Thermistor circuit

asjad · Posted: Sun Apr 04, 2004 3:37 am

Dear All,

I am towards the end of my project and I need help with a Thermistor circuit.

Which one of these configurations wiould be best and why??

A) A DC supply on a Thermistor and a series resistor, insert the voltage across the Thermistor into the PICS ADC

B) Insert the Thermistor in a Wheastone bridge and put the output in
a Differential op-amp, finally to the PICS ADC

I value your opinion/advice!
_________________
Best Regards

ritchie · Joined: 13 Sep 2003 Posts: 87

Neutone · Joined: 08 Sep 2003 Posts: 839 Location: Houston

Using a voltage divider is ratiometric to the supply voltage. Assuming you will get 10 bit readings that are accurate. (you probably will) Use excell to plot the voltage versus temperature curve. Then look at the range you will operate in and determine if the resolution is enough. The series resistor should equal the thermistor at the nominal operating temperature. You can do this all on paper and make a good decision. You should do this before looking at more expensive methods. If you have more than one channel to get readings from drop a 0.1uF ceramic across each analog input pin to rapidly switch channels.

John P · Joined: 17 Sep 2003 Posts: 331

Do what's simple first, meaning Plan A unless you find something's wrong with it.

It's true that the thermistor + resistor approach will give an output that's ratiometric versus the power supply, but if you're using the PIC's internal reference for the A-D, that's ratiometric too, and hence no problem there. If one were fixed and the other not, you might have problems.

What you do need to do is think about the maximum and minimum temperatures, and what the variation in A-D input will be based on that, and what your resolution will be once the conversion is done. Is there going to be enough voltage swing to give you what you need? It won't be 0 to 5v! Also, it's not likely that the A-D output will be linear with respect to temperature change (check manufacturer's specs for thermistor, create spreadsheet). Is this a problem, and if so, how simple is it to fix (lookup table, mathematical function, piecewise-linear)?

Couch · Posted: Tue Apr 06, 2004 11:15 am

If you continuously apply current to a thermister, it will 'self heat', this means that the current flowing through the thermister will create heat, resulting in a temperature reading higher than what you would be expecting. You could probably run some tests to check the effect of self heating.

One way to overcome this would be to 'enable' the voltage supply only when you are about to measure, by supplying voltage through of the digital output pins - be sure your worst case temperatures dont exceed the current capacity of the output pin. One of the guys also mentioned that supply voltage must be known - it would be important to measure this as well.

dyeatman · Joined: 06 Sep 2003 Posts: 1938 Location: Norman, OK

When in this situation I create a small regulated circuit off of the suppply using a resistor,zener diode and 10MFD cap.

Example:
If I am using a main 12VDC supply I use the zener circuit to create an 6 or 8VDC reference supply for voltage sensitive circuits. If using a 5VDC main supply I create a 4VDC reference supply. This isolates the voltage sensitive components (like thermistor and sensitive A/D circuits etc.) from main supply fluctuations

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

Look at Microchip's App note AN512. If you are only reading to 10 bits or so you probably don't need an A/D at all, just a couple of port pins, a cap, and a resistor. It solves the self heating problem too. It is informative reading and applies to any microcontroller (not just the 16C5X chips) even if you are committed to using an A/D for this project. The code is in C, but it is simlpe and the flowcharts and good in any language.
App note AN513 is good too.
_________________
The search for better is endless. Instead simply find very good and get the job done.