Thermistor, resistance/deg C

[Dragon Tamer]

I have some thermistors that I am using in my power supply. They are providing temp info to the thermal management of the power supply. Thermal management decides where the voltage regulators get their supply voltage from, decides when the fans need to be on, how fast the fans should be, and if the fans are working or not.

The way that thermal management works is that once it gets power, all fans should be off and the v-regs get their power directly from the bridge rectifier. Once the temp of either the battery charging transistor, 18V regulator transistor, or the v-regs exceed a certain temp.

The problem is that when power is first applied, the fan wants to be on even though nothing is hot enough, but when I adjust it so it doesn't come on when power is first applied, it doesn't come on at all. The speed of the fan also stays the same and doesn't change when the temp increase.

This leads me to believe that the thermistors are not sensitive enough for the circuit, so what I want to know is how do I find out what the resistance of the thermistor is at any given temp when its nominal resistance is 1k @ 27 C, and 10k @ 27 C?

The circuit has 3 1k thermistors in series connected to the 12V supply, and a single 1k thermistor in series with a 2k resistor to ground for the ambient temp. The op-amps that are buffer amps for the thermal management are referencing from in between the 2 sets of series resistors.

The second set of sensors uses 2 series circuits of a 10k thermistor and a 1k resistor to measure the difference between 2 heat sinks.

 

[MikeMl]

Google and Wiki is your friend

 

[Dragon Tamer]

So according to Wiki, if I have a 10k thermistor then it's resistance at 195 F (or 90.5 C) would be 905k ohms. Except that the resistance is supposed to decrease as the temp increases.

 

[dknguyen]

Well, if my part number are correct, you're not using thermistors. You're using RTDs.

If you're using an RTD, the resistance changes tend to be too small to be detected using a divider circuit. You need to use a wheatstone bridge. 1kohm are 37x larger than elements that are nominally 27ohms and would destroy the sensitivity of elements that already tend to need a wheatstone bridge. The resistances need to be closer to each other's values for better sensitivity (but then you run into the problem of self-heating which means you need higher resistance temp elements, or more in series if you want to get away with using the divider method).

To predict the resistance you really need to use the datasheet. Especially necessary for tHermistor curves since there are many different kinds and the response is non-linear (though can be linearized over a small temperature range)Platinum RTDs are more standardized and more linear.

 

[Dragon Tamer]

They are second hand thermistors and only have a color code. I am using op-amps as buffer amps so the hysteresis of one comparator does not affect the other, could I just use an offset voltage to compensate for the nominal resistance of all the parts?