> I have heared of someone accidently overclocking a 18F pic to
> seventy something megahertz by selecting the wrong oscilator
> mode (hspll instead of hs). apparently the thing ran at least to some
> extent

Could have been me, I've mentioned it before. I intended to use
PLL with a 9.8304MHz to get 39.3216MHz crystal but put an
18.432MHz in by mistake. AFAICT this 18F452 ran OK at
78.6432MHz and several tests seemed to proved that it was
executing instructions properly. As I've found with other trials
though, some modules may not be happy when over-clocked

Presumably it is still the case that Microchip sort micros by speed
group, not speed per se. So a -04 may run at 9.999MHz, but it
falls short of the 10MHz to put it in the -10 group. In that example
the PIC could be run at almost 10MHz and genuinely be within its
electrical and performances capabilities. It is in fact a -9.999 PIC.
However, you don't know its actual tested limit, and each batch is
likely to be different

IMVHO it's fair to say though that "conservative" over-clocking
(eg nudging a -04 to 4.096) might probably work reliably, as the
gaps between -04 / -10 and -10 / -20 are quite wide. And for
Microchip to absolutely guarantee -04 operation at 4MHz under
all spec conditions, you might assume that they allow a good
margin of error. Although only Microchip know what that margin
really is, and especially when costs get squeezed

Of course you have no complaint to take to Microchip if your
device fails to perform properly at greater than spec speed

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