squarewaves for headphone measurements is it useful ?

[RayDunzl]

I do agree that the HF content of music signals does not reach the same SPL as that of the test signal.

Assuming you were still talking about square waves...

The square has a fundamental and a series of odd harmonics (the higher frequencies) at decreasing amplitude.

The amplitudes are 1/3, 1/5, 1/7, 1/9, 1/11 of the fundamental, and so on.

In Decibels, that would be approximately -9.5, -14, -16.9, -19.1, -20.8, and so on (again).

With a 100Hz square the frequencies would be 300, 500, 700, 900, 1100 (yes, and so on) in this partial sequence.

The 18,100Hz, the 181st frequency in the harmonic series, would be at 1/181 of the fundamental amplitude, or -45dB.

That level appears similar to what I often see in an RTA (in-room) of music with normal (cymbals/snares/whatever) HF content.



100Hz square, happens to hit about -45dB here (in-room) :



So, I could argue that music does reach the same SPL as a square test signal.

 

[Krunok]

Assuming you were still talking about square waves...

The square has a fundamental and a series of odd harmonics (the higher frequencies) at decreasing amplitude.

The amplitudes are 1/3, 1/5, 1/7, 1/9, 1/11 of the fundamental, and so on.

In Decibels, that would be approximately -9.5, -14, -16.9, -19.1, -20.8, and so on (again).

With a 100Hz square the frequencies would be 300, 500, 700, 900, 1100 (yes, and so on) in this partial sequence.

The 18,100Hz, the 181st frequency in the odd harmonic series, would be at 1/181 of the fundamental amplitude, or -45dB.

That level appears similar to what I often see in an RTA (in-room) of music with normal (cymbals/snares/whatever) HF content.

View attachment 16233

100Hz square, happens to hit about -45dB here (in-room) :

View attachment 16234

Nope, he was talking about "HF content of music signals".

 

[maxxevv]

Despite the HD650 looking 'better' on the squarewaves and other measurements the HD800 (with EQ) to me sounds a LOT better in many many aspects.

Is the squarewaves looking "better" one the reasons why most people find the HD650 sound so effortless smooth through most of its frequency range ? Less of the 'ringing' that results in odd harshness and absence of sibilance that can be detected by ear?

 

[shoden]

it's very important yes.. lots of so called low distortion dacs and amplifiers skew the squares waves and also bad drivers cant skew the square waves.. low frequency square wav are also important to measure

it should be measured on dacs, amplifiers, drivers independantly and the ADC need to be perfect

 

[Krunok]

also bad drivers cant skew the square waves.. low frequency square wav are also important to measure

 

[solderdude]

What EQ are you using for both? I wonder if the 5 band on the ADI-2 DAC is sufficient to try this out.

I don't use software EQ but the portable Kameleon

 

[solderdude]

Is the squarewaves looking "better" one the reasons why most people find the HD650 sound so effortless smooth through most of its frequency range ? Less of the 'ringing' that results in odd harshness and absence of sibilance that can be detected by ear?

The FR and speed of the HD650 driver is the reason the 440Hz and impulse response look great. The 40Hz response is quite disappointing though as the bass-extension is not that great. This aspect only becomes obvious once you heard it EQ'ed and with real subbass in the recording.

 

[KSTR]

Great topic!
I find it hard to interpret time domain measurements of headphones to judge their quality. I mean, when we see significant ringing in a square or step response it is hard to tell if it comes from the driver/headphone itself (like diaphragm or back chamber etc resonances and standing waves) or if this is just the effect of the artifical ear with proper pinnae and ear canal shapes (at least that's what I would assume being used). For a reference of sorts we would need a measurement of the dummy head / artificial ear when exposed to an anechoic far field of sound pressure coming from an excellent linear-phase speaker which also must be anechoically flat on-axis. Probably a set of measurement under different lateral angles would be a nice data pool to compare with.

Of course a comparison between squares or step responses from different cans still tells us a lot (and I would like to thank you for all the measurements in that link, I used the HD700 curves for a starting point to EQ these somewhat misaligned phones, much better now).

 

[RayDunzl]

I found you don't need to run impulses or steps to see what an impulse or step would look like.

I originally thought "Gee, how can a quick sine sweep give you that data? It's ridiculous".

So I experimented with recording in-room a real impulse (single full-scale sample) and step (as a 10Hz square) and compared that to what gets calculated from an in-room recorded swept sine.

Impulse: https://audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44352

Step: https://audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44440

 

[KSTR]

Yes, the impulse response (IR) obtained by log sweep and convolution (most measurement programs use this because it is very robust) is the core of everthing so to say, by processing it you can get any variety of ways to display it and its effects. If you have the IR you can display it as is or convolved with a step function or a square wave or just any signal for that matter. Unless there is gross distortion at work the results are visually identical as well as numerically, at least to first order. And the Fourier Transform gives us the frequency response (mag and phase), a set of overlapping short time FFTs gives CSD/waterfalls. It's all tied together by the IR.

 

[solderdude]

Great topic!
I find it hard to interpret time domain measurements of headphones to judge their quality. I mean, when we see significant ringing in a square or step response it is hard to tell if it comes from the driver/headphone itself (like diaphragm or back chamber etc resonances and standing waves) or if this is just the effect of the artifical ear with proper pinnae and ear canal shapes (at least that's what I would assume being used). For a reference of sorts we would need a measurement of the dummy head / artificial ear when exposed to an anechoic far field of sound pressure coming from an excellent linear-phase speaker which also must be anechoically flat on-axis. Probably a set of measurement under different lateral angles would be a nice data pool to compare with.

Of course a comparison between squares or step responses from different cans still tells us a lot (and I would like to thank you for all the measurements in that link, I used the HD700 curves for a starting point to EQ these somewhat misaligned phones, much better now).

The posted plots are taken without an artificial Pinna nor ear canal. It is driver/pad/baffle effects only.
Have seen plenty of plots where most of the seen 'ringing' can be ascribed to the artificial ear canal.

 

[Sal1950]

Ran into this recording by John Atkinson on the PS Audio/Infinity IRS system. The third cut at the end has what sounds to be some incredible deep organ bass. Direct into my Emotiva DC-1 DAC > Senn HD650 the bass is very deep but does sound a bit strange with the drivers almost fluttering. Maybe it's normal and due to a weakness in either the recording process or my Senn's? Maybe someone could take a quick look at the file in Audacity , REW or something to see what is going on? Does the deep bass look clean there?

 

[derp1n]

What could you ever possibly hope to discover listening to a recording like that?

 

[KSTR]

The posted plots are taken without an artificial Pinna nor ear canal. It is driver/pad/baffle effects only.
Have seen plenty of plots where most of the seen 'ringing' can be ascribed to the artificial ear canal.

Ah ok, thanks. So a very rudimentary atifical ear, a panel with a mic capsule in a center hole, and the pad pressed on that baffle to get a realistic seal, right? Looks fine to me for comparing different phones in relative terms but might give wrong clues with regard to absolute performance as this is not the exact acoustic cavitiy shape the drivers want to see by design. Any standing wave patterns might be considerably different with strong effect on the square waveshape, this is a somewhat uncontrolled variable here.

 

[solderdude]

It is what it is.

The only 'difference' from real life is Pinna interaction, which is not that big and in the 2-6kHz range only.
The effect of it differs per headphone.