Question about Harmonic Distortion

The audibility threshold for distortion of music varies a lot with the type of music, listening level, person's hearing and listening skills, etc. But several studies put it somewhere in the range of 0.1% (-60 dB) to 0.01% (-80 dB).

The distortion shown for the headphones is in that "gray area" around -70 dB where it's hard to say if it might be audible or not. So it might not be audible with either headphone which takes it out of the picture--at least at the test level and frequency they used. Different distortion measurements could easily reverse the "winner" at a different level or test frequency (i.e. the HD598 might have less low frequency distortion but more high frequency distortion than the T1).

It's difficult to generalize headphone sound quality from those graphs unless a measurement is so obviously bad it's almost certainly an audible problem.

In general, despite people calling me Mr Objective, headphones are an area where you have to use your ears as measurements only tell part of the story.

Ideally you'd only see the 0 dB sine wave and no distortion/noise, that's not realistic however. The 2nd harmonic is the least offensive and even harmonics are generally considered less offensive than odd ones.

What's weird are those distortion products around 3 kHz, but they're also down -70 dB.

Quote:

Originally Posted by xnor 

Ideally you'd only see the 0 dB sine wave and no distortion/noise, that's not realistic however. The 2nd harmonic is the least offensive and even harmonics are generally considered less offensive than odd ones.

 

What's weird are those distortion products around 3 kHz, but they're also down -70 dB.

It's not that weird. Headphones have mechanical resonances that make them behave in rather unpredictable ways compared to a distortion spectrum of an amplifier. Good point about even/odd harmonics.
 

Quote:

Originally Posted by nwavguy 

The audibility threshold for distortion of music varies a lot with the type of music, listening level, person's hearing and listening skills, etc. But several studies put it somewhere in the range of 0.1% (-60 dB) to 0.01% (-80 dB).

 

I'd argue that you're an order of magnitude too conservative here. The number I've seen quoted quite a few times in speaker/source design discussions is -40db, a number of listening tests I've tried show that -30db becomes pretty difficult and the high 30s are definitely my limit. Of course this doesn't account for all variables in program material and distortion type, but I doubt that anything will matter once past -60db. Not the least because of the difficulty of finding a reference speaker which would offer an unobtrusive view of distortion at -80db.

Higher order odd harmonics become increasingly noticeable, so a 5th or 7th may be noticeable quite far down in level (though in practice the 2nd/3rd are almost always the highest in level).

Another annoying thing about judging HD performance is the popularity of the arbitrary 500 or 1khz single measurement that is shown in the above headroom graph. The HD performance of any transducer will begin to utterly break down at certain frequencies and it just takes one or two nasty peaks to ruin a listening experience. Tyll's new graphs at Inner Fidelity offer a vastly better view of headphone performance.

Quote:

Originally Posted by anetode 

Higher order odd harmonics become increasingly noticeable, so a 5th or 7th may be noticeable quite far down in level (though in practice the 2nd/3rd are almost always the highest in level).

 

Another annoying thing about judging HD performance is the popularity of the arbitrary 500 or 1khz single measurement that is shown in the above headroom graph. The HD performance of any transducer will begin to utterly break down at certain frequencies and it just takes one or two nasty peaks to ruin a listening experience. Tyll's new graphs at Inner Fidelity offer a vastly better view of headphone performance.

There are some exceptions to that in some rather popular headphones.

In addition, the broadband percentages given on Innerfidelity are also just about as limited as the Headroom way.  A given percentage of distortion doesn't really tell you what it sounds like in the way the spectrum at a single frequency can.  Neither are perfect and I think you need both of them to get an idea about what a 'phone will sound like.

what's up whit all harmonic distortion graphs from headroom.com beeing the same for all headphones at that biggest peak?

It's a double graph, what does this mean!? (http://www.youtube.com/watch?v=OQSNhk5ICTI (OT))

"Marqués de Arienzo" - Rioja, Reserva 2004 (wine) + At the gates(swedish death metal)

Quote:

Originally Posted by lovleylady 

what's up whit all harmonic distortion graphs from headroom.com beeing the same for all headphones at that biggest peak?

 

Because that represents the frequency of the sinewave used to do the distortion spectra plots.

se

Thanks for answer.

Don't really get it but it's ok.

Ok, let me see if I can explain it a bit better.

To test for harmonic distortion, you feed the device being tested a tone made up of just a single frequency. Let's say 1,000 Hz. Perhaps you remember back in the olden days when the local TV stations would go off the air they'd transmit color bars (or back in the olden olden days, the test pattern with the indian head on it). They'd also transmit a test tone. That was a 1,000 Hz tone.

Anyway, if the device is perfectly linear, then what you will see on the output is only that 1,000 Hz tone. However if it's non-linear, and distorts the signal, you'll see other frequencies at multiples of the test frequency. For example with a 1,000 Hz tone, you'd see frequencies at 2,000, 3,000, 4,000 Hz, etc. These are called harmonics, hence the term harmonic distortion.

The levels of those harmonics relative to the level of the test tone give an indication of how much the device is distorting the signal being fed into it.

The first big peak in the Headroom graphs represent the frequency of the test tone used for the measurement. And in order to make any meaningful comparison between headphones, the level of that tone should be the same for the headphones being tested as it's the level of the harmonics relative to the test frequency that's important.

And that's why the first big peak in the measurements are the same from headphone to headphone.

Does this help?

se

Quote:

Originally Posted by maverickronin 

Neither are perfect and I think you need both of them to get an idea about what a 'phone will sound like.

Agreed.

Quote:

Originally Posted by anetode 

Agreed.

On a side note, it would be awesome if Tyll could put together a big 3D graph for each headphone that showed the distortion products at each frequency sort of like he did in the burn in articles.  That would be almost perfect.

Thank you very much Steve Eddy, very helpful.

Really appreciate the friendly tone around here.

Quote:

Originally Posted by maverickronin 

On a side note, it would be awesome if Tyll could put together a big 3D graph for each headphone that showed the distortion products at each frequency sort of like he did in the burn in articles.  That would be almost perfect.

 

X2, I loved those too.