555 oscillator for thermistor instead of divider?

How about using a 555 oscillator with the thermistor as one timing resistor?

With a low enough capacitance, it would run at quite a high frequency and allow us to simply put a lowpass on the output to get the analog voltage, although counting cycles with the microcontroller would be far more accurate, especially at lower frequencies. Also, we wouldn't have any problems with very low resistances mucking up the readings.

I'll get a thermistor and breadboard something up, see how it looks.

Because the 555 will be running continuously rather than measuring occasionally you may run into problems with self heating of the thermistor.

The voltage across the resistor swings between VCC/3 and 2 x VCC/3 in some 555 circuits. At the high temperature end the 10K thermistor only has a resistance of about 15R. The maximum dissipation of the device I used was 18mW and I believe that 0.4mW is enough raise the temperature by 1 degree.

[hydraraptor.blogspot.com]

---

[www.hydraraptor.blogspot.com]

The 555 has a reset pin, and its frequency stabilises almost on the first cycle, so polling every so often should still work fine.

Like this:

R2 limits the self-heating of the thermistor, and since the duty cycle will (asymptotically) approach 50% as the thermistor's resistance drops, we get two measurements to use.

C1 can be adjusted to suit whichever frequencies work best for the code.

The duty cycle approaches 0 as R1 approaches 0. It asymptotically approaches 50% as R1 gets bigger.

Each cycle the energy in the cap is discharged into R1. The voltage on the cap goes from 2/3 VCC to 1/3 VCC so the energy lost is 0.5C(V1^2 - V2^2) i.e. 0.42uJ. The frequency approaches 14.4 KHz so the power is 6mW which will give 15C error. Not massive, but not ideal.

---

[www.hydraraptor.blogspot.com]