> I'm curious _why_ you have such a large capacitor on your board. A "filter"
> capacitor is named such because it is configured as a low pass filter on the
> input voltage. This is done in an effort to keep high frequency components
> that are modulated on top of the input power from getting into the voltage
> regulator. However, in some sense the regulator itself is the ultimate
> low pass filter, passing only DC. The other purpose of the capacitor is
> to act as a reservoir of energy when the current demands on the regulator
> change suddenly and the regulator is connected to the primary source via
> a network that has a limited response time. By providing the extra current
> during the short period when the regulator starts demanding it and the
> mains can produce it, the capacitor keeps the regulator operating in its
> "normal" voltage range. Now, if all you have on the down wind side (the 5v
> side) of this regulator is a PIC and a few CMOS chips, even the _worst_
> case scenario could be handled by a much smaller capacitor. Easily 1uF and
> probably .47uF.

OK.  I, however, am driving (on some of my devices), things like triacs and
LEDs.  I just picked the largest that I could fit in my tiny case to be
honest with you.  Maybe I should use a smaller cap.  If I did, then the
voltage would run down fast and I wouldn't have a problem.
>
> There are switching regulators with an on/off output. The other thing you
> can do is to hook the Dallas part to a CMOS switch (like a 4066 or something
> similar (and up to date :-)). Connect the PIC circuit in series with the
> switch and have the Dallas part control the switch enable line. When the
> reset delay has passed the Dallas part goes "on" and this turns on the CMOS
> switch and that turns on the PIC circuit. You should probably be sure and
> limit the current through the switch to under 100mA but there are other ways
> to do this (with a MOSFET for example) that will work as well. In building
> robots I find that good power management requires that I be able to turn
> power _off_ to portions of the robot that are not being used (like the speech
> synthesizer) to conserve batteries. I've started using Maxim Switching
> regulators which are more efficient than 78xx linears and they have a logic
> level "on/off" input. If you call Maxim ask about their evaluation board.

Good idea but I was hoping this conversation might produce a simple circuit
that I hadn't thought of.  I cannot possibly put even an 8-pin chip on my
board as it is.  The best I could probably do is a 3-pin device like the
DS1231.  Since my post, many people have said that I DON'T have to start the
PIC at 0V so I think that is my best bet.
>
> All in all, probably the best way to go.

Yep, that's seems to be the consensus here.  I'll try it out.

Thanks.

--

            -- Paul Greenwood --  (pablo@austin.ibm.com)

All true wisdom is found on T-shirts.