I am trying to measure the temperature of a battery pack using Zener diodes.

I want to use Energus's Tiny BMS 30A. In their battery modules they use Zener diodes to measure the temperature of the pack like this:

enter image description here

I want to copy their design as it is very elegant and simple, but I need a Zener diode which can match the voltage curve of the ones used in their design. Sadly they don't specify the diode.

I am having a hard time wrapping my head around voltage temperature coefficients. I can't find a graph showing the relation between temperature and voltage for Zener diodes I looked at.

I am.also a bit package constrained, to SOD-323, SOD-523, SOD-723. The supply voltage is +5VDC.

Help would be greatly appreciated.

  • \$\begingroup\$ I would just use a variable power supply, voltage standard or some other way of creating a stable variable voltage. That way you don't have to actually heat or cool the sensor to test your circuit. \$\endgroup\$ Feb 25, 2021 at 23:04
  • \$\begingroup\$ 40 years ago I learned that around 5 V Zeners had minimum temp coefficient. If that's still the case you'd want to avoid them. \$\endgroup\$
    – Transistor
    Feb 25, 2021 at 23:05
  • \$\begingroup\$ The minimum temp coeff occures on 5V6, 6V2 and 6V8 zener diodes. \$\endgroup\$ Feb 26, 2021 at 1:26
  • \$\begingroup\$ Unless you are relishing a science project, it would be much easier to use a dedicated temperature sensor of some sort. \$\endgroup\$
    – user57037
    Feb 26, 2021 at 5:32
  • 1
    \$\begingroup\$ If you don't need to use their sensor just use one you can use reliably, like an NTC or some integrated silicon thingy. They make them in all sizes \$\endgroup\$ Feb 26, 2021 at 7:35

3 Answers 3


The sensors are not actual "zener diodes", that's just a simplified way of thinking of the interface.

They are active circuits, that use up to four NTC thermistor sensors and allow the highest of the four temperatures to be read. For this to work there must be a reference and an amplifier for each sensor.

I suspect they are using a 3-terminal voltage reference per sensor for this purpose, and they are simply paralleled so the highest temperature one "wins". There would also need to be a reference resistor per sensor. Seems fairly innovative- maybe you should ask our Lithuanian friends if they don't mind you copying it?


2.3.5. Energus PS multipoint active temperature sensor Energus PS multipoint active temperature sensor available in the Energus PS Cell Modules has up to four NTC sensors inside. Innovative analogue signal OR’ing technique allows all 4 sensor signals to be read with two-wire acquisition system (one Tiny BMS temperature channel): output acts as a hot spot detector and reports only the maximum temperature.


To match all your specs, I found this LED which would perform very well at 1mA as a sensor and 20mA as an indicator.

https://www.digikey.ca/en/products/detail/everlight-electronics-co-ltd/EAST1608GA5/5142399 You want to use the low current range because the variation in Vf is due to bulk series resistance and not the temperature-sensitive Vth(T). It comes in a 1208m case. 1mA would be about 1.8V and could be calibrated at room temp.

  • \$\begingroup\$ Would incident light throw off the temperature sense function? Would it be necessary to black out the LED? \$\endgroup\$
    – user57037
    Feb 26, 2021 at 17:48
  • \$\begingroup\$ Incident light could affect it very slightly. Full sunlight might generate (photoelectrically) < 10 uA -- so a 1 % change in the bias current (from 1 mA). This would lead to a negligible change in VF (about 0.3 mV). Look at the data sheet -- VF has a very wide range. Likely each sample would need to be calibrated. \$\endgroup\$
    – jp314
    Feb 26, 2021 at 19:03
  • \$\begingroup\$ @jp314 Vf as in all small diodes has a wide range due to Rs bulk resistance above 50% of rated current but not <=5% \$\endgroup\$ Feb 26, 2021 at 19:09

Just use 3 standard diodes in series. At room temperature, each diode will be about 0.6 V, and the V will fall by about 2 mV per degree C rise. So 3 in series will be about 2.1 V at room temperature and fall by about 6 mV per degree C rise.

The advantage of this over NTC resistors is that the output is very insensitive to the value of the 680 Ω resistor. You can also put multiple sensors in parallel -- the output will tend to read the temperature of the hottest one.

1N914 or 1N4148 diodes are suitable.

  • \$\begingroup\$ I was thinking the same but exceeds the size requirement \$\endgroup\$ Feb 26, 2021 at 6:00
  • \$\begingroup\$ The slope of the voltage change is a function of the bias current. It will be 6mV / K at some bias current. At other bias currents the slope will be different. \$\endgroup\$
    – user57037
    Feb 26, 2021 at 17:50
  • \$\begingroup\$ OK, so just use 1 diode. Signal will be about 0.7 V and have a TC of about -2 mV/°C. Yes, slope is a slight function of bias current -- 2x the bias current will change the V on 1 diode by about 18 mV, and the tempo by about 10 %. This can be determined by a 2-point calibration and will be quite consistent over diode samples, especially from the same manufacturer. \$\endgroup\$
    – jp314
    Feb 26, 2021 at 19:00
  • \$\begingroup\$ @jp314 I agree. After cal it will work reasonably well. \$\endgroup\$
    – user57037
    Feb 27, 2021 at 5:37

Your Answer

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

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