Howard Winter wrote:
> >some really good stuff!>
> > Make sure unused inputs are properly terminated (not
> floating).
>
> Is it best to have unused I/O pins set to input (and
> terminated as you say) or as unconnected outputs?

Depends on the device. Some pins are input only, so they must
be terminated. Most PIC pins default to input on reset, so
terminating them saves you having to worry about programming
them to outputs. HOWEVER, if your program screws up and sets
them to output, driving against the rails, your power consumption
will go through the roof. Remember the very nice "Kill battery"
LED some low power systems used. If you have them as floating
outputs, and your program screws up turning them into floating inputs,
you again have high current problem. I prefer to terminate each
input separately with 1M since that is a good compromise between
drawing current if miss programmed, and noise pickup (for the
kind of systems I design).

> Is there any power-advantage to ceramic resonators vs.
> crystals?  I presume either uses less power than an
> external crystal oscillator - is this right?

Yes, of course. And less power than a crystal since you can
often get away with LP or XT mode instead of HS. They also
start up faster so wake from sleep is more efficient.

> What about decoupling/smoothing capacitors - should they
> be minimised to reduce leakage, or is this a Bad Thing?

It's a trade off, depending on just how low you need to go.
Good caps don't have a leakage problem. Avoid tantalums since
their leakage is quite variable with temp. If you must use them,
make sure they are very overrated (working V) to minimize failures.

I'm sure Dwayne R has other points to add.

Robert

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