> Does it have to be high-side? Low-side measurement is easier...

I guess it doesn't have to be high-side. Resistor-across-the-ADC has
been mentioned on the list before, detailed here

http://celesco.com/faq/420.html

A basic 250R is going to produce 1V -  5V with 4-20mA, which is
less than A2D FSD, OK for some measurements, eg ambient
temperature but maybe not others. I have a 5m depth gauge for example
that is pretty good to the mm with Maxim circuit (1.000m = 1.000V),
although of course a 10-bit PIC degrades this to 5mV steps

Maybe 300R to make 1V - 6V and then level-shift (doesn't account
for broken connection though). Does imply 0.1% or better resistors.
And what's that going to cost, plus the attendant temperature
management
 
> What is the fiddling and twiddling you had to do with opamp
> designs? After all, in your application, the MAX472 is little more
> than an opamp

Well, integral designs like the RCV420/INA168/MAX4137 are
factory-trimmed for accuracy. Internally they're still basically op-
amps. I could get reasonable results with discrete circuits, based
on their principles, but the tolerances needed are a little hard to
find OTS

A 5-pin SOT23 is such an easy answer, especially if it's cheap

> (It has> that absolute value output feature, but in a 4-20 mA
> current loop the current direction is known.)

That's true. There are quite a few subtle variations on the theme
(Rachmaninov would've been proud), some devices are unipolar,
some have polarity detection, voltage ranges, etc etc

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