Hi Kristin,
Where are the "before cal" and "after cal" values from? I see they go down to two digits after the decimal point, while the register values are all whole number decimal values.
The value for LSB/V and LSB/A should stay fairly constant across the dynamic range of the part. For the RMS values, are you reading the RMS register at least 60 times in 1 second and then averaging over that 1 second? What is your dynamic range for the current?
With the voltage and current at zero, that value of 24 V seems high. IRMS is specified to have 0.5% error in a dynamic range of 500:1, while VRMS is specified to have 0.5% error in a dynamic range of 20:1. Although active energy has a 0.1% error over a dynamic range of 1000:1, and is more accurate than the RMS registers.
This chip is meant for 3-phase energy metering and, therefore for reading energy the current has a "No-load threshold" that can be set in order to assure no energy accumulation when the current is below a certain value. Then the voltage has a SAG detection in order to recognize when the voltage falls below a certain value. The RMS registers will never be exactly 0 with no voltage or current but this assures that those values won't be used in the energy. I think it should be fine to connect the same voltage to all three phases and then measure 3 seperate currents with CT's. Yes I would still do calibration on all 3 voltage channels.
You might also want to take a look at the ADE7816, this is a part with 1 voltage input and 6 current inputs. Although you may not be able to change your design at this point.
Regards,
David