John Dammeyer wrote:
>
> Walter,
>
> I have a customer that is insisting on 'double' floating point and a PIC(48
> or 64 bits).

64 bits! Check out this month's issue of Embedded Systems Programming.
Jack Crenshaw writes an article about floating point arithmetic
(specificially about multiplication) in which he notes:

"With a 64-bit counter, we can span the solar system to an
accuracy of one wavelength of light."

Perhaps this is the resolution your customer seeks.

Scott

--
"The problem with television is not the resolution."
                                 Hugh. F. Frohbach