What are the best/worst driver designs for headphones?

[KeithPhantom]

I have been thinking about headphone drivers/transducers that regardless of frequency response if fixed by EQ (assuming other measurements to be good enough), they became great. One of these examples is the AirPods Max, it is a pretty accomplished driver but its frequency response curve is not that great, but after EQ, it completely transforms into another headphone while keeping other technical measurements under control (such as harmonic distortion, etc.). What are other examples of designs that have great transducer design but flawed frequency response.

 

[maverickronin]

Most of Audeze's drivers, if their QC is having a good day.

The HD800 series' ring radiator.

 

[KeithPhantom]

Audeze drivers are good, but don’t they vary quite a bit between units?

 

[maverickronin]

Audeze drivers are good, but don’t they vary quite a bit between units?

if their QC is having a good day.

 

[KeithPhantom]

If this is true, this is impressive to me:

 

[MayaTlab]

I have been thinking about headphone drivers/transducers that regardless of frequency response if fixed by EQ (assuming other measurements to be good enough), they became great. One of these examples is the AirPods Max, it is a pretty accomplished driver but its frequency response curve is not that great, but after EQ, it completely transforms into another headphone while keeping other technical measurements under control (such as harmonic distortion, etc.). What are other examples of designs that have great transducer design but flawed frequency response.

I've started measuring my HPs with DIY tube mics in my ear canal. While I don't really want to present this as a paragon of absolute exactitude, quite far from it, it's actually starting to be fairly reliable (around 0.5-1dB in variation ? Something of this kind) for comparative measurements during the same measurement session (ie when the probe isn't moved) - and of course it's only valid for my own samples, on my own head : https://www.head-fi.org/threads/how...guarantee-a-better-sound.958201/post-16405751

My own measurements this way are way better aligned with my subjective impressions of the APM than what I see online from test rigs. If I apply EQ based on the latter to my APM they just sound completely off (way too much energy for me in the ear canal gain region, for obvious reasons since most of them show a pretty significant deficit in that area).

Right channel only, average of 5 reseats each. APM in green, K371 in purple, HD650 (with somewhat fresh-ish pads) in blue, HD560S in red.
Note that I'm not quite as confident with my APM and K371 results as with the HD650 and HD560S. For the latter two the relative difference between the two has been repeated in at least four, five sessions, some weeks apart, while for the K371 / APM only two sessions.

I'm not sure of this at all, but experimenting with clamping force and the spring loaded cups makes me think that the APM's specific design may not work superbly well with some test rigs.

So I actually think that it's pretty decent for a closed back in terms of FR - at least on my own head -, and I actually prefer it to my K371. Bonus is that it's way more consistent because of Adaptive EQ at lower frequencies (and given the rather smooth FR up to 4-5kHz it's quite EQable), while my K371 measurably varies just by leaning forwards... and so far channel matching is quite excellent (since the acoustic design is nearly fully symmetrical front to back it's one of the few headphones you can reverse and still get valid results when measured on your own head). I've also had two copies at home for one of these sessions and they both measured quite similarly.

So back to your question, IMO that's not the best example around, because it isn't that bad in terms of FR - well at least on my head .

 

[KeithPhantom]

I've started measuring my HPs with DIY tube mics in my ear canal. While I don't really want to present this as a paragon of absolute exactitude, quite far from it, it's actually starting to be fairly reliable (around 0.5d-1B in variation ? Something of this kind) for comparative measurements during the same measurement session (ie when the probe isn't moved) - and of course it's only valid for my own samples, on my own head : https://www.head-fi.org/threads/how...guarantee-a-better-sound.958201/post-16405751

My own measurements this way are way better aligned with my subjective impressions of the APM than what I see online from test rigs. If I apply EQ based on the latter to my APM they just sound completely off (way too much energy for me in the ear canal gain region, for obvious reasons since most of them show a pretty significant deficit in that area).
View attachment 136982
Right channel only, average of 5 reseats each. APM in green, K371 in purple, HD650 (with somewhat fresh-ish pads) in blue, HD560S in red.
Note that I'm not quite as confident with my APM and K371 results as with the HD650 and HD560S. For the latter two the relative difference between the two has been repeated in at least four, five sessions, some weeks apart, while for the K371 / APM only two sessions.

I'm not sure of this at all, but experimenting with clamping force and the spring loaded cups makes me think that the APM's specific design may not work superbly well with some test rigs.

