I’ve got a few reasons for discouraging you from doing this:
It’s generally accepted that galvanic isolation is needed. This design presumably ties the incoming neutral to ground. As long as that holds true, high voltage exists only in the “Resistor Divider” box. If the neutral is disconnected, internally or externally, the line voltage suddenly appears in the “Filter” box, at the ADC, and then wherever it burns it’s way through. An electric meter is a sealed unit that is isolated from the world and is insulated to eliminate exposure to high voltage. The IoTaWatt (and other similar equipment) have other things like CTs and the USB power supply that can conduct any fault voltage outside to cause both fire and life dangers.
The above risks are equally present from simply not exercising strict polarity observance on the input. Many US, and the ubiquitous Euro two pin plug are not polarized. If you bring in two 230V voltages from two different phases, and they are not strictly polarized to match neutral to neutral, three out of the four combinations will result in a direct short of 230V or 400V into the ground.
This flat out can’t work with three-phase three-wire systems because there is no common neutral. It will always produce a 230V short.
The IoTaWatt is ETL certified as a low-voltage device. Even if this design were to be somehow made reasonably safe, the cost to maintain safety certification would go up significantly.
One of the primary virtues of IoTaWatt is that the design has made it adaptable to every power system encountered so far. One of the primary reasons is that the AC reference is normalized with VTs specific to the local voltage. This design would require that different models be available, or one model switchable.
All that said, I do use a similar direct AC line input as a reference when measuring the phase shift of a VT. I simply add a high value resistor to the hot lead and feed it into a standard voltage input. I avoid the neutral issues by placing the IoTaWatt on a rubber sheet, and powering it with a battery USB supply. The only external connection other than that line reference is the VT being measured. Oh, and I make myself stop and double check everything before I turn on the juice. I take all the measurements via WiFi.
BTW, the ADC reference voltage is 3.3V and the bias is half that or 1.65V - 1.65V peak-to-peak is about 1.16Vrms.