# temp stabilization circuit

Discussion in 'General Electronics' started by Marc, Dec 15, 2003.

1. ### MarcGuest

Hi all,

I am designing a temperature servomechanism. The layout is as
follows: I am exciting a wheatstone bridge configuration (where one
of the resistors is replaced with my thermistor) using an AC sine wave
voltage source. The bridge difference is then amplified and then sent
to my demodulator chip. The ouput of that is low-pass filtered (which
gives me a DC signal proportional to the bridge voltage difference),
sent to the PID control, and then goes to the TEC driver which of
course provides the current for the TEC.

My problem is this: I need to monitor the signal somewhere in the
configuration so I can know what the achievable temperature stability
that my circuit provides. I thought about recording the amplified
bridge difference but this is then nonlinearly related to the change
in temperature. No problem, just take the derivative of a few
equations (the diffeerence between two resistor divider equations and
the Steinhart equation) and make a substitution. But you will find
that the change in the amplified voltage is related to the change in
temperature with the value of the thermistor resistance (Rt) in the
equation. Well heck, if I had and could monitor Rt then I wouldn't
need to go through all this other business. I also thought about just
monitoring the voltage change from one side of the bridge (the side
with the thermistor obviously) but again I think Rt will show up in
the equations plus I expect the change in this voltage to be extremely
small once my circuit is properly designed, tuned, and thermally
shielded.

Perhaps I am missing something obvious here. Does anyone have any
thoughts on the best way to monitor the temperature change and
stability from this circuit? Any help is greatly appreciated. If
Thanks!

Marc Airola

JHU-APL
Electro-Optical Systems Group

2. ### Spehro PefhanyGuest

If the other resistors in the bridge are fixed, you should have a
*fixed* nonlinear relationship between voltage and delta-T (from
balance). For small differences (< 1°C, the kind I imagine you'd see),
it won't even be that nonlinear. Zero output is when Rt is the same as
the resistor on the other side of the bridge, right?

Keep in mind that the measured stability of Rt is not necessarily an
indication of what's happening elsewhere temperature-wise.

Best regards,
Spehro Pefhany

3. ### Guest

I designed an AC bridge that self oscillated in unbalance and synch
demod this was for a crystal oven in 1965!.
Even at that time the offsets that could be achieved (already good
enough) from a discrete DC amplifier made AC working unnecessary.
The limitation in my applications has always been the thermal gradients
and the thermal time constants , even with a two term controller the
thermal improvement factor can seldom exceed 200.
ie a 200 temp change externally gives a 1 change at the point of
temperature sensing.
Note at temperature sensor not the average change in the controlled
volume.
Clever thermal design can achieve infinite improvement in the controlled
volume.
Easiest to design a double oven one inside the other for high thermal
gain.
Finally to answer your primary question ; as you are using AC activation
why not measure the thermistor with a 6 digit DC ohm meter
simultaneously.