember well.  Enough room
for an 100x160mm board, the blower and the heater. The clear top is nice, you
can see what's going on, and manipulate.  Two 5mm holes near the bottom
provide input for wires.  

Experiences:

I used a cheap thermometer to "calibrate" a 1N4148 and my 15 EURO
multimeter in diode voltage mode. Scale factor was about 3mV/K.  So I
could measure aproximate temperature. OK, turn on heater and blower (never
stop the blower, and never let the device allone, fire hazard!!). After
14 minutes I got 70C.  That's the upper limit what I would assume as Ta
(ambient temp).  By manipulating the wires of my diode sensor I could
measure the temperature at different locations without opening the box,
it is the best to keep the blower always on, but also in this case the
temperature at the top was 5-7 C higher then at the bottom! Next thing
was to insert a EURO test board, mounted on 4 cm metal pins (I always
give "legs" to my test boards, because I sometimes use both sides for
components). All components are on the same level about ground, and I
could measure that the temperature did not differ more then 0.5C on the
board area! I'll have to verify this with a better thermo sensor.

Regulation:

I have no regulatur at the moment, so I was the regulator. I looked at
the diode voltage and compared it to some value (say 630mV) und turned
the heater current on/off accordingly. This is a two-point-servo loop
(sp?) and could be done with a relay and a comparator. So I was able to
keep the temperature within +-0.3C. Better then I expected!

Improvements:

1. Heating time: I can get to about 60C in 20 seconds, with the aid of
a hair dryer used with open box top.

2. Heating power: I have isolated all sides (not the top) with sheets of
plastic foam, I did not glue them (don't do this!), but used a piece of
string to fix them, like you would do a parcel. This way I can remove
the isolation, e.g. if I put the chamber into the fridge in order to
get lower then room temperature (in the summer I won't get much lower
then 27C in my living room!) In this case you want BAD isolation.
During heating I also put a foam sheet on the top, but will remove it
later, for a good view and for downward regulation loss, and for some
convection exchange. With 5 side isolation I could achieve more then 80C,
but did not try to go further.

3. Constant current: I will make a servo loop for constant dc heater
current.  This servo will be controlled by another servo loop, a simple
PID servo for controlling the voltage. This pseudolinear scheme will
suppress any ac magnetic fields (impossible to shield) and the
"eternal switching" problem of a two-point-servo-loop, because this would
lead to a continuous drift and also transient switching currents. For
some cicuits even the PWM curent of my soldering iron is too much!


Conclusion:
This temperature chamber cost me about 100 EURO so far (all included).
The improvents may cost additional 50 EURO. I would have never expected
such good results for so little money.


m.c.