> John Payson wrote:
> >
> > > Tom Sgouros wrote:
> > > > Does anyone know of an easy way to control the voltage on a 120VAC line?
 I
>
> >
> > While I acknowlege that your approach would probably work, its output would
>
> > & more stuff..
>
> Actually, no. There is a considerable amount of thermal inertia in even
> small incandescents, and the practical lower illumination limits are at
> high enough numbers that the flicker is not any more noticible than in
> your monitor or conventionally dimmed lights. The trick is to evenly
> distribute the half cycles while maintaining zero DC offset and also do
> this smoothly in the presence of changing control values. In fact, we
> designed this as a heater controller but ran light bulbs during debug
> (old trick); the result was surprising to both of us. It ended up being
> used elsewhere in a light controller.

Tom suggested in his post that the wear on the lights from an "ordinary"
dimmer was producing excessive wear (since I don't know the thermal mass/
emissivity (is that the word?) of the lights he's using.  I don't know how
much effective "intertia" they have, but was assuming that if it was a
problem at 120Hz it would also be a problem at 1Hz.

> Of course, higher frequency or DC control will do a better job, but at
> the cost of complexity and (perhaps) noise. Our design did a really nice
> job of controlling higher wattage floodlamps (for us higher wattage: 300
> - 500) in a way that would have worked well on stage. In fact, I assumed
> a similar scheme was being used commercially in some cases, because I
> have seen a similar flicker signature in watching stage lights dimmed to
> dark. You don't notice the flicker if you don't look at the lamps, and
> you don't see it at all in high end controls.

Well, it probably would be easier to keep the RF noise in check if you
limitted switching times (both on and off) to the zero-crossings.  Or if,
as someone suggested, you use an IGBiT and let lights have the "early"
part of the cycle (in which case your turn-on is clean, and your turn-off
can be cleaned up a little with a capacitor if you drive the lights with
full-wave rectified unfiltered DC).

By the way, why the concern about maintaining zero DC offset?  I can think
of a few ways to do this (some very easy) but am not quite clear of the
purpose.  The simplest way I can think of would be something like this:

Ph0:    ds      1
Ph1:    ds      1
Val:    ds      1

Init:
        clrf    Ph0
        movlw   $80
        movwf   Ph1
        retlw   0

HalfCycle:
        movf    Val,w
        addwf   Ph0

        btfss   C
         bcf    Output
        btfsc   C
         bsf    Output
                        ;       Ph0     Ph1     W
        movf    Ph0,w   ;       Ph0     Ph1     Ph0
        xorwf   Ph1,w   ;       Ph0     Ph1     0^1
        xorwf   Ph1     ;       Ph0     Ph0     0^1
        xorwf   Ph0     ;       Ph1     Ph0     0^1

        retlw   0