AKG K60 Vintage Headphone Review

[GXAlan]

This is the measurement of the headphone in the AES paper in 1967 showing the headphone measured on a human using the probe mic in the cup., and the same headphone measured in the 6 cc coupler which they say is accurate only up to 600 Hz. Not sure how you translated that measurement to what you show above

Thanks; I need to edit my post for clarity; At the time, I had not read the AES paper but wanted to find a graph to show people the difference between free field and your research. I usually try to cite my sources, but I'm not sure where I screenshotted that from.

 

[PlayGreen]

Personally, I am quite impressed by the scientific research at that time. We should not forget that there weren't many papers in the 60s about the ear canal modeling. it is too bad that the author didn't add references to the input impedance of ear.

The author had really good ideas. It is just too bad that he didn't consider the head and external ear (concha, pinna) as useful for the sound reproduction over headphones. And unfortunately, the method and assumptions described in the paper aren't entirely correct.

Let us assume a pressure source P_src with its headphone impedance Z_i and the input impedance Z_ear of ear as in Fig.1 of the paper, without the ear volume (coupling volume). Here we consider two ear simulators - GRAS RA0045 (used in Kemar) and GRAS RA0075 (6cc coupler), whose input impedances are illustrated in one of the attached figures. Following both prerequisite introduced in the paper:
1. Z_i = 0, then P_src = P_mic_entrance and there is no alteration of sound at the microphone in both ear simulator and coupler.
2. Z_i = Z_ear_RA0045 then P_mic_entrance = Z_ear / (Z_i + Z_ear) * P_src
2.a If the ear simulator is RA0045, then P_mic_entrance = P_src/2 and there is no alteration of sound at the microphone in the ear simulator (ear drum location).
2.b If the ear simulator is RA0075 (6cc volume), then P_mic_entrance = 1/(1+jwC*Z_ear) where the 1/(jwC) corresponds to the impedance of the 6cc coupler. In this case, we clearly see the contribution of the modeled input impedance in headphone impedance (Z_i = Z_ear_RA0045) in the mic response: the magnitude is constant at low frequencies and there is a notch at 6 kHz corresponding to the anti-resonance of the RA0045 ear simulator.

Now we include the ear volume in front of the headphone.
We have similar results for in the 6cc coupler: two compliances in parallel, thus a decrease of the SPL for microphone at ear simulator entrance and microphone in ear simulator (see attached figure).
If the ear simulator is RA0045, then we can see that the microphone response at ear entrance is similar to configuration 2.b, but now the compliance is the one corresponding to the coupling volume. As a result, there is a notch between the ear canal resonances for the microphone response in the ear simulator.

We can then see that the assumption of the 2nd prerequisite isn't correct because of the presence of this coupling volume.
As the author said, they didn't manage to get an equivalent value of ear impedance for the headphone. This may explain why we see differences between these simulations and the measurements they made in the coupler and at the human ear entrance.
Finally, we can note that there are differences between the input impedance of RA0045 (or Kemar), and their model used in the paper. It explains why there are peaks and dips at high frequencies in Amir's measurement. The response is flat in average because the head, pinna, and concha weren't compensated in their target (it doesn't follow the target in dotted line). And the high magnitude in the bass is probably there because of listeners' preference!

 

[solderdude]

You train your brain to use your ears.

People have a tendency to overestimate their hearing capabilities.
Use your ears to listen to music not as an analyzer.

Of course you can use your ears but they are notoriously inaccurate when used 'normally'.
The only way you can use them as a 'tool' is to perform 'blind' level matched and statistically relevant testing.
Nearly all tests *certainly when using headphones and speakers' can not be performed blind and level matching is not possible.

You can have a preference for something but one should be aware that it is not because of reliability of auditory perception but because of the lack of reliability.

 

[Mr Vinyl]

Yet, a lot of the best recorded music was made in that era...

 

[GXAlan]

Yet, a lot of the best recorded music was made in that era...

These are mine and what the FR doesn’t mention are the spatial effects. With these headphones there is no bass but the sense of front and back are surprisingly impressive. I agree with Amir's subjective impression -- you would never intentionally pick these OVER a newer headphone, but it has a clean midrange (i.e. vocals) and it is not annoying.

If you look at the Harman 2018 Over Ear target against the AKG K60 measurement, you get this, which matches @amirm 's official comparison:

But throw in psychoacoustic smoothing...

Turn up the volume to compensate for the feeble treble and what you see is a headphone that has a boosted Bass/low-mid, which is dependent on ear-seal. Again, you don't overcome the distortion, and vintage nature of the headphone, but maybe in the glass of half-full vs. half-empty, we experience a boosted/congestion in the bass/low-bass as opposed to a clean bass and completely muffled upper end?

 

[Mr Vinyl]

These are mine and what the FR doesn’t mention are the spatial effects. With these headphones there is no bass but the sense of front and back are surprisingly impressive. I agree with Amir's subjective impression -- you would never intentionally pick these OVER a newer headphone, but it has a clean midrange (i.e. vocals) and it is not annoying.

If you look at the Harman 2018 Over Ear target against the AKG K60 measurement, you get this, which matches @amirm 's official comparison:
View attachment 387295

But throw in psychoacoustic smoothing...
View attachment 387296

Turn up the volume to compensate for the feeble treble and what you see is a headphone that has a boosted Bass/low-mid, which is dependent on ear-seal. Again, you don't overcome the distortion, and vintage nature of the headphone, but maybe in the glass of half-full vs. half-empty, we experience a boosted/congestion in the bass/low-bass as opposed to a clean bass and completely muffled upper end?

View attachment 387298
View attachment 387299

I can't imagine the boominess it must have at the louder volume...

 

[GXAlan]

I can't imagine the boominess it must have at the louder volume...

That’s the weird part. There’s not a lot of bass subjectively. It’s more like a LS3/5a maybe. I think what happens is that I probably am somewhere in between where the actual perceived experience is in between the two graphs. Not as much of a bass boost as I suggest and more of a treble roll off.

This also gets to the challenge of level matching something in real life. Even if I wanted to do sighted A/B comparisons but level matched comparisons by editing the source file, I don’t really know how best to level match something when the FR is so widely different.

 

[Mr Vinyl]

That’s the weird part. There’s not a lot of bass subjectively. It’s more like a LS3/5a maybe. I think what happens is that I probably am somewhere in between where the actual perceived experience is in between the two graphs. Not as much of a bass boost as I suggest and more of a treble roll off.

This also gets to the challenge of level matching something in real life. Even if I wanted to do sighted A/B comparisons but level matched comparisons by editing the source file, I don’t really know how best to level match something when the FR is so widely different.

LS3/5a and its variants are great mixing speakers, in my opinion. They are balanced, albeit 'boring' sounding. The AKGs might be good for audio mixing as well. If the midrange, like you said, is well balanced - where the majority of the music resides - then, while mixing, you would have a tendency to compensate by increasing the high frequencies and the bass end. If you have a balanced sound through the AKGs, then in real life it will turn out to be a V shaped sound which is popular today. I'm guessing, but it may be something along the lines of the horrible-sounding NS-10 speakers that some budget studios used to use during the 70s. As far as level matching - I'm not sure how to do that with headphones - but I would suspect to match the spl or something similar? For matching source files, you can use LUFS or RMS, but then you would need to account for equipment differences down the line.