Wiring thermal probes (diode) to obtain an average?

[Toby Lawrence]

I'm trying to wire four thermal probes in a way that I can get an
average temperature between them all, but I really can't figure out how
I'd do that, exactly.

For example, I stick all four probes into a big block of ice, and leave
them in the freezer, until each probe's readout is stable within 0.01.
(doesn't fluctuate more than that) If I wire them in parallel, and keep
adding thermal probes, the value readout goes down, but it doesn't go
down in a pattern and I can't derive any function to fit it. I would
much prefer a factual way to average them as accuracy is important to me.

If there is some alternative way I could wire the probes, or if you know
the math required to get an average, please please let me know.

Thanks in advance.

 

[Jonathan Kirwan]

I'm trying to wire four thermal probes in a way that I can get an
average temperature between them all, but I really can't figure out how
I'd do that, exactly.

For example, I stick all four probes into a big block of ice, and leave
them in the freezer, until each probe's readout is stable within 0.01.
(doesn't fluctuate more than that) If I wire them in parallel, and keep
adding thermal probes, the value readout goes down, but it doesn't go
down in a pattern and I can't derive any function to fit it. I would
much prefer a factual way to average them as accuracy is important to me.

If there is some alternative way I could wire the probes, or if you know
the math required to get an average, please please let me know.

Diode probe instruments, I think, provide a fixed current and measure
the junction voltage that results. The derivative of the solved diode
equation gives something like (8.617 uV/K * ln(1+I/Is)). So with an
operating current (I) of, say 10uA, and using a guess of Is=1e-15, the
slope would be something like 1.98 mV/Kelvin -- call it 2mV/K. That's
what the instrument will be looking at, for changes. It will also
need at least one calibration point.

Paralleling junctions would firstly operate each junction at a current
lower than the instrument expects, so the conversion of voltage to
temperature would be misguided. A new table may be needed or, at
least, a two-point calibration step. But placing junctions at
different places, meaning that they are potentially at different
temperatures, and paralleling them would force all their junction
voltages to be the same and thus, to simply adjust the balance of the
current branching through them. And what that would look like would
require some analysis. I think it would roughly mean that:
T1*ln(I1/Is1+1) = T2*ln(I2/Is2+1) = T3*ln(I3/Is2+1) = ...
And that all the probes would adjust their relative I1, I2, etc., so
as to balance that equation at their respective T1, T2, and so on,
with their somewhat varying Is1, Is2, etc. But that probably wouldn't
behave they way you want.

If you placed them in series, you might get something closer to an
average (a sum of the voltages, anyway), but then your instrument
probably wouldn't be able to support the voltage compliances required
and wouldn't really drive them properly. And in any case, its tables
or math probably wouldn't apply, and again you'd be in trouble.

I don't have any experience with this, so I'm just thinking off the
top of my hobbyist head. Perhaps someone can be specific for you.

Jon

 

[Bob Masta]

I'm trying to wire four thermal probes in a way that I can get an
average temperature between them all, but I really can't figure out how
I'd do that, exactly.

For example, I stick all four probes into a big block of ice, and leave
them in the freezer, until each probe's readout is stable within 0.01.
(doesn't fluctuate more than that) If I wire them in parallel, and keep
adding thermal probes, the value readout goes down, but it doesn't go
down in a pattern and I can't derive any function to fit it. I would
much prefer a factual way to average them as accuracy is important to me.

If there is some alternative way I could wire the probes, or if you know
the math required to get an average, please please let me know.

Thanks in advance.

Since you mention that these probes are diodes, I assume
that the readout is a current source; the resistance of a
diode is a *very* linear function of temperature (negative
slope), so by applying a constant current the readout just
has to measure the voltage across a diode and adjust
for a particular slope and offset. But the actual slope constant
differs among diodes, so if you put them in parallel they won't all be
seeing the same current. Putting them in series would be better, but
then the readout would see the sum of the 4 voltage drops
and would definitely not read right.

So you either need to make your own current source and
measuring circuit, or tweak with the one you have and hope
it can handle the larger voltage range. Alternatively, you
can always use 4 separate circuits and sum them together
afterward, with a divide-by-4 scale factor thrown in.

Best regards,



Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Home of DaqGen, the FREEWARE signal generator

 

[petrus bitbyter]

I'm trying to wire four thermal probes in a way that I can get an average
temperature between them all, but I really can't figure out how I'd do
that, exactly.

For example, I stick all four probes into a big block of ice, and leave
them in the freezer, until each probe's readout is stable within 0.01.
(doesn't fluctuate more than that) If I wire them in parallel, and keep
adding thermal probes, the value readout goes down, but it doesn't go down
in a pattern and I can't derive any function to fit it. I would much
prefer a factual way to average them as accuracy is important to me.

If there is some alternative way I could wire the probes, or if you know
the math required to get an average, please please let me know.

Thanks in advance.

Diode probes are not passive components like resistors. Most of them require
a well defined constant current through it so the voltage accross it can be
measured to find out the temperature. That's what the electronics in the
instrument do for you. If you wire some probes in series or parallel, the
measured voltage will be out of its normal operating range and the result -
if any - makes no sense. So to obtain your objective, the average
temperature, you'll have to adapt some other measurement technics. As I have
no idea of you application (and your budget), I can only shoot at random:

- Find a resistor based instrument. Four NTC (or PTC) probes can be wired to
look like one probe.

- Use four independent instruments. You'll have to find a way to read them
out and do the calculation. If you can read out the voltage for instance you
can "calculate" by some opamps. If the instruments provide some serial
output, you can use a PC to do the calculations for you.

- Build a system that connects the probes to the instrument one at a time.
Reed relays may do a good job for you. You'll still need a PC or a micro to
control the relays and do the calculations.

Next time provide some more information about the real problem and the
equipment available.

petrus bitbyter