# temperature measurement

Discussion in 'Microcontrollers, Programming and IoT' started by dcac, Aug 25, 2014.

1. ### dcac

56
1
Jul 10, 2013
Hi there, what way do you measure temperature using NTC thermistor? I have made a circuit that is a Wheatstone bridge, and the 10k thermistor is one of the bottom part. The bridge is balanced at 25 degrees Celsius and the ratio of the resistors are 1k/10k.
My first question is: do you use resistance to temperature table that is found in the data sheets or use rather use Beta or Steinhart-Hart Thermistor equation to find the temperatures? Is it piratical to use the equations? The only advantage of the equations I see is no map table needed so the memory on the micro controller is saved.

My second question is: how do you do ADC by using PIC? The only way for the moment I think is to measure the two points with respect to the ground (in very short time) and then subtract one from the other. Is there is a configuration on PIC so I would need just to plug two leads in the bridge and the conversion is done without subtracting the two values manually?

2. ### BobK

7,682
1,688
Jan 5, 2010
I think some of the 16-bit PICs have a differential mode for ADC.

But, that really is not necessary. Look at your Wheatstone bridge. One side is always going to read a fixed voltage, so what is the sense of measuring it? The bridge is good when you want a differential signal, but that is not the case here. I think you just need a single voltage divider and then read the midpoint. Or better yet, you put a constant current through the NTC and read the voltage across it.

Bob

KrisBlueNZ likes this.

5,164
1,087
Dec 18, 2013
I thought the bridge was used to linearize the NTC as their output is exponential. You can also form a voltage divider as Bob has said and place the NTC across one of the resistors. This will also help to linearize the output when the correct values have been chosen.
Thanks

4. ### Gryd3

4,098
875
Jun 25, 2014
Would an instrumentation opamp be a better choice than relying on a voltage divider with a sensor? (Depending on sensor range mind you...)
The output from the opamp would be fed into an ADC which would allow the measurement of very fine changes from the sensor depending on selected gain.
Very fine readings may require a 3-wire wheatstone setup to compensate for temperature.

Of course, this completely relies on the desired resolution and accuracy.

5,164
1,087
Dec 18, 2013
I dont see how an opamp is going to improve things. The limiting factor is the A to D is it not?

6. ### Gryd3

4,098
875
Jun 25, 2014
One side of the wheatstone will be fed to the + of the opamp, and the other side into the - side.
The output of the opamp will then provide you with 0V when both sides are equal, and if you are attempting to read small differences that regularly result in readings of less than 1V, you could make use of the gain on the opamp to scale the output to a 3.3V or 5V reading at your max desired reading.
The A/D would no longer need a different external reference voltage, or would no longer need to float somewhere between the highest and lowest reading which would increase the resolution of the readings using the same A/D.

Thoughts?