On Fri, Oct 18, 2013 at 05:06:46PM -0600, Ben Brockert wrote: >You need to differentiate there from the concept of a thermocouple, >i.e. two dissimilar wires joined together in a loop, and a >thermocouple probe, which is what would be installed through the >Swagelok fitting they described. The thermocouple probe is a >thermocouple junction inside a metal sheath, with the actual junction >touching the inner tip of the probe. The wires are otherwise >electrically insulated inside the probe. There's no need to >electrically isolate the thermocouple probe from the jacket when you >want to measure the temperature of the chamber, and there is no need >to isolate the thermocouples from each other with any competent >thermocouple amplifier. Ah, okay, yes, that would do it... so long as the insulation in there doesn't melt, of course, or get crushed too hard when the fitting is tightened down. But I suppose these things are insulated internally with the sort of woven fiberglass sleeves that I've seen in high-temperature applications, which is neither prone to melt easily nor crushable easily. (I think that stuff is usually fiberglass, but doing a web search, I see other materials are used too, some with even higher temperature ranges.) >Also, thermocouples as used for temperature sensing are a voltage >source, not a current source. The voltage is, by magical physics, >always and exactly equal to the thermocouple type's Seebeck >coefficient times the temperature difference That's then run through >an amplifier which is also responsible for compensating for the >temperature of the junction inside the amp. While in theory a >thermocouple could be measured as a current, it would be significantly >more complex. Nevertheless, that voltage source doesn't have infinite ability to source current; it's not quite _that_ magical. And if you have a situation where two nominal "voltage sources" are fighting each other (as you would if you had one junction near the inner wall and one near the outer), their relative ability to source current is what determines the resulting voltage (and thus the measured temperature). -- Norman Yarvin http://yarchive.net/blog