Please recommend solution for clinical thermometer

[bchen]

I am going to make a clinical thermometer, but after a lot of research on internet, I think using a thermistor is a good way to do. But one problem comes out, as you know the thermistor is not linear, it need a temp-resistence table to be used to look up. But how can I get this table? If I measure this by myself, it seems not accurate at all.

Or can I use a digital chip for that? Do you have some suggestions? The chip at least needs to have resolution of 0.2C. I know there is a lot of chips that can do it in 0.5C, but it's not accurate enough for clinical thermometer.

 

[karesz]

Hi,
Why dont you read the datasheet & contact the vendor pls?
You need the product specific data-especially at these high accuracy(!), can not be generell answered...
I think your circuit is as a "self calibrating one" to develop, means=you must have the sensor specific characteristics knowed in a table & make 2 or 3 points on the temp. scala to exact "calibrate"(& storage on chip/E2PROM) the individuell sensors.
K.

 

[ark5230]

One solution is just make a look up table selecting few temperatures in the range of interest. The line joining the points will smooth out minor errors and you can fit it to some line or equation and this will be fairly accurate and acceptable. This is a simple solution in the absense of any data sheet. Additionally responses of rest of the circuit components will also be taken care in this manner.