STeve Childress asked for more information. My attempt at ASCII art was
aborted. If you can read MACDRAW II documents then let me know I can send *real*
diagrams. I have tried to clarify the explanation from my previous posting and
add a bit. Steve, if you are still confused and not blessed with Apple Mac, then
email me your fax no and specific queries and I will try to help on paper.

Explanation:-

One lead carrying the load current passes through the transformer. Define this
as the primary. Either choose one of the existing windings as a secondary or
wind your own with insulated wire to scale the current by the turns ratio. The
OP amp will create a virtual earth (see classic op-amp theory) at the negative
input at a dc voltage offset of the bias put on to the positive input. This will
be about 2.5 volts. For extra precision and temperature tracking, this value can
also be sampled by the PIC ADC and subtracted from the value derived from the
op-amp output.

Monitor the op amp output with a scope and ensure that it does not limit over
the range of load currents required. Adjust feedback resistor and  turns ratio
if needed. Also ensure that the op amp used is suitable for operation on low
supply voltages and with an output swing close to both supply rails.

Kalle suggested fitting a resistor across the secondary of the transformer to
avoid near infinite votages being developed if the secondary is un-loaded.
Whilst this is theoretically possible, it rarely happens because the transformer
core saturates and limits the voltage. It would however be a good idea to fit
parallel reversed diodes across the secondary to protect the op amp under surge
conditions.

Someone commented about errors with non-sinusiodal currents. This needs to be
taken into account where appropriate. If you anticipate such loads there are two
choices.
1) If the current waveform remains a constant shape the an estimate of the form
factor can be made and used to scale the peak current measured into an RMS
equivalent figure.
2) The proper way to do it  is  to calculate the rms value. After all we have
got a processor and we can use the square root algorithm that has filled the
PIClist a week or so ago.  Remember your AC theory  - RMS means ROOT MEAN SQUARE
and is calculated as  the square root of the mean of the squares of as many
sucessive samples as possible. Many samples per cycle is best but you could be
processor time limited and so instead, sample with an interval  that is NOT
equal to factors of the mains period and average for as long as possible before
taking the square root.

Final comment: if the application is destined to measure *power* consumption
then the phase shift in reactive loads must be understood. This is really beyond
the scope of this list and risks boring everyone else. I recommend that the
application be talked over with a competant analogue engineer.
It could be worth reading the two part article in Everday Practical Electronics
February and March 1996 issues which features a power consumption meter based on
a 16C84. They can supply the PCB, software and a source of preprogrammed PICs if
required.

Regards to all
Safety First

Bob Minchin
bob.minchin@roke.co.uk
Romsey UK