The problem is that the published diagram shows how the principle of phase cancellation works, NOT what is actually happening with their ANC system. To explain:
If we say the peak value of the sine wave (representation of noise) is +50 and the trough is -50, then if the peak of the noise occurs at exactly the same instant as the trough of the generated cancellation signal we get +50 + (-50) = 0, complete cancellation of the noise and cancellation signal. However, if they don't occur at the same instant, say the cancellation signal is delayed by the equivalent of quarter of a cycle (IE. We move the generated signal to the right by quarter of a wave cycle), we've now got the peak of the noise at +50 at the same instant in time as the generated signal is at -25, so we've reduced the noise by half, +50 + (-25) = 25, not cancelled it. Worse still, if we delay the generated signal by exactly half a cycle we've now got the peaks and troughs of the noise and cancellation signal occurring at the same instant, +50 + (+50) = 100, double the amount of noise!
So why would the generated signal be delayed and what are we actually talking about in real terms? The noise signal has to be picked up by a mic, sent to a processor, a cancellation signal calculated and generated, then sent to your speaker and then the sound wave has to travel from your speaker to your ears. All of this this is going to take more time than it takes the noise to reach your ears. In real terms, a 200Hz sine wave (for example) completes a cycle in 5 milli-secs, so half a cycle is 2.5ms and sound travels at just over a foot a milli-second. So, not even considering the processing/generating delay, if your head were about 3ft further from the speaker, instead of noise cancellation you'd have the exact opposite, double the amount of noise! And, the processing time is likely to take at least a milli-sec or so.
This of course is a simplification, there are a lot of variables and with careful mic positioning and clever freq compensation, you could in theory get very close to or actually achieve complete cancellation but on the other side of the coin: That's with one speaker and one head position (relative to that speaker). In practice, if your seat (or seat back) is set to any position other than that calculated (presumably for the average driver), the cancellation will be compromised, plus of course we've got more than one speaker which are at different distances relative to the driver's head and, unless the system is designed for ANC only for the driver, two or more quite different head positions. In practice, a well designed and implemented system should give a decent reduction in noise but if it's not so meticulously designed/implemented and/or if your head position is some distance from point calculated for the average driver, then you might not get any noticeable reduction in noise and possibly even an increase in those generated freqs. We're only talking about tiny slivers of time and fairly small distance differences making a relatively large difference to the summed sound wave entering your ears.
For the above reasons, I'd be very careful about what I "read in forums" from other owners. They probably don't know the principles or variables involved and even assuming their perception is accurate, a say 150Hz-200Hz hump for one owner could be a relatively flat response (with reasonable noise reduction) for another, depending on their seating position for example. As
@bigshot stated, it's a whole lot easier with headphones, the distance between the HP drivers and the user's ear drums aren't significantly different between different owners, no more than a fraction of an inch or so, rather than potentially a couple of feet or so.
G