Themocouple Voltage (Seebeck)

[mityeltu]

I have a problem. First, I'm out of my knowledge base. Second, I am working on thermocouples for a project and I am struggling to get my head wrapped around these... things.

Here's what I have. Type T thermocouple. I am sure that's the type. Red negative, Blue positive, brown jacket, rolled off reel with label stating Type T thermocouple. I say all that because I am baffled by what's going on. I'm sure it is simply my lack of understanding, but ....

All I'm trying to do is use this to measure temperature. I will eventually incorporate a thermistor to monitor cold junction temp for compensation, but right now I'm only interested in the thermocouple. Oddly enough, I have the thermistor circuit working flawlessly for monitoring ambient.

Now the real issue. All I am doing at the moment is measuring the tc cold junction with a Keithley 2001 DMM. Per the data sheet for the Type T tc, I should be reading approximately 0.911mv as the ambient temp is about 73°F or about 23°C. Here's the rub. The DMM is measuring -0.035mv. The DMM is calibrated by our lab and has been checked and verified to be working properly. I have tried 3 different sets of tc's to make sure it wasn't a fluke. All 3 measured similarly.

What am I doing wrong? Again, I presume it is just my lack of understanding, but I don't know what else to do to use these tc's to measure temperature. My plan was to simply connect them to my 24-bit sd a/d converter, add the thermistor cold junction voltage and wham-bam temperature. But what I'm measuring at the moment is just plain wrong somehow. Can someone give me a hint, push or whack up side the head?

 

[GoldDigger]

I have a problem. First, I'm out of my knowledge base. Second, I am working on thermocouples for a project and I am struggling to get my head wrapped around these... things.

Here's what I have. Type T thermocouple. I am sure that's the type. Red negative, Blue positive, brown jacket, rolled off reel with label stating Type T thermocouple. I say all that because I am baffled by what's going on. I'm sure it is simply my lack of understanding, but ....

All I'm trying to do is use this to measure temperature. I will eventually incorporate a thermistor to monitor cold junction temp for compensation, but right now I'm only interested in the thermocouple. Oddly enough, I have the thermistor circuit working flawlessly for monitoring ambient.

Now the real issue. All I am doing at the moment is measuring the tc cold junction with a Keithley 2001 DMM. Per the data sheet for the Type T tc, I should be reading approximately 0.911mv as the ambient temp is about 73°F or about 23°C. Here's the rub. The DMM is measuring -0.035mv. The DMM is calibrated by our lab and has been checked and verified to be working properly. I have tried 3 different sets of tc's to make sure it wasn't a fluke. All 3 measured similarly.

What am I doing wrong? Again, I presume it is just my lack of understanding, but I don't know what else to do to use these tc's to measure temperature. My plan was to simply connect them to my 24-bit sd a/d converter, add the thermistor cold junction voltage and wham-bam temperature. But what I'm measuring at the moment is just plain wrong somehow. Can someone give me a hint, push or whack up side the head?

The point where the thermocouple wires touch the DMM leads forms a distributed cold junction. If the hot junction and the cold junction are at the same temperature the circuit voltage will be zero.

What the cold junction compensator is supposed to do is insert a voltage into the loop equal to the difference between the actual ambient temp cold junction voltage and what it would be if the cold junction were at the standard reference temp, usually 0C.
The adapter into which the TC wires connect to adapt them to the meter jacks usually contains the actual cold junction. For type K (copper/constantan) TCs the meter is effectively inserted into the copper wire, so you have a well defined single point cold junction.

For other TC alloys you cannot avoid having multiple junctions of dissimilar metals where you insert the meter, so you want to keep all of the extra junctions at the same temperature.

 

[winnie]

GoldDigger is spot on with the problem.

A thermocouple is simply a junction of two different conductors. Every such junction develops a thermal voltage. This includes the junctions formed between the multimeter probes and the TC wire.

It turns out that if you look at a complete circuit, composed of a bunch of junctions, if the entire circuit is at the same temperature than all of the various junction voltages will add up to zero.

You will only see a net voltage if different parts of the circuit are at different temperatures.

The classic way of doing this is to make _two_ junctions with your thermocouple alloys, so you have :
<meter probe> to <alloy A wire> junction <alloy B wire> junction<alloy A wire> to <meter probe>. One is your 'measurement' junction, the other is your 'reference' junction. You put the reference junction into an ice/water bath. Since both meter terminals are connected to alloy A wires, the meter junctions balance out. The voltage that you measure will depend on the temperature difference between the two junctions.

These days the reference junction is handled by a temperature measurement at the meter end of the thermocouple wire, but the basic theory is essentially the same.

-Jon

 

[Smart $]

The point where the thermocouple wires touch the DMM leads forms a distributed cold junction. If the hot junction and the cold junction are at the same temperature the circuit voltage will be zero.

