Thermistors: A worked example

The following example summarizes the procedure for a SPice10200.

A SPice10200 is being used in conjunction with an MMi99 to control a wine fermenter. The ideal fermentation temperature is 33°C. Available thermistors are 2K, 5K, 10K and 20K, all with Beta=3920. Find the best possible design to get best resolution at 33°C.

The SPice10200 gives a choice of Rfeed of 100K and 9.09K. All the available thermistors will be less than 20K at the temperature of interest because it is above 25°C and the thermistor resistance drops with heat. Remember, best resolution occurs about where Rfeed and the thermistors resistances are equal. Hence we intuitively expect a 20K thermistor and 9.09K Vfeed to be best.

Set up the spreadsheet with all the preliminaries as shown below.

The last value, Vfeed, is the result of clicking [FindBestVFeed]

The Tmin value has been selected so room temperatures will give a sensible reading, when the vessel is not in use. What effect does that have on our resolution where it counts, at 33°C?

With the above settings we get a resolution at 33°C of 0.305°C.

If we change Tmin to 28°C, and click [FindBestVFeed] again, we get 0.26°C resolution at 33°C. If we shoot for Tmin=0 we get about 0.35°C resolution. In this case Tmin is clearly not making very much difference.

If we try the 10K thermistor we get a slightly better resolution of 0.3°C at 33°C with Tmin=0°C. The 5K thermistor gives a very similar result, but with better resolution down at 0°C and worse above 33°C. Overall the 10K unit is the best choice, so our intuition was a bit off target!

Now here comes the surprise: The 2K thermistor gives 0.2°C resolution at 33°C, though to get there we have to lift Tmin to 16°C.