Initial Thermistor Testing

4 Apr 2011

Ho All! I'm Charlie and I imagine that I am a materials scientist - designing and manufacturing such things as rare earth magnets and medical devices. My task is to address some of the temperature measurement aspects of the BioBoard - we want something that is inexpensive, robust, sensitive enough for bio-systems in the appropriate range of interest (say -20 C to 150 C), and reasonably accurate. We considered a bunch of different choices - thermocouples (TC), resistance temperature devices (RTD), digital temperature sensors (DTS), and thermistors being some of them. My training says "Go with the thermocouple, it is well known, extremely robust, has a wide temperature range, etc.", but we got some reality checks once we started looking into them - while they indeed work over a very wide temperature range, it is much bigger than we need for biological systems. And what good is it if there is the additional cost of an amplifier circuit ($12 here http://www.sparkfun.com/products/307) to boost the millivolt signal, and also TCs are just a bit hard to find and/or make - great for professionals, but not students. We have been unable to initially choose between a DTS (like here http://www.hacktronics.com/Sensors/Digital-Temperature-Sensor-DS18B20/flypage.tpl.html for $4) and a thermistor (available from all kinds of places that deal in electronics components - Digikey, Newark, etc. for less than $1) so we are testing both! Rolf is managing the DTS efforts and I have the thermistor work on my plate. They come in all shapes and sizes, but unfortunately they are no longer as easily available in retail stores as they used to be - everything is mostly on the web - so I had to initially take what I could get locally, a big clunky sensor (a centimeter in diameter!) rated at 100 ohms at room temperature.

Besides the size, which will be determined by your application but small is usually preferred, the two features you will be looking for will be that room temperature resistance value (anywhere from 1 ohm to 1 mega-ohm) and whether the resistance increases with temperature (positive temperature coefficient, PTC) or decreases with temperature (negative temperature coefficient, NTC); as always, Wikipedia is a good first resource for finding out about such things (http://en.wikipedia.org/wiki/Thermistor). In any case, we'll be using the analog inputs to the Arduino boards, so we mainly need to determine what simple circuit we can use to measure the electrical resistance of our thermistor as the temperature is changing in our bio-system. We'll do it with a voltage divider, and it really is no different than using the Arduino to monitor a photoresistor or a potentiometer - here is my first simple circuit:

I've borrowed Rikke's Diecimila (just one of the Arduino models that we are exploring for our sensing platform) and while I was buying that 100 ohm thermistor I picked up a resistor with the same value to use in my voltage divider (big ones are no more expensive than small ones, so I bought the largest one they had) - that is the thermistor up at the top of the image. We're using a 5 VDC excitation and I had it up and running in a matter of minutes, just using the AnalogReadSerial sketch that is widely used as an example tutorial when beginning with the Arduino. I tested that everything was working OK by first dunking the sensor in ice water and then dropping it into boiling water - we'll talk later about calibrating, but this just told me all was fine. I am just using the serial monitor to view the streaming data (logging at 1 Hz for now - plenty fast enough for most bio-systems, and certainly faster than this high thermal mass version of a thermistor warrants) and next need to drill into the details of the Steinhart-Hart equation to convert the analog input to the board (in volts) to an accurate temperature. And I am waiting for my improved sensors - 10 kilo-ohms at room temperature with a very tiny sensor volume - to appear in the mail. Stay tuned and next time we'll talk about selecting the right thermistor for your application, mathematically fitting the resistance vs. temperature curve, distinguishing between precision and accuracy and resolution, etc.