Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › Headphone non-linear distortion measurements
New Posts  All Forums:Forum Nav:

Headphone non-linear distortion measurements

post #1 of 12
Thread Starter 

This thread is a spin off from the CSD thread. Some HF'ers had asked that I take a non-linear distortion measurements (harmonic + intermodulation + whatever else.)


So what got me thinking is this hypothetical question: "What exactly does 1% harmonic distortion mean at 50Hz @90db?". Well, IMO, it means "not very much." The reason it means very little is that we don't know the exact nature of that distortion. It's best to show via pictures. The following three graphs below show the frequency spectrum of three headphones (HD800, HP1000, and LCD3) being fed a 50Hz sine wave signal so that it measures 90db.


The extra crap that is seen (anything other then the 50Hz spike) is distortion. There are some caveats to these graphs: the sub-bass 20-40Hz consists of a lot of environmental noise - it should be ignored. The noise floor is fairly high - you can definitely see extraneous crap up high in the spectrum and in the bass and lower midrange regions.


Our main focus will be on harmonic distortion, which would be multiples of 50Hz.


HD800. For anyone who's heard the HD800, it is a very good performer in the bass department. Good extension. A little dry sounding. The 3rd harmonic (150Hz) accounts for most of the distortion of the 50Hz signal. It's -40db down. The 2nd harmonic (100Hz) is over -45db down from the fundamental. Overall very good, especially for an open headphone.




HP1000. Anyone who's had the opportunity to hear the HP1000's bass knows it's pretty slamming. Maybe even a bit muddy. That's what second order distortion sounds like. There is quite a lot of second order distortion (100Hz), with it being less than -30db down from the fundamental.



LCD3. Everyone who has heard the LCD3 knows how amazingly loud and clean the bass on these headphones can be. I've wanted to take this measurement for a long time because I thought it would be perfect and indeed it is. We can't even see any distortion at the 2nd harmonic! This is really amazing. The 150Hz spike is environmental noise (I don't know why it's there tonight), so effectively 2nd order and 3rd distortion of the 50Hz signal at 90db is not measurable - it's below the noise floor of the system and environment.




I'm going to keep things simple for a bit and just focus on 50Hz sine waves. I'll throw up a few more measurements of some headphones I have lying around.

Edited by purrin - 2/23/12 at 10:35pm
post #2 of 12


post #3 of 12

Are any of these headphones modded?

post #4 of 12
Originally Posted by gurubhai View Post

Are any of these headphones modded?

LoL!  I'll let him purrin answer that one!  tongue_smile.gif


post #5 of 12
Thread Starter 


Originally Posted by gurubhai View Post

Are any of these headphones modded?

All of them. I measured them unmodded - did not record results - not much of a difference.


post #6 of 12
post #7 of 12

Nice ! I'm looking forward to see the double-tones results with the other frequencies.


Are you planning to make a comparison between single-tone VS double-tones results ? IIRC you've been stating in other threads that double-tones results are more relevant, because more faithful to the real-world use (i.e. music), and I'd be really interested to see how it translates into distorsion graphs. 

post #8 of 12

Awesome work.

post #9 of 12

Awesome.  Thanks for doing all the dirty work (i.e. tedious measurements).  Its about time some decent measurements were taken biggrin.gif.

post #10 of 12
Thread Starter 

thanks - i'll be keeping with the single tone measurements for now - the double and multi-tones will be later.


Up next for the 50Hz @ 90db tests.


V-Moda M80

T50RP (stock)


post #11 of 12


Thanks for beginning these investigations into the non-linear distortion of headphones. Harmonic distortion components certainly play a role in the sonic character of headphones, and THD tells us very little. I'm sure the information you provide in this domain will prove important for headphone evaluation.


The DIY loudspeaker design crowd has settled on the following style of graph as the best single view of driver/loudspeaker distortion. Using a stepped sweep or continuous sweep, these plots show the levels of second through fifth harmonic (along with total level [or first harmonic]) over the whole frequency range. By showing the whole range, any small problem areas can be seen and different speakers/drivers are easily compared.


Please consider sweep based distortion analysis and this style of plot in your investigations.







post #12 of 12
Thread Starter 

Yeah, I can't do the Zaph audio style graphs unless I cough up more money to upgrade my ancient PC-based CLIO system. frown.gif The disadvantage of the above is that it only plots harmonic distortion. IM distortion is not indicated - and this is worse sounding than HD IMO. You'll see once we get to the two and multi-tone tests.

Edited by purrin - 2/25/12 at 8:54am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Headphones (full-size)
Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › Headphone non-linear distortion measurements