Thanks for the information, Robert. It allows a more detailed look at the numbers - if I can get my sums right!
From the schematic, Vcc for the atmega328 is +5V. The burden is 150 ohms.
analogRead() uses Vcc as an ADC reference so a 1023 count is equivalent to 5V, so 1 count ~ 4.89mV.
The ‘zero’ reference for the counts is 511, so the actual voltage is (count-511)*4.89mV.
The turns ratio of the CT is 1 (primary) to 2000 (secondary).
So for a 1kW load at a nominal 240V AC the peak CT o/p voltage would be
sqrt(2)*(1000/240)*150/2000 = +/-442 mV pk
From the numbers in Robins’ text files
Power min count max count min mV max mV theory
1kW 424 597 -425 +421 +/-442 mV pk
2kW 374 678 -670 +817 +/-884 mV pk
3Kw 367 748 -704 +1159 +/-1326 mV pk
So it looks like with the 150 ohm burden, the YHDQ negative output is being limited to about -700 mV at 2kW and above. Perhaps the SMD code of WP does mean a zener diode!