> ... and wire resistance.

Kelvin sensing. Annoying.

> The sensitivity (microvolts per degree) of this approach is not that much
> higher than that of Pt100 RTD or thermocouple

No calibration needed for sensible sensor choices.

Sensor cost trivially low.
Pt100 cost would make you think it used a precious metal ! :-)
Thermocouples this good are getting costlyish and need moderate
processing power for linearisation.

> and the complexity is not that
> low.

Certainly non zero.
ADC measurement is a nothing in modern environments.
Differential constant currents annoying but can probably be PWM out
and one op amp.
ie you will probably need an opamp for good precision but most other
requirement is on IC in modern micro controller.

> Since the output is proportional to absolute temperature, a 0.1% error
> represents several degrees C, but a single-point adjustment could be used=
..

And it's not temperature dependent :-)

> In a way, this idea is similar to the internal circuit in your garden-var=
iety
> band-gap voltage reference, except rather than using two closely matched
> (fabricated on the same monolithic substrate) BJTs operating at different
> currents, a single device is used with a switched current.

Yes. Although temperature needs to be factored into the bandgap device.

> It's a fine technique for certain high-volume niche applications (such as
> measuring CPU die temperature, but I'm not sure I see much in the way of
> general application areas for it given the limitations it has.

Seems potentially useful to me.
Once you have basic system going you can eg monitor N sensors with
little more effort and minimal delta cost. Switching issues to be
sorted.
..

R
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.