Temperature measurement

wrote:
I have a temperature measurement problem. I have implemented the
measurement
using two different methods, a platinum RTD and a thermocouple. Each
method
has its own unique problem.

Background: The measurement takes place in a small vacuum dewar. The
dewar
contains a CCD imager mounted on a thermoelectric cooler (TEC). The
cold side
temperature is about minus 40 degrees C. Signals in and out of the
dewar are
through vacuum sealed glass bonded pins (alloy 52). The pins have to
be
located in the heat exchanger plate that forms a thermal path for the
hot
side of the TEC to a heat sink. The heat exchanger plate rises 15 to
25
degrees above ambient. Therefore, the pins eventually rise to these
temperatures also.

Thermocouple Method: The Thermocouple leads are welded to the feed
through
pins. Because of the difficulty of vacuum sealing (and other
logistics
problems)using feed through pins of the same material as the two
thermocouple
wires is problematic. Therefore, the feed through pins are alloy 52
also. (I
should note that the feed through pins are compression sealed.) The
difference in pin materials might be OK since the heat exchanger is
fairly
massive and should be isothermal. Therefore, the temperature of the
two feed
through pins should be the same and by the law of intermediate metals
there
should be no temperature measurement error. Notice that there are a
couple of
"should bees" in the previous discussion. This method of measurement,
therefore, leaves me with a bit of uncertainty.

The isothermal condition is along the length of the wire and not
between the 2 wires. Going from the inside of a vacuum chamber to the
outside through a feedthrough could sustain large temperature gradients
across the feedthrough wire leading to measurement error that is
proportional to the delta T that you miss by using alloy 52 instead of
TC wires through this gradient. I can't really get a good picture of
your application so I can't say for sure what might be your conditions.

A way around this would be to measure the cold junction temperature at
the inside of the feedthrough and use homogeneous (e.g. copper) wires
from there on out to the instrument and then correct the readings later
on. But of course you said you don't have extra feedthrough wires for
the cold junction sensor so that idea won't work.

Two Wire Platinum RTD. This method does not have the dissimilar
method
problem. But, the RTD needs to be coupled to the temperature meter
(through
the pins) using high electrically conductive leads and as a result
high
thermally conductive leads. These leads are over 100 times more
thermally
conductive than the Thermocouple leads. As a result the thermal
loading as a
function of the cold side of the TEC and the hot pins creates a
temperature
measurement error, i.e., the heat conducted up the leads is trying to
raise
the temperature of the RTD and reducing the cooling affect of the
TEC. The
higher thermal resistance of the thermocouple leads greatly reduces
this
error. Changing the RTD lead material to something that has lower
thermal
conductance also increases the thermal resistance. This can be
compensated
for in the reading. Compensation wouldn't be required if we used a
three wire
RTD, but, we don't have a spare feed through pin for this. But,
getting an
RTD with high thermal resistance leads from the manufacturer is not
possible
and modifying the RTD ourselves in a production environment is not
cost
effective.

Have you looked at using Constantan wires? That's typically the best
low cost way to lower the thermal conductivity and thus reduce shunting
errors. You can measure the lead resistance and subtract it to correct
for lead errrors. Also don't forget to compensate for the
thermoelectric errors in resistance measurements when using DMMs. Read
Agilent's (HPs) website on resitance measurement for an explanation. I
don't know why you can't get RTDs from your supplier with special
leads. We can certainly provide them so double check with your
supplier.

An alternative to RTDs would be a thermistor or a diode. Check out
Lakeshore Cryogenic's offerings. With a thermistor you need to keep
the excitation current from a DMM low to avoid self heating errors.
With either RTDs or thermistors you can check how much energy you're
dissipating with a simple I^2R calculation.

Bill Schuh
Watlow



Does anyone have any suggestions?

Vince Kasprzak


-----------== Posted via Deja News, The Discussion Network
==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your
Own