[goodsound]

Whenever I saw a "distortion chart" for a headphone that shows distortion for just one frequency I always wondered - what about the rest of the band ?

Well here it is - the full range distortion measurements for the headphones I own - KSC-75, AKG-K501 and ipod nano. As a bonus it includes the frequency response also. (White graph is freq response, Green is distortion).

I did these tests using RMAA in acoustic mode.

What I find interesting and amazing is how the sonic signature of each headphone matches the distortion measurements. See my comments below.

KOSS KSC-75:

AKG K501:

Stock ipod nano headphone(just for comparison):

It is important to not just look at the relative but absolute distortion also.
Distortion worsens for all of them in the low frequencies but look at what the absolute distortion is at 100hz and 500hz for all three. That explains why the nano headphone sounds like it does!

The distortion mountain at ~4Khz in KSC-75 explains the high-end harshness.

The very low (well atleast relatively) distortion of the k501 from 1Khz all the way up to 10khz confirms its smooth and pleasant midrange and highs.

As a next step I would like to split up the "total" distortion into 2nd and 3rd harmonic. Kind of give an idea about the odd and even harmonic content of the distortion, the ratio and proportion of which also defines in a big way how a device sounds.

oh btw, the measurement equipment is an ECM8000 measurement microphone, Behringer UB802 mixer that servers as a mic pre-amp and the Audigy 2 NX USB soundcard.

 

[sound_man]

Ok.

I'm a bit shocked concerning how bad the distortion of the Koss @ 4KHz is...
maybe I should finally go ahead and look for some new phones

Some time ago I wrote a program which reads these frequency response plots (headroom-style) and creates an inverse filter (basically an equalizer) and I have been using various versions of such filters over the past few months with my phones. It's an incredible improvement (it has become impossible for me to spot any serious colorations any more), but one can't get rid of the harmonic distortions so easily!
If there were some IEMs which are as good as your AKGs in this with respect to this I'd buy them right now (since frequency response plays next to no role for me), bur I fear that most IEMs are far worse than even the Koss is (at least headrooms plots @ 1000Hz do indicate that), so it would be very cool if you could measure a few of these (maybe Shure E500, though I'd be most interested in J-Jays, b/c their distortion seems to be on par with the HD650)...

Btw: It might be possible to generate a volterra filter from the raw data which you recorded for RMAA. With this (and some luck) you could even be able to eliminate the harmonic distortions (there are a lot of papers on this topic, but of course manufacturers of expensive speakers would wish that no one ever makes progress on these things

 

[goodsound]

The other side of it is that the distortion is in a relatively narrow range. Its a spike in a small range of frequencies so when listening to music it might not be as prominent as it looks. But it is discernable nevertheless.

But its almost unbelieveable that you would reply to my post today! Just this morning I happened to look at the Grado SR-60 advertised in the paper and I thought to myself - wish I had done these full range distortion measurements on the Grado SR-80 before I sold it. That would have thrown some useful light on its very obvious tonal characteristics. And I thought maybe I could reply to this thread of mine and see if someone is willing to loan me a SR-60 or SR-80, then I can still do that. Maybe its a sign!

 

[Lad27]

@OP: any idea to which order is this THD calculated? Just 3rd order or does it go up to 7th (or possibly even higher)?

Just curious. Thanks.

 

[Liver]

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif Ok. I'm a bit shocked concerning how bad the distortion of the Koss @ 4KHz is... maybe I should finally go ahead and look for some new phones Some time ago I wrote a program which reads these frequency response plots (headroom-style) and creates an inverse filter (basically an equalizer) and I have been using various versions of such filters over the past few months with my phones. It's an incredible improvement (it has become impossible for me to spot any serious colorations any more), but one can't get rid of the harmonic distortions so easily! If there were some IEMs which are as good as your AKGs in this with respect to this I'd buy them right now (since frequency response plays next to no role for me), bur I fear that most IEMs are far worse than even the Koss is (at least headrooms plots @ 1000Hz do indicate that), so it would be very cool if you could measure a few of these (maybe Shure E500, though I'd be most interested in J-Jays, b/c their distortion seems to be on par with the HD650)... Btw: It might be possible to generate a volterra filter from the raw data which you recorded for RMAA. With this (and some luck) you could even be able to eliminate the harmonic distortions (there are a lot of papers on this topic, but of course manufacturers of expensive speakers would wish that no one ever makes progress on these things

how did you apply your inverse filter? was it tied to a specific program, or did you have to do coding on your own, or was it a global windows driver (or other OS)?

please tell. very interested.

 

[goodsound]

Quote:

Originally Posted by Lad27 /img/forum/go_quote.gif @OP: any idea to which order is this THD calculated? Just 3rd order or does it go up to 7th (or possibly even higher)? Just curious. Thanks.

Its a total of 2nd and 3rd. RMAA has options to do just 2nd, 3rd or both together.

 

[Lad27]

Quote:

Originally Posted by goodsound /img/forum/go_quote.gif Its a total of 2nd and 3rd. RMAA has options to do just 2nd, 3rd or both together.

Thanks. I thought RMAA could help me, I have specific requirement to measure THD up to and including 5th harmonic.

 

[sound_man]

@OP: Yeah, that was pretty good timing! I hope that someone lends you his/her phones for measurement.
As you say, the distortion peak is relatively narrow, but still it should be quite audible since its at -20db.

Quote:

Originally Posted by Liver /img/forum/go_quote.gif how did you apply your inverse filter? was it tied to a specific program, or did you have to do coding on your own, or was it a global windows driver (or other OS)? please tell. very interested.

Since it's a linear filter, it can be applied by any program which has a proper convolver. But I use (and recommend) foobar2000 as music player / testing ground.

 

[kramer5150]

Thanks... isn't THD normally measured in percentage? Or is that only for electric circuits?

 

[goodsound]

Kramer, happened to see your reply only now.

nope they are interchangeable. % is just more commonly used.

here's a quick calculator - www.sengpielaudio.com/calculator-thd
and some theory - www.dogstar.dantimax.dk/tubestuf/thdconv.htm

 

[pfloding]

This is very interesting!
I'd be interested to know more about the test setup. Any chance of measuring the K701s, BTW? How do you rate the 501s with 701 pads -better than the 701s?
Worth mentioning (unless I missed it above) is that the ear is most sensitive around 3kHz, so any distortion in this area is likely to be more significant than other areas.