We're currently trying to build a portable temperature sensor for the purposes of estimating human core temperature. This is a bit of an engineering problem (more to do with choice of materials than choice of IC), but it does involve electronics and related issues like heat transfer, so I feel it's relevant enough to this board. Please correct me if I'm wrong in this regard.

There are quite a few articles out there on various attempts to achieve this, but we've been having trouble finding a solution that fits our criteria.

Since the device is portable and must be in contact for extended periods of time, invasive (rectal, esophageal) or cumbersome (axillary, in-ear/tympanic) methods of measurement are infeasible, leaving skin temperature as the next best alternative, since core temperature can normally be fairly well estimated from these values given certain constraints, such as known position on the body, shielding from environmental effects and so on. Contactless detection via infrared is unworkable since the uncertainty is greater than ±0.5°C, which is outside the acceptable range for clinical significance (i.e. the uncertainty needed in order to diagnose issues such as fever). Thus, the sensor needs to be in direct contact with the skin.

However, this creates further issues, the most pressing of which is the immediate cooling of the skin upon contact with the sensor, which will invariably be the temperature of the environment rather than the skin. While the skin will eventually return to its previous temperature, it normally takes around 20-30 minutes for this to happen. This is considered too long for our purposes. While insulation has reduced the difficulties caused by the sensor module's different thermal conductivity "permanently" reducing the skin temperature near the module, the additional thermal mass has not helped with stabilization times. As can be seen in the above articles, this has been approached in different ways - for instance, using a thermocouple and feedback-controlled heater adjusted for skin temperature to detect whenever temperature flux is zero (and thus whenever the sensor has the same temperature as the skin). This has a faster response since less heat will travel out of the skin, creating a lower discrepancy that needs stabilizing - however, the energy and power requirements of the heater are outside the scope of the battery and will quickly drain it, making this unsuitable for portable use.

Would anyone know of any way to get a good response time (<10 minutes at the very least... 2-5 minutes as a rough estimate perhaps) for such a device while keeping within clinical significance parameters of <±0.5°C - either by active (but low power) or passive (perhaps by superior thermal insulation) means?


2 Answers 2


I would consider a thermistor or platinum RTD in contact with either the skin or something very thin between the skin and the sensor, the whole thing covered by an insulator (there are some very high performance foam-like insulators) and an adhesive patch over the lot. The wires to the sensor must be thin. Optimum accuracy might be from using something like AWG 36 wires (2 or 4) that are of low thermal conductivity, such as manganin or phosphor bronze, but using copper and "heat sinking" enough length to the skin is probably enough.

Key points- use a high performance sensor such as a Pt1000 thin film type, minimize heat loss out the leads, and maximize thermal insulation overtop of the sensor by using high-tech insulation materials.


Is the sensor to be used as a probe (making single measurements) or is it to be worn 'continuously' and monitored? In the latter case its response time should matter less. With a small enough thermal mass (of the active element) and sufficient thermal insulation from the environment, the initial response and the lag time ought to be able to be brought into a suitable range. I'm imagining something like a tiny thermocouple bead overlaid with, f/ex, a centimeter or two of closed cell foam (not necessarily those technologies; that's just an off the cuff first concept).

  • \$\begingroup\$ It's worn continuously, but apparently a better response time is still a requirement. \$\endgroup\$ Commented Jun 20, 2014 at 8:48
  • \$\begingroup\$ You mentioned 10 minutes - does the current technique really take longer than that to settle? How is it being done now? \$\endgroup\$
    – JRobert
    Commented Jun 20, 2014 at 12:48
  • \$\begingroup\$ Yes, it takes about half an hour to settle. \$\endgroup\$ Commented Jun 20, 2014 at 14:56

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.