On Tue, 13 Aug 2002, Scott Gordon wrote:

> Hi Brusque,
>
> I found exactly the same problem (programmer + PICALL software).
>
> I believe that it's a bug in Microchip's assembly-language `.inc' file for the
> 18F series.
>
> The include files are set up to construct a configuration byte value by
> `AND'-ing a series of options. This means that unimplemented bits (and bytes)
> end up set to `1'. However they read as `0' when the software goes to verify
> them, resulting in a discrepancy [ if you remove all unimplemented bits then
> FEFF and 2600 are the same thing ].
>
> I manually edited the configuration words in PICALLW after the loading the
> program, to mask out the unimplemented bits, which works.
>
> By the way, GPASM (Linux, etc) has the same problem.


I don't know whether the config bits are wrong are not, but I do know that
you have to assemble 18fxxx code with case sensitivity (i.e. in gpasm no
-c). Some of the config definitions are repeated with all lower case and
all upper case. If you assemble with case in-sensitivity then you get
"duplicate" define error.

Scott

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.