the Iac pin is not "regulated" it is designed to be fed from a current source, in this case the rectified mains via 1Meg ohm or similar, max nominal input is 500uA so for 265Vac mains this would be 375V / 500uA = 750k, but you can use higher.
As it converts the current to other things, one would expect the V on the pin to be in the region of 1v2 - 2v5 to possibly as high as 7v5 ( as it was on the original UC3854 ).
Measurement would confirm - pretty easy to do with a scope.
Thanks very much. Thats interesting that the UCC3817 doesnt need the VREF_____IAC bias resistor.
Its a little odd too, since the problem of mains zero crrossover distortion happens, AYK, due to the vin being very low at zero cross, and thus the chip cannot make the inductor current slew up fast enough...therefore the current error amplifier saturates high, and when the voltage rises up further, a huge overcurrent spike results.....i dont see anything in the UCC3818/7 datasheet showing any mitigation of this whatsoever....as such, i dont see how ti.com can say the VREF_____IAC bias resistor is not needed
Obviously IAC is a current input, e.g. current mirror and you don't need to care for it as long as you are keeping the 500 µA Imax specification.
I guess you have a problem in a different part of your circuit and are simply looking at the wrong place.
...ahh, OK, thanks to Easy Peasy, the circuit that you post actually gets rid of the zero cross problem......or rather, adding the VREF bias resistor (to stop light load noise) causes zero cross distortion , then the two R's and NPN solve that problem.....but yes, it looks like UCC3817 doesnt suffer the typical UC3854's zero cross problem, as per Easy Peasy's kindly donated ti.com doc of the 2nd post above.