Essentially I am shifting my voltages up with the diodes to eliminate the negative voltage sensing issue, right?
Do you think I could get away with using only one diode ?
As I am only going to be sensing ~-20 degC to +50 degC (-0.2V to 0.5V), adding one diode should allow for all positive readings (-0.2 + 0.6 = 0.4V margin) while still keeping my upper voltage (0.5 + 0.6 = 1.1V) below my Vref+ target of 2.048V. I would then set Vref- to MCU GND to keep things nice and simple.
I'm not shure about it. The output of the sensor is also pulled down with a resistor, maybe there is something inside,
what needs this room.
To find the forvard voltage: You may want to buy such a diode, and measure the forward voltage at the sensor's supply
current. I think diode selection is not critical, any silicon PN junction diode will do, but use that what You measure.
The 0.6 V was a common approximation for common silicium diodes around 1 mA current around room temperature.
If You like simulation, You may want to simulate just the diode with Spice or Pspice or similar.
If You like calculation: the diode If curent and Uf forward voltage for a given diode is approx:
If=I0*(exp(Uf/Ut) - 1). If you know a given If - Uf pair, then You may calculate Uf for another point.
The Ut is approx. 26 mV for room temperature ( 25C ) and is 1:1 related to absolute temperature ( in Kelvin ): Ut2=Ut1*(T2/T1)
The philips 1N914 datasheet has graphs about Uf/If at 25C.
I think the simplest and best way is to measure :-)
the max fwd voltage is 1.0V
O.K. but in this case you don't need this parameter.
If I wouldn't need 150C sensing just 50C then my choice would be leaving the 2 diodes in the cicruit, and the 100C drop in sensor output voltage would bring down the upper limit below Your 2.048 V, and leaving 2 x diode forward voltage room to the sensor
in the case it needs it ( I don't know what's inside ).
post edited by iabel - 2008/02/28 17:33:29