I have two K-type thermocouples that look exactly like the one below. One was sold as good to 800C, and the other was quoted as having a temperature range from 0-400C. I have been swapping them between a multimeter and PID controller while trying to get control of an electric lead furnace, which consists of a steel pot, resistance heating coil wrapped around it a few times, and an outer shroud.

I have tapped the shroud in a few places to thread the head of the controller's thermocouple tight against the pot. And I am getting very bad results, using infrared thermometers to check the actual temperature at the contact points.

  1. The thermocouple pressed tightly against the outside of the pot has very high latency, and it never catches up. E.g., the pot will heat to 400C in two minutes. The thermocouple will increase its reading more slowly, peak a minute or two after the pot reaches steady temperature, and never indicate much beyond 250C.

  2. If I drop the other thermocouple loosely inside the pot, touching the side, it will react more quickly and get further, but it will still stabilize about 100C lower than the infrared thermometer indicates for its contact point.

Since this occurs with two separate thermocouples (that, for all I can tell, are of identical manufacture) I'm wondering if I am doing something wrong? For example, is there supposed to be slack between the nut and the head of the thermocouple? Or are these just poor pieces of equipment and I need to buy something better?

K-type thermocouple

  • 2
    \$\begingroup\$ Possibly, you are experiencing these problems because you have a grounded thermocouple probe, but you try to use it as a floating thermocouple. More here. \$\endgroup\$ Feb 23, 2016 at 2:41
  • 1
    \$\begingroup\$ Infrared thermometer may not read accurately on shiny surfaces without calibrating "emissivity" Have you verified the temperature. A grounding issue is possible, but otherwise how are you terminating at your pid? You must use special alloy wire for thermocouple leads if extending. The thermocouple is actually just a soldered wire inside the nut, which adds a thermal mass and attachment point. Loosening the nut will affect the thermal junction, yoh should use thermal grease (if possible) and tighten well. \$\endgroup\$
    – crasic
    Feb 23, 2016 at 2:51
  • 1
    \$\begingroup\$ @crasic: The IR thermometers are reading correctly: The pot has acquired a dark, matte patina. I'm connecting the thermocouple's leads directly to the PID's contacts. Both PID and multimeters read correct temperature when the temperature is low. The nut rotates independently of the nub at the head of the thermocouple. The nut threads into the shroud and, when tight, holds that nub in tight contact with the pot, which is what I want the temperature of. \$\endgroup\$
    – feetwet
    Feb 23, 2016 at 4:17

2 Answers 2


The most likely cause of a relatively large error in that direction (if the junction is actually at the correct temperature) is a double-reversal of extension wire. Since color codes vary by region, the most reliable way to identify the polarity (assuming you are sure the extension is type K) is to match magnetic to magnetic (positive) and non-magnetic to non-magnetic negative).

Proper polarity extension or thermocouple wire must be run all the back to the measuring instrument.

It is extremely unlikely that anything is wrong with a thermocouple- they are far more reliable and stable than IR sensors. If thermal contact is poor the junction may well be reading correctly, but be considerably cooler than what you are trying to measure- the leads draw heat out.

Ideally, immersion length should be 10-15x the diameter of the probe.

Edit: If you cannot achieve that try shielding the probe from air (make an insulated cup against the vessel) and running the leads against the vessel so they don't leach as much heat from the probe tip. You would probably do better with a bead thermocouple- at least you can visualize it more easily.

  • \$\begingroup\$ I assume that if I get a precise ambient temperature reading when I connect the thermocouples then I have connected them properly, right? The thermal contact is tricky: I'm using the nut to hold the head of the thermocouple against the surface I want measured. There's no way to immerse it because the contents of the pot are corrosive and mixing. Is there some high-temperature thermal grease or thermally-conductive contact patch that could mitigate that? \$\endgroup\$
    – feetwet
    Feb 23, 2016 at 4:22
  • \$\begingroup\$ Yes, You should get close to ambient if connected correctly and the sensor sitting out, and if you short the T/C terminals of the instrument with copper wire you should get the temperature at the terminal block (typically a bit warmer). You an try the sensor in an ice-water slurry- should be very clos to 0C and boiling water (around 100C, but possibly not as close). \$\endgroup\$ Feb 23, 2016 at 7:29
  • \$\begingroup\$ Aren't these essentially "encased bead" thermocouples: I.e., isn't the resistive component still just a little bead at the tip of the leads? Then I see that the problem is the "case" and even this one's sheath are strong thermal conductors that will pull the measurement back towards the ambient temperature. \$\endgroup\$
    – feetwet
    Feb 23, 2016 at 15:21
  • \$\begingroup\$ Typically the junction is right at the tip of the tubing. From your description ,the tubing etc. is not helping. \$\endgroup\$ Feb 23, 2016 at 16:02

The first thing you need to check is if the thermocouple is actually calibrated correctly: Do you have a calibration certificate with the probes? If not you could well be using a probe that is out of tolerance to start with.

The major issue from what I can see is that you are not using the correct method of measurement. Attaching the thermocouple to the outside of the crucible means you are only measuring the outside temp of the crucible and not the temperature of the material inside. You may have all sorts of drafts that could easily create big temperature anomalies.

Check your PID is set to type K and not a different type. Many PIDs have either a set thermocouple type or if it accepts all thermocouple types ensure its set to Type K and not say Type J.

In truth this is a very poor system of measurement. If you have exposed the Type K element to temps over 250 degrees C you have already drifted it and so you don't really ever know what it's reading.

To do this properly you are going to need to buy a Type S or R probe with a ceramic sheath material suitable for temps up to 1100 degrees C. You can then insert this into your molten liquid directly, it will always have a degree of latency, the smaller the probe's cross section the faster it will respond but the more fragile it becomes.

Using a multimeter is a no-no. You have no cold junction (unless you are using a multimeter designed for thermocouple use which has automatic cold reference... cheap ones are useless). The multimeter used in our calibration lab for example is over £5,000 and that is backed up by a dedicated cold junction reference unit.

Measuring temperature is a lot harder than most people think!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.