At 09:27 AM 07/12/2010, you wrote:
>Is there really one reference junction, though?

The main effect is from the pair of junctions where the thermocouple
materials connect to (typically) copper.

>I am under the
>impression that any joint of dis-similar metals in the thermocouple
>circuit will cause a voltage difference which is temperature
>dependent. This includes even things like bond wire connections inside
>the IC used to read the thermocouple.

Usually that is a relatively small effect, and usually it (mostly)
cancels out in a well-designed circuit. If you want to invite problems,
place high power devices (eg. an SSR or a hot linear regulator) on the
same PCB such that there are large gradients.

You need a good 'cold junction' (which might actually be the warmer
junction in some applications) measurement to get an accurate reading.
You'll get around one degree error for every degree your sensor and
measurement of the sensor does not match the actual 'cold junction'
temperature (with most thermocouples- pathological examples like 'B'
notwithstanding).

>  So, I'm not sure that it makes
>sense to couple the cold junction sensor so closely with the TC
>connector unless you are also coupling it closely with the rest of the
>electronics in the circuit which reads the TC.

You need to measure the points where the T/C alloys connect to
copper-- ideally all three are exactly the same temperature. It
depends on the accuracy you need. For better than a few degrees
C you probably shouldn't be using the PCB type of connector at all.

>Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the rewar=
d"
speff@interlog.com             Info for manufacturers: http://www.trexon.co=
m
Embedded software/hardware/analog  Info for designers:  http://www.speff.co=
m



--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.