> > -----Messaggio originale-----
> > Da: pic microcontroller discussion list
> > [mailto:PICLIST@MITVMA.MIT.EDU]Per conto di Byron A Jeff
> > Inviato: venerdl 15 gennaio 1999 22.23
> > A: PICLIST@MITVMA.MIT.EDU
> > Oggetto: Re: SUNSET SWITCH WITH 12C508
> >
> I have a similar project (to finish) to replace an old unsatisfactory
> sunset switch.
> >
> > I used almost the same technique for hysterisis. I sample the 555 65536
> > times, counting the number of light and dark intervals. If either light or
> > dark gets the majority of the samples (the other getting no more than 255)
>
> If you have the same number ,(a cloudy day ) how does it work ?

The default action is that if there isn't a clear cut winner is to do nothing.
Most cloudy days are light enough to trigger the lights off. However a deep
cloudy/stormy day will trigger the lights on, which is OK. There is the
possiblity of oscaillation at 2-3 minute intervals, but I haven't observed
it yet.

>
> > then the relay is triggered. Otherwise it's called a wash and nothing is
> > changed.
> >
> > It takes about 30 seconds for the circuit to register light and
> > about 2 minutes
> > for it to register dark. I threw in a 30 second or so delay after a relay
> > change to make sure that the circuit doesn't oscillate.
>
>  That is a good idea ! I had used the same time.
>
> >I solved this by paralleling a 33 Kohm resister with the LDR.
>
> Me too
> I also have the problem to limit the low level of LDR to keep the A/D
> conv. in acceptable range.

I couldn't get my LDR to register under 100 ohms even under direct 300 watt
halogen light. You could always series a resister to ensure that you have
enough resistance to get it to register.

BAJ