ashly Dearden wrote...

>I'm thinking of doing some temperature sensing using the approach that
>Dallas and Maxim has with the current source and measuring the voltage
>across a diode-connected transistor.  They don't go into alot of detail =
on
>what they are doing...other than around 100uA of current. I assume this =
must
>be a stable current source and just measure the voltage that is =
generated?

Chips like the MAX1805 are a bit more sophisticated than that; they're
actually pretty clever.

Simply taking a diode (or diode-connected transistor) and putting a
constant current through it will give you a temperature-dependent
voltage that you could measure; and its temperature coefficient will be
pretty predictable, at -2.1 mV per degree C; but different diodes and
transistors will give you very different voltage drops at any given
temperature, making it necessary to calibrate each individual sensor and
making interchanging of sensors very difficult.

What MAX1805-type devices do is modulate the current through the diode
or transistor, alternating between two levels of current that have a
fixed ratio to one another- 2:1, 3:1, 10:1, etc.  The chip measures the
diode voltage drop at each level and subtracts the two voltages,
measuring the difference.

It turns out that this difference in voltage at the two current levels
is a function of temperature and current ratio only, and is very nearly
independent of diode/transistor characteristics and the actual current
levels used.  The voltage difference is proportional to absolute
temperature.  Transistors/diodes can be interchanged without more than a
couple of degrees C error.

>I'd like to do this in just a PIC, since its there doing some other =
stuff as
>well.  The alternate is to use the MAX1805 and interface the I2C to the =
PIC.
>But one chip is better than two...

If you try to replicate the functionality of the MAX1805 with discrete
stuff, you'll very likely end up with way more hardware than if you just
used the MAX1805 itself.

Sometimes Maxim just can't be beat...

Dave

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu