From what I understand, the 8kHz resonance was chosen in order to maximize accuracy given how screwy things get near and especially after 10kHz. This has lead to complaints that 711 results don't reflect well what humans hear, especially in the treble region (a fair complaint).
Crin is the one that seems to have started this trend of aiming for a resonance at 8kHz. The reason was mainly convenience, most IEM designs can hit that resonance fairly easily in a 711 coupler + canal extension.
711 aren't "screwy" above 10kHz per se. It's rather more correct to say that they're designed to adhere to a standard that's specified at a specific reference plane up to 10kHz only.
Type 4.3 fixtures also are specified at a specific reference plane, but the way the eardrum simulator is designed means that in practice the notion of a "reference plane" doesn't really have much significance.
This then certainly underlines the limitations of the respective measurements, but even so I do put value in standardization even at such costs. I really do think that inconsistency in squig.link sites with respect to the resonance frequency reflect poor measurement practice rather than a showcase of the grand variety of IEM design. I mean I often don't even see the right and left channels match with the worst offenders! So it is still to me a useful tool to ascertain the quality of the measurer and there is value in that alone.
That seems less like a Squiglink issue rather than an operator issue to me, regardless of the fixture they use.
I can see value in a standard, so what if the standard was to measure IEMs with a set of standardised ear canal extensions of a different length ? Would you find that problematic ?
I'd suggest separating the notion of poor measurement practices from the notion of which resonant frequency to hit. I would even suggest that comparing a pair of AirPods Pro 2 and a pair of Moondrop Chu at the same resonant frequency could be considered "poor practice" given the different nozzle / eartip / body designs (the APP2 is likely to hit a slightly lower resonant frequency than a Chu in a given person's ear).
In terms of what I have seen from resonant frequencies produced by the 711 clones, I haven't come across such ranges as those mentioned,
The ranges mentioned concern variance in real ear canals' length, not variance in what you can get with different IEMs / eartips in a 711 coupler.
One other thing that I find interesting is that software such as the JBL headphone app on my phone tells me that I am inserting my IEMs wrong more often that I want to admit. (I believe that it measures the attenuation of the low end frequencies.) This also tells me about the other side of the coin, which is the limits of humans and their expectations and ability to properly insert their IEMs. It would be great if we could integrate something like that to the couplers so that measurements better reflect "proper fit."
I'd love to have more data on how effective people are in putting IEMs in their ears to minimise leakage issues. Companies that put microphones in the front volume of their IEMs, whether a wired add-on for research purposes, or because they're using TWS earbuds with a feedback path, may have an easier time accessing this data (although for the latter as these earbuds are often designed with copious built-in venting they're less sensitive to leakage in the first place).
Still, for the 5128, the put it in and see what you get attitude leaves me a bit uneasy even if the coupler should extract more consistent and precise results. It would be great to know more about the general practice of the measurers. Even something simple like, "I mark the tips and insert them both the same depth" would make me rest a little easier.
This seems to me less like a 5128 issue than a pinna simulator issue. GRAS' anthropometric pinna should concern you just the same I think ?
It's a lot easier to get repeatable measurements in a coupler finished with a circular extension indeed, which is why I'd love to see a fixture roughly adhering to type 4.3/4.4 impedance in such a format, to provide a complementary set of data.
But why feel uneasy about an ear simulator that's designed to physically represent the entrance of the ear canal in a more accurate way compared to an average ear canal ? If one wants to know exactly what resonant frequency a specific IEM is most likely to hit, it's this, isn't it ?
Personally I'd like to see IEM measurements in both types of fixtures : a fixture (or several) that aim at representing the average ear canal (or a small sample of ear canals representative of a wide population), and a fixture designed to provide easier, more repeatable measurements, as a way to characterise the behaviour of an IEM when varying a specific variable.
Do you have a link to the source of the graphs? It would be interesting to find out more information about how these results were generated. Were the IEMs reinserted each time? Were the canal extensions changed with the IEM remaining still? This all makes a difference.
I measured them. I've 3D printed a set of ear canal extensions with a varying length :
For that set I inserted the May at several insertion depths for each canal extension length (typically four-five times, this results in a bit of overlap across neighbouring extension lengths), and measured the IEM several times at each of these positions, with at least three different seatings (sometimes more) for each extension / insertion depth combination.
Ex here :
I've also occasionally done it with a different approach (try to insert the IEMs at exactly the same physical insertion depth in each extension).
I'd love to be able to measure how an IEM behaves when continuously vary one specific variable (for example with a screw-like design), while keeping the IEM's seating the exact same throughout the course of the test run, but alas that's not easily done given the way ear simulators are designed in the first place.