Cold junction temperature measurement in Multichannel Thermocuple

[electronx]

In a multi-channel thermocuple signal conditioner (for example, 5 channels on a single PCB), should I add 5 separate junction temperature measurements to the thermocuple input of each channel to measure the cold junction temperature, or is one temperature sensor on the PCB sufficient?

or

[TizianoHV]

Usually one sensor is enought. But you should add a ground plane and keep away heat sources to reduce temperature gradients .

You can search for agilent 34901A, it's a scanner card, it uses two sensors for 20 TC channels (and there's a big gnd plane).

[mzzj]

Really depends on your accuracy target. CJC compensation is the weak point in many otherwise excellent dataloggers if you want anything better than 1-2 Cel accuracy.

[voltsandjolts]

Yeh, lots of CJ measurements or alternatively ensure good thermal coupling between CJ's.
e.g. the iso-thermal CJ copper block in the Keithley 2001-TSCAN card.

[electronx]

[MathWizard]

On the 34901A scanner card, whats the row of grey things above the relays? It looks like terminal blocks, but this is a DMM or something, is it for some external connections if you needed them? Is there wires that get screwed there? Or is it a pin header connector, and something snaps into it ?

Maybe it's some block of trimmers?

[nfmax]

They are terminal blocks. The card fits in 34970A & 34972A dataloggers, normally you hard wire the inputs to the scanner cards.

[Berni]

All depends on how much accuracy you need.

Lowest: Use a existing temperature sensor for ambient temp and assume the scanner card is at that temperature.
Low: Use a temperature sensor to measure the actual temperature of the scanner card.
Medium: Use a single temperature sensor, but thermally bonded to keep temperatures similar.
High: Use a dedicated temperature sensor next to each connection point.
Highest: Use a dedicated temperature sensor that is thermally bonded to the TC connector terminals using a special TC connector.

Usually only the professional dedicated thermocuple measurement instruments do that highest level, since the simpler solutions are usually close enough for what is needed. If you want accuracy go for a high quality PT100 or something like that