Some questions about thermocouples.

[UltraTrunks]

So, I'm getting started on a little project to build my own reflow oven out of a toaster oven.  I've found this little inexpensive K-Type thermocouple for my project.  I've done a little reading on thermocouples and the magic seems to happen when you have a junction of 2 different metal types.  So presumably, the positive lead is one metal type, and the negative lead is another.  I've also read that, when "extending" your thermocouple, you need to use special extension wire as mention here and here.  I'm guessing the same kind of wire that's used in the original leads are used so that you don't form new junctions of different metal types.

So this is where I have some questions.

• Well, unless your contacts inside your multimeter or circuit are made of those special metals, wouldn't you ALWAYS be ruining your reading by forming junctions by simply attaching your thermocouple to your multimeter or circuit?
• I experimented a little bit and used the alligator attachments on my multimeter test probes to attach the thermocouple to my Fluke.  When I took the temperature reading it appeared to be correct.  I'm sure my test probes are not matched with the same metals in thermocouples, so technically I "extended" the thermocouple with the wrong metal types.  So why did it still work?
• So what is the best way to connect a thermocouple to a circuit, because it seems that you can ruin your measurements if you do it wrong.

[apelly]

Dave has a pretty comprehensive video on thermocouples. I can't be bothered finding it for you right now, but it's well worth a watch. I suggest you look for it.

[nsayer]

Thermocouples work effectively by generating a delta voltage that represents the temperature differences between *two* junctions. The first is at the business end of the thermocouple. The other is at the end where the measuring apparatus is.

Most thermocouple amplifiers have a local thermometer, which they use to stake the temperature at the "local" end. So as long as the place where the thermocouple terminates (that is, plugs in) is at the same temperature as the temperature measuring thing for the thermocouple amplifier, you've got no worries.

If you improperly extend a thermocouple lead, then you've moved the termination point from where the amplifier is to the place where the incorrect extension ends. If that isn't at the same temperature as your reference, then your measurements will be wrong.

I learned all this from working on Toast-R-Reflow. I also make and sell a little AD8495 thermocouple amplifier breakout board that may help you if you're breadboarding a reflow oven controller.

[nctnico]

A high impedance, low offset op-amp will also do fine to amplify the thermocouple signal and feed it into an A/D converter. IIRC an amplification of 30 or 40 is more than enough.

[mikeselectricstuff]

A high impedance, low offset op-amp will also do fine to amplify the thermocouple signal and feed it into an A/D converter. IIRC an amplification of 30 or 40 is more than enough.

Do TCs need a particularly high impedance?

[poorchava]

It seems that they do.  When I was developing amplifier for my jbc handpiece controller, I used a BAT54 to clamp the input from the TC. The leakage through this diode was enough to screw up readings. For reference,  1N4148 worked just fine.

JBC uses a proprietary thermocouple composition,let though,  so it may have had some influence.

[hamster_nz]

That thermocouple only goes to 250C- close to the 230C or so you will be aiming for. Most likely due to the insulation melting or burning.

I used http://www.adafruit.com/products/270, which had glass braid. It is good to 500C if you remove the heatshrink...

I also used this amp/ADC board - http://www.adafruit.com/products/269 - was a bit expensive but all worked fine.

Mike

[SeanB]

A high impedance, low offset op-amp will also do fine to amplify the thermocouple signal and feed it into an A/D converter. IIRC an amplification of 30 or 40 is more than enough.

Do TCs need a particularly high impedance?

They do not, but if you are using them as a current source you do need to control resistance in the circuit. Most simple industrial indicators ( dial type) use a fixed line resistance and a variable calibration resistor to set the desired full scale current. Works well for high temperature readings where you do not really worry about ambient temperature cold junction compensation, where you are reading with a 10-50C resolution and either take ambient to be 25C or ignore it.

[LukeW]

I'm guessing the same kind of wire that's used in the original leads are used so that you don't form new junctions of different metal types.

That's right.

http://en.wikipedia.org/wiki/Thermocouple

And there is never just one thermocouple junction, there are at least three - the actual measurement junction and two reference junctions.

Any accurate thermocouple circuit requires a thermistor or some other temperature sensor next to the reference junctions, which is used for cold-junction compensation to correct for the temperature of the reference junctions.

If you look at a thermocouple socket like this:

http://www.omega.com/Temperature/images/PCC-OST-SMP_m.jpg

It's made of certain metals welded to a copper lead at a certain point, allowing the cold-junction temperature sensor to be mounted so that it is always very close to the cold junction, the cold junctions occur at a consistent place, and they are always very close together meaning that their temperatures are the same.

[UltraTrunks]

Dave has a pretty comprehensive video on thermocouples. I can't be bothered finding it for you right now, but it's well worth a watch. I suggest you look for it.

I found it, EEVBlog #419



Good video, answered most of my questions   I feel vindicated as well as some of my suspicions where actually onto something.  I had no idea about needing a thermometer to operate a thermometer though  .  I never would have guessed you needed a cold junction temperature sensor to properly operate a thermocouple.  Nothing is ever easy is it?

I used http://www.adafruit.com/products/270, which had glass braid. It is good to 500C if you remove the heatshrink...

I also used this amp/ADC board - http://www.adafruit.com/products/269 - was a bit expensive but all worked fine.

These are pretty much perfect for what I'm doing.  I'm going to give it the ol' college try first and do it myself because this is more about learning for me than actually having a reflow oven.  But its nice to know I can buy my way out of trouble if it its not working out the way I hoped.