measuring temperature via NTC and op amp

Hello,

We are doing water temperature measurement and control for domestic showers.

A NTC in the water flow is used to measure temperature, and the temperature is measured thus by us:-

CIRCUIT DIAGRAM:
https://i56.tinypic.com/125qqs9.jpg


(Vcc = 5V)


..However, the above is not recommended by Texas Instruments....... they recommend using current sources, like this:-

https://focus.ti.com/lit/an/slyt156/slyt156.pdf

...please see figure 1 of first page.

Is it really worth it for us to change our method and use the current sources as in the texas application note?

(you can see that other than the current sources, our method is very similar to the Texas method)

they recommend using current sources

Click to expand...

Surely, it's a current source application note.

I have a different approach. Considering you are using an ADC with VCC as reference, implementing the sensor circuit as a simple voltage divider forms a ratiometric circuit, that has resistor accuracy (and ADC imperfection) as only error terms. As a disadadvantage, you get an additional non-linear term y= a/(a+x) respectively y=x/(a+x), depending on the sensor position in the circuit, in the signal chain. It has to be reversed to calculate the temperature. But a NTC is highly non-linear any way, needing a table or ln() calculation to get the temperature, so the additional term shouldn't be a problem. An ADC reading to temperature table could include the voltage divider.

If the ADC resolution is sufficient, I would also drop the OP circuit.