Andres,

A half wave rectifier will drop .7 volts. A full wave bridge will
drop 1.4 volts. I recommend that you rectify the voltage *before*
reducing it with  a resistor divider circuit. That way you will
lose .7v out of 120 volts instead of .7 volts out of 5 volts!

As to filtering using a capacitor, that is useful *if* you can tolerate
a response time of a fraction of a second or so. (You would *not*
want to alter the response time in any way if you are specifically
looking for voltage droputs, line glitches, etc.)

I take it you just want to monitor the line voltage and give a
readout that tells the approximate RMS voltage. As long as the
waveform is a sine wave, you can actually monitor the PEAK DC
and convert this directly (via your look-up table) to the
corresponding RMS value.

120 VAC >---------->|-*----|
(Hot)                 |    |
                      |    R2 150K
                    + |    |  25K pot
               1 ufd ===   POT------>
             250 vdc  |    |
                      |    R1 5.00K
                      |    |
120 VAC >-------------*----*------>

Let's assume you want to be able to read voltages from 0 to 127.5 volts
AC RMS. (255/2=127.5)

127.5*1.4142135=180.3122213 PEAK
Subtract one diode drop to get the actual peak output
180.3122213-.7=179.6122213  or 179.61 is accurate enough.

If we want the max output to the PIC A/D input to be 5.00 volts
when there is 179.61 peak dc volts, then we need two resistors
that have the ratio (179.61-5.00):5.00

Since it is difficult to get resistors having the ratio 172.61:5.00
it makes sense to form the R2 resistance from TWO resistors, one of which
is variable. I would use a 150K resistor in series with a 25K
multi-turn potentiometer for R2, and a 5.00K resistor for R1.
I would take the output from the wiper of the pot.

 [When calibrating the unit later, apply a known AC RMS value
 and adjust the pot for the closest displayed value. Note that
 in this case the increments are in .5 volt steps, so don't
 expect greater than +/- .5 volt accuracy!]

Scaling is very straight-forward: 127.5 V AC RMS in will yield a count
of 255 from the A/D converter. This corresponds to .5 volts per count.
In this case a lookup table can be dispensed with, since you can
arrive at the correct value by shifting the count to the right,
and masking the MSB. This will divide the value by two. (If the
original LSB is a "1", then the decimal point value is ".5"
otherwise it is ".0")

 [If you decide to use a lookup table, you can then correct for the
 .7 volt diode drop within the table. In this case I would recommend
 coding the decimal point portion in the MSB immediately. That
 will save you the shift operation later.]

The algorithm for the RMS input value corrected for .7 volt diode
drop is:
InputRMS=(MeasuredPeak+.7)/1.4142135  Again, you can code the
portion after the decimal point into the MSB or LSB as desired.

**************
There is another kind of meter that is also quite useful, and that
is the expanded range meter. Let's say you are really interested
in the range from 100 to 125.5     That is a VARIATION of
25.5 volts and we can use 255 counts to represent 25.5 with
an accuracy of .1 volts.

This type of expanded range meter requires that you have some
method of "removing" the initial 100 volts. There are a number of
ways this can be done using op amps and "scaled down" voltages
so that the output will go from 0 to 5.00 volts for an
input change from 100 to 125.5 volts. (by the way, I choose
voltages that scale well to the value 255 on purpose, so that
we can get a scaling factor that maps well to the inherent
0-255 count of the 8 bit A/D. Other scale factors can be chosen,
but they are not as clean and simple)

There is also a simple method using a Zener diode, but it is
limited to the accuracy of the Zener, and it has a certain amount
of sloppiness in the output depending on the sharpness of the
Zener at the chosen current level.

Regardless of which method is used to remove the 100 volts,
the hardware/software simply adds it back in by displaying a
leading "1" in the hundreds position. This digit can be
blinked when there is an under or over voltage condition.

I have to get something to eat, so I will terminate this post
now, but if there is interest in the expanded range method,
I could write some more about that later. Of course, if there
is no interest, then I will just save myself the trouble
and do something else with my time.

Fr. Tom McGahee