Using LTC1966 with CT (current transformer) in an harvesting circuit.

So i work on a design which measures AC RMS current via CTs and performs energy harvesting in time multiplexing.

This is achieved with a rectification circuit and two burden resistors which can be disconnected during charging mode. 

In measurement mode the rectification circuit is still connected but does not seems to impact much due to the low impedance of the measurement path (16.4R). 

 
Until now i have used differential and then single end ADC to acquire the waveform and performed RMS calculation with digital signal processing, this works well but i face issue with noise, increasing with the input current, which impacts the measurements precision. 

So I want to use the LTC1966 to reduce noise and error in the RMS conversion. 
Here is a view of 3 channels.



CVCC is the power harvestin / supercap rail, MEAS_EN is the control line to switch between measurement and charging mode, RMS A, B and C are the AC signal outputs.

The acquisition circuit for channel A is shown : the MCU DAC is used to create the DC bias at the midpoint (0.55V) of the ADC range (1.1V), the AC waveform is sampled at the center of the 2*100K divider.

Here are my question and concerns regarding the integration of the LTC1966:

1) how should i connect the LTC1966 in this specific case?

• should i connect RMS_A to inputA and GND to inputB?
• should i connect RMS_A to inputA via a capacitor and a 1.65V DC bias (half VCC)  to inputB?
• another way?

2) will the LTC1966 be happy with a 3.3V VCC?

My only power rail is 3.3V, from an ultra low IQ DC buck, this system is designed for ultra low power operation and each uA counts, it charges a 3F 5.5V supercap and operates on its discharge (5.2V down to 3.3V), it would be inefficient and inconvenient (in term of  PCB realestate) to add a DC boost to create a 5V rail, i want to avoid that. A voltage doubler would not work, i think, since the supercap rail (CVCC) is constantly moving up and down between 3.3V and 5.2V (following the charge and discharge cycles).

Thanks