Temperature sensors are used for a wide variety of industrial, scientific, medical and domestic purposes, and differ by their design and/or operating principles to better suit a particular application. Measuring temperature distribution requires use of many sensors, and their selection should balance cost, accuracy and convenience of networking. Most accurate are resistance temperature detectors (RTDs), of which most commonly used are Pt100 and Pt1000 platinum RTDs. These sensors are expensive, require a separate electronic driver for each RTD and complicated networking. Thermistors are the least expensive on their own but ensuring accurate measurement requires their individual calibration, provisions to shield them from noise and complicated networking. Semiconductor temperature sensors frequently integrate an appropriate electronic driver and provide a standardised interface, which eases of their networking, at a medium price point. However, their accuracy is usually specified rather conservatively by the manufacturers.

In this study we compared readings, collected every 19 seconds from Pt100 and DS18B20 temperature sensors, placed in a cardboard box close to each other in a typical dwelling, over several months’ period. For periods of time both sensors were placed inside the same plastic bag, or were wrapped around with packaging materials for thermal insulation, or were left exposed to the environment on their own.

The vast majority of collected data show that the DS18B20 provided substantially better accuracy than the one specified by the manufacturer (+-0.5°C) when the sensors were close to each other and well insulated. Their readings were well correlated over time even when the absolute readings were notably biased. These and some other past and present observations, to be presented at the conference, led us to the conclusion that DS18B20 sensors, after some suitable calibration, can reduce complexity and cost of the temperature distribution measurement networks.