I have a desire to do some really rough-scale temperature sensing with the
10 bit AD in my 16f873.  I am looking for something like +/- 1-3 deg. C
over the range 0-100C.  I think I can do a single point cal at the time of
manufacture.

Here's my initial idea : find a stock resistor with a really crappy temp
co. and measure a voltage across it.  At 300PPM/deg. C, that gives me a
delta of 3% over the temp range.  If I did some clever voltage referencing
I think I ought to be able to get at least 100 clicks in AD over that
range, I just can't think of how.  Keep in mind the operating word here is
cheap.  We want to temp sense stuff inside of our cases so that if there
are failures we can go back and look at what the reported temp was in the
box and see what is killing us.

Thermistors seem to be in the $1-$2 range, and I'd like to be in the
$0.10-$0.20 range.

You guys got any ideas?

Are crappy temp co. resistors linear, or at least monotonic at those temp
ranges?

Maybe I'd get a better temp variance with a capacitor and timing an RC
delay....

TIA

-E
Erik Reikes
Software Engineer
Xsilogy, Inc.

ereikes@xsilogy.com
ph : (858) 535-5113
fax : (858) 535-5163
cell : (858) 663-1206