Not knowing what's on the EVAL board in terms of additional circuitry, I looked at the datasheet.
The AD7763 has "current sense" inputs, but these are voltage inputs that are meant to receive a voltage proportional to the current (or to di/dt). Therefore you need an external "current sensor" which will do the translation from current to voltage.
1) put a small series resistor into the mains wire and use the voltage drop acrosss the resistor as input to the AD.
advantage: very simple, low-cost, good accuracy
disadvantage: no potential separation, AD will be on high potential (dangerous)
2) put a current transformer into the mains wire. A current transformer with e.g. 1 primary winding and 1000 turns of secondary winding will give a 1:1000 current ratio. Thus 1 A of primary current forces 1mA of secondary current through the transfomer. Put a burden resistor across the secondary terminals of the transformer, then the current will generate a voltage drop across the resistor which is proportional to the secondary current or to 1/1000*primary current.
advantage: sav due to potential separation
disadvantage: costly, bulky, less accuracy, risk of transformer saturation in case of overcurrent, no measurement of DC components
3) use a hall sensor to convert the magnetic field from the primary current to a proportional voltage.
advantage: safe (potential separation), smaller than a transformer, can transfer DC components of the current, too, typically higher bandwidth than a transformer
disadvantage: costly
From the original specs I assume the device is to be operated at life potential of a 240 V mains. Therefore I suggest using one of methods 2) or 3) for safety reasons.
I also suggest using a transformer (or more transformers for more than 1 phase) for the voltage input(s) instead of a voltage divider. Thus the AD is completely isolated from the mains and you can safely handle it.
Furthermore there are two aspects to building a power meter.
1) hardware
2) software
If you are going to build both from scratch, you are likely o have a hard time getting both set up correctly simultaneously. CocaCola's suggestion of using the eval board is a fairly good compromise, although at a prize, I admit. By using the eval board you can concentrate on getting the SW part right. Once this works and you have developed an understanding of the AD7763, you can still build your own hardware - but now you have a working reference.
You will need to build the sensors (aka transfomer interfaces) for the eval board anyway and can reuse them for your later hardware project.
Harald