What the cold junction compensator is supposed to do is insert a voltage into the loop equal to the difference between the actual ambient temp cold junction voltage and what it would be if the cold junction were at the standard reference temp, usually 0C.
The adapter into which the TC wires connect to adapt them to the meter jacks usually contains the actual cold junction. For type K (copper/constantan) TCs the meter is effectively inserted into the copper wire, so you have a well defined single point cold junction.

For other TC alloys you cannot avoid having multiple junctions of dissimilar metals where you insert the meter, so you want to keep all of the extra junctions at the same temperature.

GoldDigger is spot on with the problem.

A thermocouple is simply a junction of two different conductors. Every such junction develops a thermal voltage. This includes the junctions formed between the multimeter probes and the TC wire.

It turns out that if you look at a complete circuit, composed of a bunch of junctions, if the entire circuit is at the same temperature than all of the various junction voltages will add up to zero.

You will only see a net voltage if different parts of the circuit are at different temperatures.

The classic way of doing this is to make _two_ junctions with your thermocouple alloys, so you have :
<meter probe> to <alloy A wire> junction <alloy B wire> junction<alloy A wire> to <meter probe>. One is your 'measurement' junction, the other is your 'reference' junction. You put the reference junction into an ice/water bath. Since both meter terminals are connected to alloy A wires, the meter junctions balance out. The voltage that you measure will depend on the temperature difference between the two junctions.

These days the reference junction is handled by a temperature measurement at the meter end of the thermocouple wire, but the basic theory is essentially the same.

-Jon

These are fine explanations except two things...

1) Type K is Chromel^®-Alumel^®. Type T is Copper-Constantan.

2) When using a Type T thermocouple, you will [essentially] only get _one_ cold junction. With other non-copper types, you get two. This is because DMM meter probes are [basically] copper... and copper to copper is not a TC junction. If you want to get uber technical, the probes are plated, but mated probes should be plated the same so you get offsetting junctions. After all, we are talking millivolts here. Best to always use mated probes/leads.

 

[GoldDigger]

Thanks for the correction!

 

[mityeltu]

Ok, so, I'm not even sure what question to ask at this point as I don't really understand the problem.

I don't understand this. I get the idea that I have more than one TC because of the connection to the DMM (dissimilar metals). So I really have 2: actual TC and another from the DMM lead to TC connection. The other connection is copper to copper, so no TC type junction there. Got that ... I think. So why the small negative voltage? If I had DMM leads that matched (materially speaking) the TC leads, would I read 0 mv with all junctions at ambient? If this is the case, then how in the world am I ever going to be able to measure anything with a TC? My connections will always be dissimilar somewhere along the line, I will have a reference temp at the cold junction, but if I turn that reference temp into a TC voltage and add it to the TC voltage, then I will still be pretty far off because of the weird effects at my TC interface (cold junction connections). It seems like I will always be calculating the wrong temp. And this is only going to get worse if I include an amplifier. Now the voltage I get has a giant unknown amount of error that I don't know how to compensate for.

How in the world is this done?

I know I don't understand this, but it seems to be alot more difficult than it should be.

 

[winnie]

Remember you need to have a temperature _difference_ to see a measurement. You need at least 2 junctions, where one junction is at a different temperature than the others. If all of the junctions in a system are at the same temperature, then the measured voltage should be zero.

As I described before, the _classic_ way to do this is to intentionally make two junctions, and then put one of the junctions into ice water (defined 0°C).

So for example you could have: <meter lead> to <Copper wire> (junction 1) <Constantan wire> (junction 2) <Copper wire> to <meter lead>
This then allows you to measure the temperature _difference_ between Junction 1 and Junction 2.

Current practice is to have a single measurement junction, and then at the instrument you have a temperature sensor (such as an RTD) that measures the temperature of your 'unintentional' junction(s), and then somehow compensates.

-Jon

 

[mityeltu]

Not sure I really get that yet (why I read the small negative voltage if all junctions are at the same temp, but I'll keep thinking about it), but... moving on.

So, if I measure the cold junction with my thermistor, back calculate that temperature to a thermocouple voltage then add it to the thermocouple voltage and re-calculate the temp using the tc polynomials, I should wind up with the absolute hot junction temp. Does that sound right?

 

[Smart $]

Not sure I really get that yet (why I read the small negative voltage if all junctions are at the same temp, but I'll keep thinking about it), but... moving on.
...

In your first post, you said -0.035mv. If you look this up in a reference table you will see that voltage indicates between 1-2°F (~1°C) relative. While that assumes a reference junction temperature of 0°C, your measured voltage at other than 0° at all junctions in the table will be a fairly accurate approximation of the relative temperature (difference). Add to that the typical accuracy of a Type T thermocouple is 1°C and I'd say the voltage measurement you are observing is nominal.