midnight solder / temperature chamber

Martin Czech martin.czech at intermetall.de
Wed Nov 10 10:28:39 CET 1999

Remember my temp chamber?  I picked up the thread again (cycling is
over for this year).  Mechanical construction is a bitch, but I managed.
I tested it with hand-regulation this weekend, I can get to 70C, better
would be foam isolation (saves 5min or so from 25C to 70C).

The whole thing started on friday, I went to the opticians to get new
cycling goggles (glasses), and looked at the thermometers. All 20 EURO.
No, wait we have one on sale for 7 EURO. Bingo.  It is one of those
inside/outside things (two sensors), with min/max memorry and 1.5V
operation. Arrived at home, the inner sensor was quickly turned into 
a second external sensor. Then I made some ice and tested with 0C and 42C
(digitla fever thermometer). Wow, 0.1C deviation! A bargain.  I grinded
the encapsulation of the external sensor away, this reduces mass and
gives much better thermal contact, up to temp in seconds.

Now the chamber: I bought another blower and some cooling metal pieces,
where the 4x 25W (safety) resistors and the 2N3055 workhorses sit, as
voltage followers. Actually they are combined darlingtons with BD139 now.
The voltage follower has the advantage of low gain, thus stable.

The two blowers make a very nice air stream, I couldn't detect any temp
variations inside the box, but I have to investigate that further.

The high air velocity together with the cooling metal keeps the temp
difference between the 3055 and far away air down to 10C, which is very
good for safety. I didn't expect this. After all, this devices consumes
8-10A, it could be even more , but the transformer is too weak. This *is*
current, and one has to be carefull not to burn the semiconductors. The
resistors are on the safe side (these golden, mountable wire wound from
Burns have 25W lowest rating anyway).

The next step was the servo loop, yesterday. I use one KT84 as temp sensor,
it is good enough for this purpose. The voltage difference across the resistor
is small, I get only 100mV signal from 25C to 60C eg. So large amplification
is needed, OP07 is no bad choice here, they're cheap.

Here is the result (sloppy):

Voltage divider, 27K upper leg to 15V, KT84 lower leg (about 800 Ohm)

Temp C   sensor voltage V
60.8     0.4030
56.4     0.3905
54.0     0.3834
50.1     0.3730
47.6     0.3679
43.5     0.3567
39.9     0.3475
37.7     0.3424
31.3     0.3264
27.3     0.3176
24.3     0.3104

Then a difference amplifier to compute the servo deviation, and in the
first step, a compararator to switch the 3055. As expected, this two
point servo arrives at some selected temperature, but will never settle
within 1C. always kep on moving, which is no good.

Reducing the "comparator" gain gives a linear servo. At least I
thought so.  In fact the amplifier is heavily overdriven most of the
time for reasonable regulation! This and the extremely lossy integrator
characteristic of the temp chamber make the loop very friendly and
stable. I couldn't obtain oscillations for large step responses, only
for small steps, when in linear mode. The natural frequency is then
about 1/30s.

All in all I'm very satisfied with the chamber dynamic, but the servo
should be expanded from P to PID. Of course, the integral part has to be 
clamped to 0 as long as the opamps are overdriven, otherwise I expect
extreme oscillation problems.

Is this all absolutely necessary for synth diy ? No.
But I will learn a lot more about temperature, and it is fun.

Schematics, fotos etc. will turn up if finished.


More information about the Synth-diy mailing list