EGT

[ducharme at eiss1.psf.ge.com]

Fred Miranda wrote:

> can someone explain "cold  junction"
> does it have to do with the voltage produced at the junction of
> the thermocouple wire and your measuring circuitry?

> how do you interface a thermocouple? a cmos op amp?


A thermocouple is a junction between two dissimilar metals which produces a
voltage dependent on temperature.  When you join a Chromel and an Alumel wire
(type K thermocouple used for EGT's) you have a junction.  You also have two
other junctions that are also thermocouples:

  
      Chromel        Copper
               - v2 +
  +   /-----------^----------------    +
 v1  <                               Meter
  -   \-----------v----------------    -
        Alumel  + v3 -  Copper        


The voltage at the meter is v1 + v2 + v3.  In a laboratory environment, you
could put junctions 2 and 3 in an ice bath, and read the voltage v1, and
by using a thermocouple reference table, look up the temperature.  In the real
world, most measuring devices either tie junctions 2 and 3 to a thermal mass of
known temperature, or in the case of an analog meter, dismiss it entirely.  The
reference junction temperature (- 32 deg. F) is then added to the measurement.  
Another method is to use a cold junction compensator, which is a temperature-
dependent voltage source.  National Semiconductor makes one, although I don't
remember the chip number.

To measure the voltage, an opamp wired as a differential amp or high impedance
single-ended DC amplifier can be used, or special purpose instrumentation amp
can also be used.  Differential amps are required when the junction is grounded.
While laboratory instruments have a high input impedance (100 Meg is typical) 
as long as the junction is of sufficient cross-sectional area and the overall
wire length isn't outrageous, a differential amp with a 200K ohm impedance will
give reasonable results.  Remember, a thermocouple can deflect an unpowered 
analog meter movement! 

Most types of thermocouples are somewhat non-linear, although for EGT, type K 
is reasonably flat (maximum variation of 10 degrees through the range of 
800-1800 F). 

Cliff Ducharme