>   I am designing a small digital voltmeter and current meter to build into a
> piece of equipment. I'm probably going to use a 16c73 or 16c74.
>   I would like to be able to measure 0 to 40 Volts and 0 to 20 Amps. In the
> lower range ( 0 to 25 Volts) I wanted .1 volt of resolution. If I feed the
> input voltage through a 5.1:1 voltage divider network and  into an A/D pin
> clamping the voltage at 5 volts I should get .1 volts per step ie. 255 =
> 25.5 volts, using Vdd = 5 volts as the reference.  My problem is what do I
> do to read above  25.5 volts with lower resolution? I was thinking about
> switching in a further 2:1 divider network with a FET when my A/D reads 255
> and scaling the result in software to reflect the 2:1 division.

Before you procede with this design, I would highly suggest that you consult
Microchip's page of errata regarding the 16C74.  If memory serves, the ADC is
not accurate for any input voltage greater than about 3 volts.  I would sug-
gest that you feed the input through both 10:1 and a 20:1 divide-down net-
works and use a VREF value of 2.5 volts.  This will then give you 0 to 25
volts with 0.1 volt resolution, and 0 to 50 volts with 0.2 volts resolution.
Since the A/D parts include multiple analog inputs, you could feed both in-
puts to the part all the time (if the input is at 50 volts, the low-scale
input will be at 5.0 volts; outside the useful range of the ADC but not so
high as to cause clamping or device damage).  One simple wiring approach would
be to use 5 resistors in series:

  47K, 33K, 100K, 10K, 10K.

The left side of the 47K should be tied to the voltage to be measured.
The node between the 100K and the first 10K should be tied to one of the
PIC's inputs, along with a 0.01uF cap to ground.  The node between the two
10K's should be tied to the other PIC input along with a second cap.  The
other side of the second 10K should be tied to ground.

>   For  measuring current  I am planning on using a .01 ohm shunt, ie. 20
> Amps would yield .2 Volts. I would scale Vdd down to .2 volts (25:1) and use
> that as Vref. For measuring voltage as described above  I would change the
> A/D settings and use Vdd as the reference voltage.

For current measurement, I think you will need to use an amplifier; if you
try to scale Vref down to 0.2 volts, your measurements will be nominated by
sampling noise and other nasty effects.  I would suggest that if you wire
three op-amps so as to produce an instrumentation amplifier with a 10:1
gain, your current measurements should be very accurate (you would connect
the two sides of the instrumentation amplifier to the two ends of the shunt.
By doing this, you cancel out any resistance between the shunt and ground.)

>    Can anyone suggest some way to simplify this project especially the
> voltage measurement scaling. I would like to do away with the FET if
> possible. Any comments or suggestions would be appreciated.

Well, my current-measurement idea is more complex than what you propose (but
should be much more accurate) but the voltage method should be simpler.  That
sound like a good compromise?