Thermocouple junctions

[cheese9988]

I am learning about how thermocouples work. I am trying to understand
why it is necessary to have a junction for you to accuratly measure a
thermocouple. I understand from another topic that with any wire, when
heated produces am emf due to energy being given to that particular
section on the wire. Now, do you need the disimilar metal with a
separate cold junction to produce what would be two separate emfs, and
through some math be able to determine the temperature from that? What
is it about the dissimilar metals with the extra cold juntion that
makes it easier to measure?

 

[Spehro Pefhany]

I am learning about how thermocouples work. I am trying to understand
why it is necessary to have a junction for you to accuratly measure a
thermocouple. I understand from another topic that with any wire, when
heated produces am emf due to energy being given to that particular
section on the wire. Now, do you need the disimilar metal with a
separate cold junction to produce what would be two separate emfs, and
through some math be able to determine the temperature from that? What
is it about the dissimilar metals with the extra cold juntion that
makes it easier to measure?

Circuits involve a loop. How do you propose to connect your proposed
homogenous wire with a temperature gradient to a voltmeter without
effecting another temperature gradient which precisely cancels out the
voltage resulting from the first one?


Best regards,
Spehro Pefhany

 

[John Popelish]

I am learning about how thermocouples work. I am trying to understand
why it is necessary to have a junction for you to accuratly measure a
thermocouple. I understand from another topic that with any wire, when
heated produces am emf due to energy being given to that particular
section on the wire. Now, do you need the disimilar metal with a
separate cold junction to produce what would be two separate emfs, and
through some math be able to determine the temperature from that? What
is it about the dissimilar metals with the extra cold juntion that
makes it easier to measure?

Any piece of conductor creates voltage when it passes through a
temperature gradient. Unfortunately, you can't complete a circuit to
measure that voltage, except by having another conductor also pass
through that same gradient. If you complete the circuit with the same
metal, you get two equal and opposite voltages that cancel, and you
get nothing. But if you use two different metals to pass in and back
out through the gradient, their different voltages produce some net
total.

In order to have that voltage represent the total temperature change
from a known temperature requires that that couple be in series with
another to a known temperature (usually at ice temperature) or in
series with a temperature controlled active voltage source that
simulates this second couple.

 

[John Larkin]

I am learning about how thermocouples work. I am trying to understand
why it is necessary to have a junction for you to accuratly measure a
thermocouple. I understand from another topic that with any wire, when
heated produces am emf due to energy being given to that particular
section on the wire. Now, do you need the disimilar metal with a
separate cold junction to produce what would be two separate emfs, and
through some math be able to determine the temperature from that? What
is it about the dissimilar metals with the extra cold juntion that
makes it easier to measure?

Another way to look at it is that you need an asymmetry in the
circuit. If you used a single metal type to make a wire loop, intended
to measure temperature, what would be the polarity of the result? Why
would one output be positive or negative relative to the other? OK, it
wouldn't.

The thermocouple uses different metals so that when you create the
voltage gradients as a result of temperature gradients, the two legs
are different so you get something with a predictable voltage (and
polarity!) that you can measure.

John

 

[cheese9988]

So if I understand this right, the other conductor creates a different
emf, and becuase of the difference a small current is allowed to flow?

 

[cheese9988]

So if I understand this right, the other conductor creates a different
emf, and becuase of the difference a small current is allowed to flow?

 

[John Larkin]

So if I understand this right, the other conductor creates a different
emf, and becuase of the difference a small current is allowed to flow?

Here's a type K thermocouple:


chromel
________________________
/ B
/
T1 A * T2
\
\________________________ C

alumel


The wires are made of chromel and alumel, and are welded together at
A. The junction is at temp T1 and the other ends are at T2. We call T2
the "reference junction temperature". Call the absolute circuit
voltages A, B, and C.

Each wire generates a voltage that depends on the temperatures at T1
and at T2. These voltages are A:B and A:C. Note that these voltages
depend on both T1 and T2, *not* (accurately) on the difference T1-T2.

For a given T1 and T2, the chromel generates some voltage A:B, and the
alumel makes A:C, and we get to measure the difference B:C.


Now add a voltmeter, with copper-wire probes, to measure B:C...


chromel B
________________________ copper
/ *---------------+
/ |
T1 A * T2 meter
\ |
\________________________ |
*---------------+
alumel C copper



Heat up junction A, and the voltmeter indicates positive voltage,
around 40 microvolts per degree C.

We've created two new thermocouples * and *, but they cancel out, so
we still measure voltage B-C.

We don't want current, usually, we just want to measure the voltage
with the meter. If we know temperature T2 and the metered voltage, we
can calculate T1. Thermocouple wire has high resistance relative to
copper, so we usually want a very high impedance voltmeter to avoid
loading the loop and making errors.

Sometimes you do want current, like for holding in the gas solenoid on
a water heater, where the pilot light heats the thermocouple. But in
that case you're not trying to measure temperature.

I'm writing this same stuff today, in a manual for a thermocouple
simulator. Good practice.

John

 

[John Popelish]

So if I understand this right, the other conductor creates a different
emf, and becuase of the difference a small current is allowed to flow?

No. Ideally, the current approaches zero. A metal passing through a
thermal gradient is a thermal battery (different metals are batteries
that produce different millivolts per degree), but you get a most
accurate measure of the difference between the two battery voltages at
zero current (so wire resistance doesn't get involved).

 

[cheese9988]

So the difference between ab & ac is what your looking for. The two
produce two different voltages.
The copper wire really has nothing to do with the circuit? So long as
it is either kept in an ice bath as a reference or temperature
compensated, because if you know what the voltage produced at the
reference would be, then you can figure out the voltage per degree of
the thermocouple would be from that reference?

 

[John Popelish]

So the difference between ab & ac is what your looking for. The two
produce two different voltages.
The copper wire really has nothing to do with the circuit?

It completes the circuit to a voltage measurement. It will have no
effect on the voltage, as long as the temperature is the same at both
points where the copper connects to the thermocouple leads and also
the same temperature (though possibly, a different temperature) at the
far ends where the voltage measurement is made. That is, any voltage
generated by one copper wire will be balanced by an equal and opposite
voltage from the other copper wire, as long as the pairs of ends share
temperatures.

So long as
it is either kept in an ice bath as a reference or temperature
compensated, because if you know what the voltage produced at the
reference would be, then you can figure out the voltage per degree of
the thermocouple would be from that reference?

There may be a second thermocouple in series with the measurement one
that is kept at ice temperature, so you know what temperature to start
with when you calculate the degrees from the volts. But many systems
just measure the temperature of the terminals where the thermocouple
wires connect to copper, and either generate a voltage based on that
temperature (to simulate the voltage generated between an ice bath and
the terminal temperature) or communicate this temperature to the
measurement process so it can calculate the couple temperature
starting at that temperature.

 

[w_tom]

This temperature compensation is also why some IC
manufacturers make cold junction compensation ICs. See app
notes for those ICs to better appreciate the concept.