Oh dear...... this is a simple question that needs to be answered by looking at so many factors.....

You say a general temperature measurement, so no clues as to accuracy there.... what about ambient conditions? etc... etc...

Not being helpful am I?

Generally a thermocouple is good for general measurements as its easy to use and replace, its not so good for high electrical noise environments or where the sensing is remote from the instrument.... due to its low signal...

If the ambient temperature is changing, i.e. the instrument is handheld sometimes, outside inside etc... then the temperature compensation of the cold reference junction will effect accuracy....

If you're talking about general temperature measurements why not just use a thermister? Much easier to use and replace for low temperatures....

There is no definitive answer to your quetion without much more details I'm afraid....

John.

Simply by looking at the issues you have raised, it seems that there will be no 1 general solution for all applications. However, I shall attempt to tie down this particular instance a bit better.

Accuracy: we would like a reading which is certainly within 5 DegC of the actual temperature (preferrably within 2.5deg).

Ambient conditions: This varies even in the same application. External temps could be around 50degC, coolant temps around 80degC, internal temps could be up to 160degC (maybe 180).

General: although normally we use externally connected and sometimes handheld apparatus, this particular application will have a dedicated condition monitoring unit fitted on the equipment itself.

Platinum probes are generally used as a primary control TC on percission furnaces and testing equipment. I would recommend going to the Omega site and researching your specific needs in the use TC technology. Not being a smart annul type of person, it is really not as simple as you might think to configure your needs. For general temps between -40 to +750*F, you have a simple choice of either type "J" or "K" TC.

If you are going for extream temps, then you would be using type "R" or "S" TC.

Here is the link to the perfact site and information for you too relish through.

http://www.omega.com/toc_asp/frameset.html?book=Temperature&file=SEL_GUIDE_THERMOCOUPLE

Good Luck.

Omega is a great site to find anything for intrumentation, they have a huge selection for temperature. You can request their free catalog for temperature, pressure and such.

But back to the question, RTD,s work great as long as you have the room to install them. Thermocouples are just as accurate, but in my experience with them they tend to fail more. Cold junctions coming apart, leads breaking apart. Another thing I have found is that RTD's are easier to troubleshoot and replace, but everybody has their preferences.

Having worked in the instrumentation/process control industry for many years, it has been the normal practice to use Platinum RTD's for most measurement.They are accurate enough for general use, very rugged, and easy to troubleshoot and replace, especially if installed in a well.We use the European curve, which is a little more accurate.I recommend using at least 3 wires for remote locations, to compensate for lead resistance.4 wires if you really want to split hairs.(The extra leads are simply place in the opposite sides of a wheatstone bridge from the sensing element, therby cancelling the lead resistance)Usually,in the converter, the output of the Wheststone is applied to an op-amp for amplification, and then to signal conditioning/a/d conversion,scaling, etc.

We use thermocouples in very high temps, like boiler grates.They have a very poor durability recors.

Hope this helps.

Adding to the mix, have you considered using shielded vs. unshielded designs? A long time ago, while testing comressors and turbines to ASME PTC-10, the unshielded designs tended to fail at the rate of 1 or 2 (and sometimes more) per test. On one compressor, the design inlet volume was 350,000 cfm. The massive moving volume vibrated the TCs apart. The shielded designs had a slower response and the shield had to come to temperature too. But, as we wanted a stable temperature for each test point, this was a good thing. We wanted a long time constant. While initially more expensive, the time NOT lost replacing the damaged TCs paid for a new set for each test.

Yes, good point. However, these also take up a lot of room where space is already at a premium. There will need to be so many of these sensors (probably thermocouples now) since this is a development prototype and we are trying to get as much data as possible about operating conditions.

And... we haven't even begun to think about how all these leads are going to be brought out to the condition monitor!

And... it's still supposed to be a professionally finished article presented to the customer (not the development prototype created without the specialist tools that would normally be necessary in production)!

And... it's still supposed to look as much like the final production thing as possible!

Not that I'm complaining, after all it's these sorts of challenges that bring us into engineering in the first place.

Maybe I did not read about that "space" problem in earlier posts. Running bunches of wires together may contribute to an extra electomagnetic signal that affects the apparent temperature. The guys at Ansoft have some (very expensive) software that examines this effect. While your design is not nanoscale (think about the path-to-path interference in sub-micron ICs) there may yet be some impact of running too many wires together. You may be able to get some experimental single vs. multiple readings to ensure there is no significant interference.

Yes, yes. The general environment is one wherer there are some very high voltages loafing around. Generally there has not been a problem but then normally we have chosen RTDs for this kind of measurement. We are only re-visiting this now because there is a chance, if we're successful, of producing high numbers and so need to keep costs low. Also, we seem to have lost along the way the reason for having selected RTDs in the 1st place!

Space wasn't mentioned previously simply because the real issue was trying to decide which technlogy to use. If we feel interference might be an issue then we can simply switch everthing off and see if there is any sudden change in apparent temperatures. Thermal mass is quite large and actually temperature continues to rise even after switching off.

Many thanks for everyone's input.