>> Less completely:

:-)

> Less completely, to resolve N values of a resistor chain, you have to
> have an accuracy for the chain of better than 1/2^N.

Yes. But note that that's "for the chain" and not "for any resistor"
(plus various other caveats) as I discussed.
That's typically a factor of 2 worse for the top resistor - which may
push you uner a tolerance value in some cases.

> For 2^6 values with 6 resistors, this gives you the same 0.26% that RM
> got (but the last schematic posted had N+1 resistors for N switches...)

Yes - I was adressing a solution using a constant current source as
input - which has some advantages, burt adds the accuracy of the
current source as a factor.

> Perhaps equally obvious, for a binary ladder, the error in the largest
> resistor has to be smaller than the smallest resistor value.

Smaller than HALF the smallest - and it's the sum of all errors, as
above - but the largest resistor predominates.

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