I'm working out the resolution of my sensors, to see what changes I will be able to detect. With my thermistor in a simple 50/50 voltage divider, the values given in the data sheet for the thermistor, and 5V source, I'm looking at 4mV change per degree C. Given the onboard ADC is 10 bit, that works out to about 4mV resolution. That's about 1°C resolution.

That doesn't sound good enough to do what I want, granted I have no idea what temperature change will be associated with a particular air flow. I also don't know how hot I will be able to get the thermistor. This is important since the higher above ambient the device is, the easier it will be to detect increased heat dissipation. I also am planning to use the micro to control the voltage to set the average temperature of the array. So I don't know the voltage across the sensor either. Also since this is used outdoors, I have to assume the voltage required to keep a certain level above ambient will vary.

It's doesn't sound like I know a lot, which presents a challenge in designing a circuit that will work across that range, or ah least be fixed by swapping a few resistors or in software. I do plan on at least leaving a footprint on the sensor board for a higher resolution ADC. Speed isn't an issue, so hopefully a reasonable resolution and cost effective ADC could be found. Also, another option is to use a op amp to provide high gain on the output and adjust the signal to use more of the ADCs range. Some quick math shows a gain of 8 could be used, of we wanted to use the thermistors full range. That would give 1/8 degree resolution. Higher gain could be used of we refine the temp range we will use it at.

I'll put down some of these calculations later, to hard on mobile. (Lunchbreak)