Well, the low-tech solution is two ferrite toroids. Wind around 8 turns
of light-duty but mains-rated wire for the primary - this is in series
with the bulb, of course, and 16 or so turns of normal hookup wire as
the secondary. Assuming a 100W lightbulb you have around 0.5A per bulb
flowing which gives you plenty of current in the toroid. The secondary
voltage will probably be a volt or so - feed this to a comparator of
some sort - IC or a couple of transistors, depending on your preferences
and supply voltages etc, and there you are. Fully isolated and very
cheap. Also won't blow up if the light bulb shorts out.

> ----------
> From:         Reginald Neale[SMTP:neale@SERVTECH.COM]
> Reply To:     pic microcontroller discussion list
> Sent:         Tuesday, December 09, 1997 3:34 AM
> To:   PICLIST@MITVMA.MIT.EDU
> Subject:      Re: Interfacing to ac powerlines
>
> >Hi all!
> >
> >I need to do some simple interfacing to 220V powerlines.
> >A PIC16C84 must monitor eight different lights to determine if they
> >are switched on or not. All the switches are situated together, so if
> >I could find a way to "tap" the powerlines, and feed this to the PIC,
> >I don't hafta lay km's of wire to the lights themselves (this won't
> >be practical anywaty). I'm a little nervous about connecting mains to
> >the PIC. Also, I hafta keep the component count really low, mainly
> >for budget-, and space-related reasons.
> >
> >Anyone have some simple suggestions that could help me?
> >
> >Thanx!
> >Gaffs.
>
> It is possible to sense the voltage directly. Use a couple of
> high-value
> (several meg) resistors in series to a PIC input. Senses only the
> positive
> half-cycle, of course; the internal diode clips the negative one.
> However,
> there are safety issues here; doubtless you know about the potentially
> fatal danger of having your logic circuitry share a common with the
> power
> line. One safe way is to use a special optocoupler that HP makes for
> providing a logic output from a power-line input.
>
> Also, monitoring the switched voltage doesn't tell if you if a lamp
> has
> burned out. Can you steal a few milliamperes from the lamp current to
> feed
> the optocoupler? That would be more fail-safe.
>
> If the lamp current is high enough, it might be possible to use a
> clamp-on
> inductive sensor to sense lamp current and provide isolation at the
> same
> time, but that's probably a more expensive approach.
>
> Reg Neale
>