Understanding Headphone Measurements & Hifiman HE-6 Review (video)

[Robbo99999]

For grins I set up a pair of calibrated GRAS 40AG microphones; one on the bench next to the APx555 and APx1701 with cooling fans running. The 2nd microphone was set up across the hall in a bedroom with two closed doors in between.

The ambient noise was significantly lower across the hall from the instrument fans.

Thanks DT

View attachment 120822View attachment 120823

Although at 94dB headphone sweep measurement the gap between the two lines would be a lot less, because that's just how dB works (as well as our sonic perception). (I'm betting you could work out or measure the difference of an identical headphone frequency sweep (no headphone position change on the device) in the two different rooms to really ascertain the difference between the two in terms of how much it can affect the reported frequency response.)

(Plus the headphones will attentuate the background noise to varying extent depending on headphone design - eg closed back vs open, and you measured without any headphones on the ears).

You'd really have to measure a few different headphones in the different rooms to see the effect of the background noise, and you'd have to make sure that the headphone position wasn't changed at all when comparing measurements between the two rooms, i.e. you'd have to leave the headphone on the device without remounting it as you move it to the new room, and be certain that the headphone hadn't changed position accidentally as you moved the device (ie you'd measure it back again in the original room to confirm the measurement was the same).

 

[amirm]

For grins I set up a pair of calibrated GRAS 40AG microphones; one on the bench next to the APx555 and APx1701 with cooling fans running.

As noted you at least need to put a headphone on the fixture to run such a test. Even an open back headphone provides some filtering.

Outside of that, as soon as I got my APx555 analyzer, I bought a rack and put it in there to quiet it down. It was so noisy sitting on the desk that I could not think straight!

In addition, I have the GRAS fixture about 6 feet away at different elevation to the AP (sitting higher). It is also sitting on an isolation pad. We also live in a quiet area. Combined with high measurement level of 94 dB, there is no problem at all for frequency response measurements and even distortion. I have shown distortion measurements down to 40 dBSPL and even lower:

Note that 30 dB lower noise floor represents 1% error in frequency response. So we definitely don't need more than that.

 

[B4ICU]

Don't kill the messenger...
I thought that a good thing can get better. The tests are kind of a routine (after reading this site for a while, I'm getting the impression that they kind of repeating them self), projecting data: plenty of it, but sorry to say so, they do not provide a full peace of information. That's the message.

A chinese eats at a restaurant, and the food is not good!
If he likes the owner, he will let him know about it.
If he doesn't likes the owner, he will leave without saying a word...

 

[DualTriode]

Thank you @amirm for sharing how you measure headphones and attenuate the ambient test instrument noise in the same space.

I calibrated the GRAS 40AO measurement microphones with a 94dB calibrator and reran the tests one microphone on the bench and one across the hall.

I ran another test with both of the microphones hanging about 6 feet away and 2 feet higher than the bench.

Thanks DT

The next plots are in the same room 6 feet away and 2 feet higher than the bench.

 

[phoenixsong]

Drop + Sennheiser PC38X Review: Inappropriately Musical – In-Ear Fidelity (crinacle.com)
Hi Amir, with regards to measurements what could explain Crinacle's displeasure in its Technicalities section? It is shown to have close-to-desired frequency response, low distortion and has angled drivers too. Is group response responsible for headphone performance in the aforementioned area?

 

[amirm]

Hi there. Hard to know without us repeating the measurements to see if that is what we get, and what our subjective results are. Clearly it is a quandary for him and "does not compute." Maye a distortion measurement helps there.

 

[don'ttrustauthority]

So are you saying Harmans target is for preference, not transparency or accuracy?

 

[DualTriode]

So are you saying Harmans target is for preference, not transparency or accuracy?

Hello,


The Harman curve is the result of extensive science and research. Many topics were studied with multivariate statically analysis. Google, multiple regression.


Sean Olive is a PhD. Research Scientist.


Frequency Response is the most representative single variable to chacterize the Harman Curve.


You cannot have transparency or accuracy without first having the correct Frequency Response.

Vocabulary is the single most important variable in determining intelligence.

Thanks DT

 

[amirm]

So are you saying Harmans target is for preference, not transparency or accuracy?

There are no standards of transparency or accuracy in audio when it comes to final sound reproduction. We simply don't know what the production sounded like when the talent and label approved it. Somehow though, we all have some internal compass of what we think is "correct sound." Decades of research have results in this being flat on-axis for speakers. And for headphones, it is more or less the sound of those speakers being heard by a headphone fixture.

For intermediate devices, we do have standards of transparency because it is expected that they don't change what goes into them and what comes out. With sound reproduction devices, we don't.