Thermocouple makes erratic readings when the probe casing is touching a ground at high temperatures

[ETULK]

I am having a rather odd problem with my Thermocouple measurement system.

To describe the setup:

I have k-type thermocouple probes which are wired into TC02 boxes inside an sc2345 unit.

This sc2345 unit is then wired into a NI card on the back of the computer (can look up unit if its required)

we are measuring the potential drop through a metallic sample as we heat it up to 1100 deg C.  A current of ~2A is applied to the sample and the potential drop is measured with a dmm.

 The thermocouple measures temperature properly until it is around 800 deg C.  At this temperature and above IF the thermocouple probe is touching the metal furnace tube the temperature signal drops off significantly.  When the probe is not touching the tube at high temperature, it makes the correct reading.  

When the thermocouple probe is hooked up to an auxiliary TC measurement system AND is touching the tube at a high temperature, it makes the correct reading.  This indicates (to me) that there is a problem not with the thermocouple or probe, but the DAQ system AND that the change in signal that this temperature probe revives is only picked up by the DAQ system and NOT a thermocouple DAQ device.

Any help?

[Kyle_A.]

Hello,

How is your thermocouple wired up?  Have you grounded the thermocouple?  If you have, be sure to check that ground and make sure it stays at ground through the entire heating cycle.  It could be possible the metal furnace tube changes potentials as heating occurs that causes the measurement to skew.  Does current flow through your thermocouple at all?  Is it shielded? 

[Henrik_Volkers]

While I also suspect a grounding issue, a second thought is the element isolation. At such high temperatures not all materials, even if they can withstand the temperature, still behave as good isolators.

[Henrik_Volkers]

How about a pyrometer?