4

Please forget the "why" for now, but I am attempting to determine the temperature of a 1N4148WS-E3 part by doing two measurements at different current setpoints. I've done some tests and some calculations and I don't understand where the error is coming from.

The theory

Derived from the Shockley diode equation, the following formula should give me the absolute temperature of the diode:

$$T=\frac{\Delta V_D}{\ln\left (\frac{1}{N}\right )}\cdot\frac{q}{n\cdot k}$$

Source: This video by Arda Yilmaz

Where:

  • ΔVD: the difference between two measured voltages over the diode
  • N: the ratio between the two measured currents corresponding to those voltages
  • q: Charge of an electron, 1.602176634 × 10-19 coulomb
  • k: Boltzmann constant, 1.380649 × 10-23
  • n: ideality factor of the diode, this is 2.60607 according to Vishay

The measurement

I've measured voltage and current over/through this diode using two setpoints. The measurement data I got is:

  • U1: 0.5819 V, I1: 582.8 µA
  • U2: 0.4886 V, I2: 78.3 µA

To get this data, I used two multimeters, a 1 kΩ resistor in series with the diode and a lab voltage supply to set the setpoint. The current measurement is in series with the resistor, while the voltage measurement is directly over the diode.

schematic

simulate this circuit – Schematic created using CircuitLab

Feeding this into the formula, ΔVD is 0.0933 V and N is (78.3/582.8 =) 0.134351

This results is T = 207 K which is obviously incorrect. I'm expecting something close to room temperature.

Does anybody have any idea where this error is coming from?

ocrdu
  • 8,705
  • 21
  • 30
  • 42
Graafvaag
  • 61
  • 4
  • Try taking a few more points and see if they track exponentially; and take a few at another temperature (as well calibrated as you can; can be tricky without a thermal chamber), see if you can solve for \$n\$. Depending on where you got the parameters from, the manufacturer may be prioritizing different parts of the curve, or have other parameters that modify the basic equation. – Tim Williams Mar 23 '23 at 14:25
  • 2
    I think all you have done is prove that the ideality factor is under 2. See this web page for your information: https://2n3904blog.com/1n4148-diode-forward-biased-i-v-curve/ See also this discussion of IF: https://electronics.stackexchange.com/questions/559667/how-to-determine-the-ideality-factor-n-of-a-diode-from-the-datasheet – Andy aka Mar 23 '23 at 14:35
  • I don't think you can use the same value of n for all 1N4148s ever produced. The fact that this value is quoted to 7 sig. figs. is a big red flag to me about that source, anyway. Even devices from the same batch will vary. A typical SPICE model uses 1.9, [this answer](https://electronics.stackexchange.com/questions/559667/how-to-determine-the-ideality-factor-n-of-a-diode-from-the-datasheet) calculates it to be 2.3 from Vishay's own data. Your best bet is to make some measurements and find n for your particular device. – Simon Fitch Mar 23 '23 at 14:43
  • Look at Linear Technology's application note 137 on how to do temperature measurement. A transistor is used for the sensor and \$ V_{be} \$ is measured at two different currents. Also beware of self heating of the sensor. – qrk Mar 23 '23 at 17:17
  • 2
    Did you cover the diode to prevent light from making it conduct like a photo-diode? – Audioguru Mar 23 '23 at 17:18
  • Diodes are much less ideal "diodes" than diode connected transistors. Diode connected transistors are commonly used as temperature sensors for fan controllers or temperature measurement for PC motherboards. For example https://www.analog.com/en/products/max6639.html. – Kevin White Mar 23 '23 at 23:18
  • TimWilliams I'll do that and post the results when I have them, good idea. Andyaka My first guess was 1.82 since that created usable results, but then I found the spice model. SimonFitch I'm specifically testing with the same model from the same mfr as where the data is from. Audioguru the SMD package is black and probably blocks all light. All, I tried to convince my customer that he should use a 2N3906 instead but he cannot be convinced and prefers a backwards compatible worse solution. – Graafvaag Mar 24 '23 at 08:53

1 Answers1

3

Ideality factor is a strong function of temperature (and varies a lot from unit-to-unit on top of that). I suggest you use a diode-connected transistor, which has an ideality factor very close to 1 (and is known to work well in this application).

enter image description here

A 10:1 ratio in measuring currents works well (yielding a change of about 200uV/K), but it's also possible to use three currents and eliminate the effect of the resistive component in the "diode" behavior.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842