Thermistor Circuit Help

[Pheetuz]

Hi folks,

I am working on a university project and just trying to get some information about whether something that I want to do is viable.
Basically I want to measure accurately (to within 1 degree C) over a temperature range of 0-200C, is this possible with a thermistor? I have seen thermistors that have the necessary range but the output resistance changes are very small at the temperature extremes, I do think that with a 16bit A2D I should be able to pick up the differences, however I am worried that my accuracy will be limited by the other resistor in the potential divider changing temperature.
Does anyone have an experience doing this kind of thing that could offer me any advice?

Cheers,
Pheetuz

 

[sky_123]

Probably you want to use an RTD device, which has a defined resistance at different temperatures (mnfrs will publish tables for their devices).
Something like an AD7715 would be a suitable ADC, and there is an example circuit in the datasheet.

 

[Pheetuz]

Thanks Sky, I should probably point out that for reasons unbeknownst to me I have to use a Thermistor for the project!
Has anyone used a Wheatstone bridge for this and could offer any knowledge on how effective they are?

 

[sky_123]

A normal potential divider should work. I've never tried a thermistor to such a range, but I noticed that GE has a **broken link removed** that could help figure out the resistance delta per degree C at the extremes. Farnell have some GE thermistors (although the couple I looked at had the £15.95 US warehouse charge : (
You'd then have to store the table in memory to perform the translation to get the temperature.

 

[chuckey]

I think from a stability point a thermocouple device is the way to go. As their characteristics are well know, the calibration is easy. Thermistor tend to have wide tolerances on their value and law, so would need to be calibrated at 4 points over the wanted temperature range. With a change in resistance, the current through it must be accurately maintained (+- .1%?) over the instruments working temperature range, likewise the voltage measuring/linearising circuits must also be very stable.
Frank

 

[chuckey]

Had another thought about your use of thermistors, their characteristics are base on degrees K, so 1 degC at 473 deg K, is .2% accuracy. So every gain setting resistor must be trimmed to better then this figure. The thermocouple amp (https://www.analog.com/static/imported-files/application_notes/28080533AN106.pdf fig 31) is stuffed with .05% resistors, to get 1 deg C though that up to 1000 deg C.
If you have the kit , I would do an experiment, measure the thermistors resistance to better then .1% at some specified temperature (stabilized lab temp?). Set the thermistor up to dissipate its max power with a PSU, leave overnight, re-measure its resistance next morning, repeat for a week. This is to try and pre-condition the thermistor, to stop (or at least minimise) long term drift.
Frank