> Your argument is sound re the input and output impedances,
> although I do question the calc based on max sourcing and
> sinking as these are recommended max I, not actual max I.
> I'm sure the PIC pin will drive more than 20-25mA if
> driven into a short...

My thinking, as I mentioned, is that these values are based on the Rdson
with determines max specced output current. Mchip isn't about to give you a
lot more silicon than they need to to get the Vol and Voh values, therefore
Rdson can't be much smaller than the numbers I gave. If you really wanted to
be sure, you could measure Rdson across a range of temps and Vcc values.

> But it still raises the point, why are people so harsh
> re the overclocking argument but quite happy to discuss
> the "correct" method of driving a PIC pin into a dead
> short. You could probably come up with some good arguments
> why overclocking by 20% will probably perform great too,
> but I'm not going to rush to try either grey area.

No, overclocking is a very different thing. At a given clock speed,
temperature, Vcc, aging of chip, etc. a signal inside the chip is either on
time or not on time. If they test to a given limit I would _not_ want to go
beyond it (unless I was willing to test to my new limits myself). This
disregards the thermal issues in overclocking, which are probably pretty
small for modest overclocking rates.

> Out of spec is out of spec. Next we'll be discussing the
> best way to drive a 10amp load with a 3amp transistor,
> using time constants and such as our argument??? ;o)

This is often legitimately done. Some transistors are specced at other than
100% duty cycle just for this reason.

Bob Ammerman
RAm Systems
(contract development of high performance, high function, low-level
software)

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