link1896 wrote:
the area filled in on the graph to the left, like in the noctua graph, are bass frequencies.
it is of importance as lower frequency sound is harder to attenuate then mid frequencies, and can excite cases panels and case cavity resonances, but conversely, mid range sounds are more annoying to humans due to their sensitvity to mid range frequencies (matches up with frequency range of the human voice)
In addition, lower frequencies roll off much more quickly as you move away from the source, and at the levels that we're looking at, the frequencies below <100 dB are almost guaranteed to be inaudible outside of a foot or two. I have to wonder why a subjective comparison of the two fans did not reveal low frequency differences. I find it interesting that you attribute the "presence" to a larger amount of low frequency noise, but I don't understand why you draw this conclusion.
I would interpret your FFT graphs in a very different way. To me, the most important feature of the Noctua graph is not the higher amount of low frequency noise, but the fact that it doesn't have many significant peaks. The noise is a little more broadband than the Nexus, especially in the lower frequencies. My gut feeling is that *this* is the explanation of the "presence", since our ears find it easier to pick out pure tones than white noise. With most of its distinctive characteristics hidden by the white noise, the Noctua is very difficult to perceive. Nevertheless, the noise is there, and is presumably taken in by our ears. We do not hear it because our brain filters it out as something to ignore, so it is only perceived subconsciously — in the form of the psychological "presence" that I described.
I'd be very cautious about reading too much into the extra bass in the FFT graphs, and I agree with Mike's decision not to include FFT information in future reviews. I think I have shown that the same graph can be interpreted multiple ways, and it is not clear how to determine which interpretation, if any is correct.
In any case, the audio recordings and our subjective comments do a more accurate job of capturing how the fans noise character than an FFT graph can. And, if the noise character can be captured through these simple methods, why should we try to
explain it via an FFT graph? Including the graph as an explanation is akin to rationalising our subjective impressions, as though we don't have enough confidence in our listening skills to accurately describe our experiences without mechanical aids. Remember that it's the experience we're trying to capture here; explanation should play a secondary role if it has one at all.
In my opinion, the value of an FFT analysis is not explanatory, but predictive. Our subjective impressions can stand on their own without explanation, but the FFT analysis might be able to help predict how the fan would sound under different circumstances. As you said, the additional low frequency noise might make the Noctua more prone to resonance. However, I'm not confident that I can make such predictions with any accuracy, and the difficulty of interpreting the graphs correctly makes me wonder whether such predictions
can be made.