Invasive High Frequency Mains ADC Signal Conditioning

[madteckhead]

Hey Analog Designers and Artists,
Recent graduate new to the field and first time posting so be gentle. I've attempted to do my research before posting so please be kind

The problem: I'm attempting to sample mains power line with a 5MSPS ADC (AD7356). I'm unsure of how to convert the bipolar signal to a unipolar signal for use with with the ADC.

So far I've got a resistive divider on both the live and neutral to drop the voltages down, I was planning to drive the AD8275 (G = 0.2, Level Translation, 16-Bit ADC Driver). Which claims to have an bipolar input of +/- 10V of a 5V (unipolar) supply.

I've created an artificial ground on the isolated circuit at the bottom of the live/neutral resistive dividers. I've got 5V (single side supply) from an isolated DCDC converter which I want to run the signal conditioning and ADC off. Is it possible to convert these artificial ground referenced bipolar differential inputs (from mains) to a unipolar signal suitable for the ADC (0 to 2.5V). In addition do I need a buffer before the AD8275 in due to the high impedance of the resistive divider? The AD8275 has a -Vs pin... which is confusing me, as it is is claimed to have an bipolar input of +/- 10V of a 5V supply (unipolar). Does this mean I can get away with having the single rail 5V DCDC isolated supply?

Is there a smarter way to do this... it starting to get complicated if I need all the below.

1. Resistive dividers
2. Low pass filter
3. Buffer
4. Differential driver (AD8275)
5. ADC (AD7356)
6. Isolation
7. DCDC converter for isolated power.
8. Power supply.

I've spent countless nights trying to find a way to achieve this while keeping the complexity down. I'd greatly appreciate your input as I'm just starting to confuse myself.

Many many thanks,

Cheers,

Nathan

Future Design Challenges I'd appreciate input on:
- Low pass filter placement without distorting signal phase.

 

[madteckhead]

This is a quick sketch of what I was thinking...(http ://madteckhead.com/tmp/Screen shot 2010-10-18 at 1.08.16 PM.png) but ideally I could do it without the 5V DC offset. However am I correct in assuming that I need it to bias the signal due to the 120deg phase difference between the active and neutral lines. In addition, can I just represent the active and neutral lines as one 240V phase? As I understand it, with the two phases there will be a portion of the difference that is negative, thus requiring the bias VDC. however if I can represent the two AC sources as one 240VAC source, then the lowest potential is always 0.

Please help me out of my confusion! Many thanks. Nathan