One leads to the other, more or less.
The problem is seen up to 700kHz.
Measurements comparing a zero crossing controlled triac vs a opto coupler turning the Triac on during the zerocrossing:
There is no CM inductor, due to the high current of the triac outputs. There is both x and y caps on the triac supply lines.
It's not my design, I am just helping them to pass EMC. The layout is really crappy. CM inductor for a switch mode is placed next to the PSU, but other circuits on the hot side are placed close to the mains side, so it effectively shorts out the performance of the CM inductor. It is not the root cause, since shorting the triac M1/M2 removes the problem, and the implementation I did with turning on the triacs during the zero crossing had the same effect, removing the problem.
The load is a 75W Papst advanced fan, which has build in fan speed control. As seen in the curves I linked to, there is a charging of DC link capacitor for each half wave, so the speed control of the fan is done with vector control probably using a 3 phase power module on a PM motor.