6
\$\begingroup\$

I'm looking to prototype a low-cost thermocouple interface board. It needs to output a voltage, as opposed to using SPI like several of the ICs I have looked at, since it would be fed into a 10-bit ADC present on a servo drive. I found a potential IC in the AD8497, but now I have to figure out how to connect the thermocouple to the IC on my board. Long term, it would need to survive linear accelerations of up to 6m/s^2. Vibration is not much of an issue, but I haven't quantified that yet.

I have read that it is possible to solder a Type T thermocouple, but this requires a fairly expensive thermocouple (I found them for about $70), and I haven't yet found an amplifier IC... so I think I'm stuck with the Type K thermocouple.

I've read a helpful post that outlines a few connection methods, such as high temp solder, adhesives, and constant mechanical force, but none of those appeal to me. I'd prefer to go with a PCB-mounted mini receptacle like the one used in the Arduino Thermocouple Shield, but Omega seems to only sell them in lots of 100. I found another nice one, but could only find sources in India and the UK!

Can anyone suggest other options? I would have just bought Ryan McLaughlin's Single Thermocouple Interface since it's exactly what I need (but maybe not the right form factor), but he doesn't have them in stock anymore due to part shortages. Ideally, I would like the board to be flatter, so a different receptacle is preferable (like the RS Electronics one I linked to in the UK)

\$\endgroup\$
4
  • \$\begingroup\$ m/s is not a unit of accelleration. \$\endgroup\$ Commented Jun 25, 2012 at 14:48
  • \$\begingroup\$ Why do you need a voltage? That complicates things because offset errors are a big problem when amplifying such small signals. There are amplifiers for that, but lately I've been running thermocouples straight into high resolution delta-sigma A/Ds. That gets around the amplifier offset problem nicely. You could even use that method to read the thermocouple, then produce a PWM that gets filtered to make the analog signal, although digital is more useful in the first place. \$\endgroup\$ Commented Jun 25, 2012 at 14:50
  • \$\begingroup\$ I need a voltage because I'm limited to the 10-bit ADC that's on my servo drive... though to be honest, I had not checked into the ADC onboard. I also don't have any experience in this area, so I guess I need to learn more about these high-res sigma-delta ADCs that you have referred to. \$\endgroup\$
    – Dave
    Commented Jun 25, 2012 at 14:55
  • \$\begingroup\$ 6m/s/s is not much acceleration to worry about. It's less than gravity. \$\endgroup\$ Commented Jun 25, 2012 at 22:42

1 Answer 1

6
\$\begingroup\$

There is more to be aware of in the connection issue than "just making a connection:

I've read a helpful post that outlines a few connection methods, such as high temp solder, adhesives, and constant mechanical force, but none of those appeal to me.

That's a useful article and worth reading by anyone wishing to use thermocouples for measuring small low mass temperature sources BUT it is not principally aimed at electrically connecting the thermocouple to the measurement system - it's mainly about connecting the thermocouple thermally to the target heat source.The two tasks have quite different requirements.

I have read that it is possible to solder a Type T thermocouple, but this requires a fairly expensive thermocouple (I found them for about $70),

Type T thermocouples are Copper-Constantan. You can make your own from two wires of the appropriate sort, spot welded at the tip. Price can be under a dollar for tye actual thermocouple. Add sleeving, handle and beautification to increase price as desired. It is possible to but rolls of thermocouple wire which use the correct materials and which are intended for extending the thermocuople connecting lead. A length of this wire may be cut off, the end spot welded and a thermocouple produced at very low cost.

BUT soldering this material MAY produce problems if not allowed for - see below.

Cold junction issues:

It is crucial to realise (and you probably do) that the conductive materials used in the connector for a thermocouple will influence the result as they form a second thermocouple at ambient temperature.

Knowing both the ambient temperature at the point of connection and the materials used is required for "cold junction compensation". If you just solder the wires to eg a PCB or to some other wires (copper or tinned copper or ...) you will introduce differing metal-metal connections in each case and the low temperature thermocouples formed will be different, and probably unknown with any precision.

This is one reason why the tip of the thermocouple where the two wires join is usually welded rasther than soldered or brazed. Quite apart from melting point issues, introduction of another, or several other metals in a solder or brazed mix will create a new thermocouple of unknown characteristic.

Can anyone suggest other options? I would have just bought Ryan McLaughlin's Single Thermocouple Interface since it's exactly what I need

Ryan's interface is "more real" than may be obvious. As it notes in passing " ... while also using the proper metals in the connector required for a thermocouple connection ... " - ie, he knows his stuff. Additionally, the converter IC involved is located on the same PCB as the thermocouple socket so any additional junctions formed by track to IC pin transitions are (approximately) balanced in both leads and so any potentials generated are invisible to the system.

There are a number of connectors on the market which may suit you but, whatever is used must address the "correct connector materials" and "equal materials in both paths " issues.

The RS socket that you mention could be sent to you from afar off - and RS will send internationally if suitably encouraged.

Omega have a range of potentially suitable products and here and also here and this wire-wire connector of theirs looks adaptable to your purpose and seems to be available in 1's.

Here is a useful discussion of the low temperataure juntion issues that I mentioned above

comment in poassing on the same issues

BUT regardless of what you use - something from above or elsewhere,
you will need to be certain not to introduce extra "junctions" into the connection path
OR to introduce the same ones in both legs
OR be aware of what extra junctions are and how to allow for them AND to be aware that soldering may cause differences.

If soldered, any connector you use will ideally use the same leg material into the PCB.

\$\endgroup\$
1
  • \$\begingroup\$ Russell, thanks for the comprehensive answer. I'm going to start looking through your suggested links. As you stated, I am aware of the issues when introducing the connector, but honestly I figured that as long as I used one like Omega sells, things would magically work in my favor since the ICs I have looked at perform the cold junction compensation. However, in order for them to really work, I assume I have to use a specific connector? \$\endgroup\$
    – Dave
    Commented Jun 25, 2012 at 15:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.