I somehow managed to damage the temperature sensor in my thermostat while cleaning corroded battery contacts. Apparently the only fix is "buy a new thermostat", though it is a pity to replace this for a component worth a few pennies.

This is a "British Gas" branded RF-linked Drayton Digitstat thermostat.

The temperature sensor seems to be a two-pin ceramic device with blue body and white tip. As far as I can tell there are no other markings, and there is no standard colour scheme.

Picture of unknown temperature sensor in thermostat


  • before desoldering I checked the connectivity, and it seems to be connected to the traces
  • after desoldering I measured resistance, and room temperature I've got ~55kOhm. Holding it between fingers it goes to ~45kOhm
  • 3
    \$\begingroup\$ What makes you so confident that you have damaged the sensor? It just a kind of thermistor, it's not as delicate as you might think. \$\endgroup\$ Nov 26, 2021 at 10:40
  • \$\begingroup\$ My guess would rather be that the mentioned battery contacts aren't conducting any longer. Just measure resistance across them when connected? \$\endgroup\$
    – Lundin
    Nov 26, 2021 at 10:50
  • \$\begingroup\$ Switching it on shows "E1" which is apparently "internal sensor failure". If it goes past the error it shows a temperature of 40 degree Celsius, which is a bit warm for this part of the world, especially with the heating off. \$\endgroup\$
    – user1387
    Nov 26, 2021 at 10:57
  • \$\begingroup\$ It could be that there is no contact anymore, you should use some flux or solder to redo contacts, but the photo shows only the sensor and not the whole board damage. \$\endgroup\$ Nov 26, 2021 at 11:02
  • \$\begingroup\$ The photo is very blurry, but it seems that a track is discontinued, don't know it could be an artifact of the blurry image. \$\endgroup\$ Nov 26, 2021 at 11:06

2 Answers 2


If I was a betting man, I would guess it's a 10K\$\Omega\$ \$\beta = 3977\$ or so thermistor. But there is really no way to know from the appearance.

You could take it off and measure it, and refer to a R vs. T table to see if that's close. It should be 10K at 25°C if my guess is correct and about 4% lower or higher for every degree C the thermistor temperature is different from 25°C. Be sure to let the thermistor cool and avoid touching it if you are trying to make that measurement.

If you get the base value right, chances are if the \$\beta\$ is a bit different it won't make much difference in a substitution, since thermostats are usually used very near to 25°C.

Alternatively, if the thermistor is actually bad (in which case the measurement won't make much sense), then substitute fixed resistors for the thermistor and note the temperature reading for each resistor and compare with data sheets of likely potential substitutes. For example, if you substitute a 10K resistor and the display reads 25°C you have one point. If you then substitute a 5K (say 2x 10K in parallel) resistor and the reading is close to 41.5°C then my guess was correct.

Edit: Based on your measurement, maybe it's a working 50K @25°C thermistor.

Instead of looking at the sensor, I would take a close look at the power supply.

  • \$\begingroup\$ Don't you think that measuring a thermistor characteristics is possible only with a working thermistor? \$\endgroup\$ Nov 26, 2021 at 10:55
  • \$\begingroup\$ @MarkoBuršič If it's really damaged, yes. I suspect it's not. But see edit in case it is. \$\endgroup\$ Nov 26, 2021 at 11:02

It was a 50k @ 25 degree resistor. I connected a variable resistor to the solder points, then graphed the reported temperature for a few different values. The biggest problem was buying one! It was far below the minimum order for Digikey, Mouser, Farnell, etc, and eBay had a massive markup.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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