This likely isn't "static" as in like a random noise generator. Its more likely to be what we HP users would call "crackle" or "ticking", or some combination of the two. That drivers will generate if they are driven too hard.
The sine sweep, within the resolution limits of the test, will cover a range of frequencies and not just 50 Hz. It looks to be odd for about 10 measurement points. My inclination to see it suddenly change like that would be to investigate further. Also it looks like the distortion measurement may only apply to the cup with the dip in the bass response:
It is not a sweep, it is a serie of discrete tones.The sine sweep, within the resolution limits of the test, will cover a range of frequencies and not just 50 Hz. It looks to be odd for about 10 measurement points. My inclination to see it suddenly change like that would be to investigate further. Also it looks like the distortion measurement may only apply to the cup with the dip in the bass response:
But it could be an average of both cups. But for some reason I remember it only being done for one channel in the measurement. If that is the case, then we have no idea what the one that doesn't have the shortfall is doing. Maybe Amir can elaborate more if the THD is an average for both, but either way the measurement will be incomplete or inconclusive if that is true since both sides are marching to their own tune in the region where the concern is.
Edit: I just noticed that the one with the dip has the higher response at low frequencies where the issue was, so it may actually be the worse of the two.
Amir used the word static. After all the crap we give to other reviewers for using jargon that they are unwilling to exactly define, it would be nice if Amir clarified what he heard
Do you have a reference to the specifics of how the measurement is done, out of curiosity? At any rate, there doesn’t look to be a b-spline fit, but a series of line segments that join each discrete point. Given that it gets very granular at high frequencies, it has to be a linear tone spacing that are only on the order of a few Hz apart.
I don’t think it’s done for the same reasons that audiophiles do. He sometimes uses odd terms and wording at times. Sometimes more clarity would be much more helpful, but subjective descriptions being what they are we don’t always agree, and he is also doing a large volume of reviews to boot.
I wouldn’t be that categorical. I see where you come from but it’s not that obvious to differenciate a driver that goes beyond it’s physical limit and digital clipping. they are both quite similar to “static”. I have been fooled before, We are not int the realm of analog electronics saturation which is very recognizable.
I thought it might be extracted from the multi-tone as you say, but I think that may only be for the electronics. I think it may be from the chirp with speakers and headphones, unless you know he does so otherwise. Would be nice if it wasn’t a chirp as the multi-tone measurements are also available in the dataset.
It’s not “ extracted” but it can be the resolution you want, and in real lit’s a set of tone not a sweep. the number of point will tell you how much time you are willing to wait for computation. Only @amirm knows the resolution of this graph, it’s my ”educated” guess that it‘s only one point that make the graph go down that way.
I mean it could be a stepped sine as you say, but the graph gets very granular as the frequency goes up, which is more indicative of it coming from a chirp since that’s the easy way to measure it if you have a good analyzer. Otherwise it could be very lengthy. Maybe Amir can confirm. Edit: Unless you are referring to the nature of the chirp itself. At low frequencies it will be more coarse. From the segments it looks to span maybe 6-7 Hz per point around the 50 Hz dip with no fitting done, just lone segments between each computation point.
No it is a log sweep. Same thing I use for speaker measurements.
Ah ok. Good to know they are indeed the same
That's what I remember from your earlier video on the subject:Thanks for correcting me. You can do both yes.
If it's not obvious then it should be noted in the review, but Amir is blaming headphones. Every such mistake is damaging to the manufacturer (if it turns out it was noise from source).
Unless he did his due diligence and verified that it's not digital clipping.... You seam to assume that it's a mistake. Me I think it is possible, but still unlikely. If the source don't clip, verified by Roon indicator, I would assume that the ADI-2 can only apply digital attenuation, not digital gain, but I am not positive about that. That said yes, it's Amir's platform to test headphones so yes at this point He should know exactly what's going on with the digital gain structure, the USB protocol drivers and handshake process and such considerations as a technical reviewer. I don't think a headphone review is the place to go trough all that but it can make an interesting video. Don't get me wrong, we all want to know how this crackle happened and it does not show in the distortion measurments. For the record it doesn't clearly show in the THD measurments of the Focal Big cans series neither. Amir acknowledge that his set of measurments fails in showing such flaws. It may be something below 20 Hz, it may be other considerations that comes with music and not sines, I don't know, you don't know, Amir's don't know and he admits it. Not sure what more you want. Improve his measurment sets probably. I am one to believe that everything can be measured, but do not think that everything about audio performance is shown by the standard sets of performance metric measurments. Product designers are the one that should and normally can assess to that. They normally perform many measurments that are not presented here to assess issues.
— Steve Temme, Sean Olive et al, The Correlation Between Distortion Audibility and Listener Preference in Headphones
That's interesting, although most headphones show increased distortion at increased SPL from distortion measurements done in various headphone reviews by Amir, and indeed in the distortion results that Oratory sends out to people who have sent him their headphones to measure.......or are you saying that at a lower bracket of testing SPL's that this behaviour you linked can happen?
I suspect this is at lower SPL levels since Sean mentions the hypothesis of it being from changes in compliance at low displacements. I seem to recall this as well. It’s possible if the dynamic driver in this headphone does not exhibit much excursion for a given SPL level, but then its behavior when excursion is needed will be an unknown. I strongly suspect that the pendulum will likely come back the other way with Seans case if we were to test them here at high SPL levels that are normally tested at during the reviews. This has been the first dynamic to seemingly “buck” that trend, but again I think this may not bode well for it in terms of applying EQ.
I have seen this in speakers as well, and it often happens when they sit for a long time and have taken a set. It goes away after some exercise, but the drivers still fall apart at high SPL as one would expect. Still, it could be a possible mechanism to explain he reduced distortion seen. This is especially true given the test signal is a short “chirp” as Amir has confirmed. It would be interesting to see this headphone pushed to higher SPLs or run in a short-term multi-tone test. As a matter of fact, that might even justify performing multi-tone tests in some instances as part of the reviews, but this can stress the transducers, so care is needed. That’s the main advantage of the log sine sweep is that it’s short so it keeps dissipation down and only spends a brief time at each respective frequency.
They measured at 82-100 dB. Goes to show headphone distortion can't really be assumed to necessarily monotonically (in the mathematical sense!) increase with SPL over all frequencies, and distortion at lower SPLs should probably be measured as it could in fact be higher than at higher SPLs (dependent on headphone and frequency), in addition to considering most of a music track is around the lower average rather than peak level, and distortion audibility is actually greater at lower levels as a higher signal level will better perceptually mask its harmonic distortion products.