So I actually think that it's pretty decent for a closed back in terms of FR - at least on my own head -, and I actually prefer it to my K371. Bonus is that it's way more consistent because of Adaptive EQ at lower frequencies (and given the rather smooth FR up to 4-5kHz it's quite EQable), while my K371 measurably varies just by leaning forwards... and so far channel matching is quite excellent (since the acoustic design is nearly fully symmetrical front to back it's one of the few headphones you can reverse and still get valid results when measured on your own head). I've also had two copies at home for one of these sessions and they both measured quite similarly.

So back to your question, IMO that's not the best example around, because it isn't that bad in terms of FR - well at least on my head .

Umm, I am not talking about FR, but the other metrics (harmonic distortion, group delay, impedance/sensitivity). FR (assuming good performance in the other measurements) can be easily fixed by EQ, but this requires that the transducer is good enough to handle this without becoming a mess in other aspects. Example, you could fix the HD 600 FR in the bass by using a bass shelf, but the driver already has ~5% THD without any boost around that area, I would not use the required EQ since the distortion measurement is too high even before EQ.

 

[MayaTlab]

Umm, I am not talking about FR, but the other metrics (harmonic distortion, group delay, impedance/sensitivity). FR (assuming good performance in the other measurements) can be easily fixed by EQ, but this requires that the transducer is good enough to handle this without becoming a mess in other aspects. Example, you could fix the HD 600 FR in the bass by using a bass shelf, but the driver already has ~5% THD without any boost around that area, I would not use the required EQ since the distortion measurement is too high even before EQ.

I am not sure I listen at levels where the HD6... series' THD is a problem. But I certainly like them a lot for EQ because of the lack of significant high Q peaks or dips, and good channel matching - and of course the very low THD above a few hundred hertz. Given when they were engineered I still find them one of the most impressive headphones I've ever tried (and my HD650 remains my daily driver at home).
FR can't always be easily fixed by EQ even when other metrics are OK. My first copy of the Sundara (cf. link above) was an un-EQable mess.
One thing of note in regards to the APM's THD : it is possible that the ANC circuit is used to reduce it (like on the QC35 : https://diyaudioheaven.wordpress.com/headphones/measurements/bose/qc35-ii/). So if that is indeed the case, at this point it's a little difficult to be talking about the driver in isolation, it's probably designed as an overall system.

 

[KeithPhantom]

I am not sure I listen at levels where the HD6... series' THD is a problem.

Well, going by Fletcher-Munson, we need a boost in the bass area since the human ear is less sensitive in that area. So, even if we measure distortion at 94 dB SPL at 1 kHz, the bass has to be increased to higher values due to insensitivity, also driving distortion to higher levels.

Even at 94 dB SPL, from 20 to 80 Hz, THD is at >1%. This is assuming that these frequencies are playing at these amplitudes. But in day-to-day use, at an average volume of 80 dB SPL, bass (at least the region of 20-200 Hz) will be playing at least 10 dB SPL higher than other higher frequencies (specially 1-5 kHz).

Audibility is also what I am not looking for in this post, it is more of an academic post where we can all learn about interesting transducer designs and their engineering.

One thing of note in regards to the APM's THD : it is possible that the ANC circuit is used to reduce it (like on the QC35 : https://diyaudioheaven.wordpress.com/headphones/measurements/bose/qc35-ii/). So if that is indeed the case, at this point it's a little difficult to be talking about the driver in isolation, it's probably designed as an overall system.

Great, I hope other manufacturers start using these clever solutions instead of purely relying on physical and on-driver attempts that don't really lead anywhere

 

[JWAmerica]

I am not sure I listen at levels where the HD6... series' THD is a problem. But I certainly like them a lot for EQ because of the lack of significant high Q peaks or dips, and good channel matching - and of course the very low THD above a few hundred hertz. Given when they were engineered I still find them one of the most impressive headphones I've ever tried (and my HD650 remains my daily driver at home).
FR can't always be easily fixed by EQ even when other metrics are OK. My first copy of the Sundara (cf. link above) was an un-EQable mess.
One thing of note in regards to the APM's THD : it is possible that the ANC circuit is used to reduce it (like on the QC35 : https://diyaudioheaven.wordpress.com/headphones/measurements/bose/qc35-ii/). So if that is indeed the case, at this point it's a little difficult to be talking about the driver in isolation, it's probably designed as an overall system.

While I have no measurements to validate this assertion, my Tozo NC9 TWS IEMs sound better with ANC active. Transparent mode does not sound nearly as good.