As I recall, Omega Engineering has good literature on this, and is probably Web-accessible.

Hi there,

The recommended practice is to always use an instrument with full isolation as even insulated thermocouples will suffer a breakdown of insulation.

The TC tip will then have contact with the sheath and associated well thereby creating a ground loop and introducing measurement errors.

Regards,

Craig

But in some plastic injection molding machines that I work on have both types. Of course, the controller also is different for accepting these inputs. But having said that, I could not clearly understand from application stand point why we need to choose between either.

The isolated thermocouples are generally quieter and free of ground loops. The down side is that the electrical insulation also provides thermal insulation. This gives them poorer transient response.

So, it means ISOLATED types are used when the distance between sensing and measuring is quite far and also when the temperature variations are more "sluggish." Whereas, the NON-ISOLATED can be used where the distances are short and temperature variations are "rapid."

some controllers will only work with isolated, [allen bradley plc]

hi limit cards may be more reliable with simple non-isolated type

different parts of the process don't require faster response times, such as molds heated by cartridge heaters, The controller may have a more complex program to account for the distance between the heaters, thermocouple & working surface.

That's generally true, although even short thermocouples are susceptible to RFI and can be made quieter by enclosing them in an isolation sheath.

On a previous job I tested large centrifugal compressors. One had a rated input of 350,000 cfm. To test these machines you needed to get the exact inlet conditions (temperature and pressure). In the high velocity input and outlet streams, the isolated thermocouples with the exposed bead lasted from hours to minutes. The bead would vibrate off. The jacketed (read non-isolated) lasted for many tests. So the answer is, it depends on what environment the TC is exposed to. The cost to replace a busted thermocouple may well be worth a slight decrease in accuracy. Or you could measure the stray voltage and do a direct compensation and get the benefits of a more stable temperature.