Famous Quotations from Patz R.  :

> I am currently designing a temperature sensor using a 16C54 and the
> AN512. This involves charging a capacitor via two resistors
> consecutively and then using the measured time to calculate one of
> them. As mentioned in the application note I am using the RTCC pin
> for triggering. For testing I use a fixed resistor instead of the
> temperature sensor. So I should get always the same result. But this
> is unfourtunately not the case. The result differs by a few counts.
> I don't why that happens. Could it by caused by noise?
>
> Does anybody have experience with this method?

I've used it on several projects.  I had a lot of problems with drift
and noise.  I had one project that wouild hold a temperature setpoint
then slowly drift off by 25 C per hour.   One project ended up
hinging on a good enough averaging and noise immunity algorithm - any
ONE measurement was completely unreliable.

Since then I've tried to stay away from this method whenever I have
the budget.  The advantage of the resistor capacitor method is it is
cheap.  It uses a standard microprocessor without analog inputs, and
can measure temperature with a thermistor which is also a cheap part.
 If you need any kind of accuracy at all, use an analog input and a
thermistor or better yet one of those nice Dallas temp sensors.
Don't rely on this method if you are designing anything but a toaster
oven.
Best Regards,

Lawrence Lile