Content-Type: text/plain
X-MIME-Autoconverted: from 8bit to quoted-printable by harrier.prod.itd.earthlink.net id NAA23521

I thought about something like Souto's circuit, but there is a problem.
The sine wave is small, lets say 8 to 12 Vac to a 5V power supply. It
means that the transistor will conduct when the sine wave is above 0.8
or 1.0 Volt, so it will exist a large gap meaning "zero crossing" and
not just a narrow pulse. 1V into a 10V sine wave means aprox 5.7¡ down
and up, or 11.4¡ per zero crossing, as it happens each 180¡, it means
6.33% of the time. It means the transistor will be off (collector up)
during 527.56µs from the 8.33ms period between zero crossings.   The
zero crossing circuit should get a pulse as close as possible from the
real zero volts at the AC line.  The above slack window will represent a
AC line window with a maximum of 10Vac (in a 120Vac line), what can not
be considered zero crossing. Of course that timing circuits of 263µs
could be implemented after the transistor collector goes up, but it will
always be ac voltage dependent, and never precise.  The other problem at
your circuit, it gets only one side of the sine wave, it would require
another diode at least.

Using the opto-coupler idea, driven directly by the ac line, it means
that the opto emitter would start to lite around 2 to 3 Volts, so it
would represent 3/117, or 1.46¡ x 2 = 2.92¡, or just 1.62% of the time. 
This ac line zero crossing slack would be exactly the necessary voltage
to lite the LED.  Remember that the LED would be attached to the ac line
via a resistor, and if this resistor value is high, the LED would not
lite when ac line is only 3V requiring a higher voltage, if the resistor
is small enough to do it, when the ac line reaches high value will fry
the LED.  Protecting the LED with zeners or something like that would
always require a low value resistor, so it will ask for a power resistor
(and zeners).

The big problem here, is that to sense the zero crossing we are trying
to measure "when it is not zero", what is wrong.

Now, if you double rectify a sinewave from the output of a power
transformer, you will have a double half sine wave, and exactly at the
zero crossing you will have a sharp edge, or... high frequency pulse. 
Using a simple RC filter it is possible to recognize exactly when the
double sinewave reverts state, so it is exactly zero crossing, or in
this circuit you can call it zero peak.

I don't have the minimum idea how to calculate what will be the peak
pulse width, so its equivalent frequency.... anybody?

Wagner
Attachment converted: wonderland:hfp.gif (GIFf/JVWR) (00018872)