Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › How to proof burn-in (and other experiments)
New Posts  All Forums:Forum Nav:

How to proof burn-in (and other experiments)

post #1 of 47
Thread Starter 
You are only given one chance for a first impression.
On the other hand; a first impression with new cans is somewhat controversial because of the burn-in that seemed to be required.
That’s why I decided to let the first sound that reached the drivers of my new W1000’s to be pink noise and record it. That way, I could compare it later on with the sound sig of fully burnt-in cans.

This is my way of working:
I put home made binaural mikes in my ears.Here's a pic.
http://www.tollenique.nl/Head-fi/inEarMike.jpg
I place the cans under test on my head in the normal way, play some pink noise over them on a moderate level to exclude distortion and record to minidisk (Hi-md).
After that I also recorded some music. I made notes of the levels so I can repeat the procedure after burn-in.
I'll inform you of the outcome after a lot of musical hours.


The measurements and sound sig you get are not accurate and don't lead to absolute values because of the quality of the mikes (not calibrated), but of course you can compare different cans. Which led to the next experiment.
Pink noise on DT880's
http://www.tollenique.nl/Head-fi/pinkDT880 1.mp3
Pink noise on W1000's
http://www.tollenique.nl/Head-fi/pinkW1000 1.mp3
Mind the levels! Notice the very different sound sig.

Since this is not a pleasure to listen to, I also recorded a Paquito d’Rivera track with DT880’s and W1000’s. If you are familiar with either of these cans, you can somewhat predict what the other would sound like to you. You should however keep in mind that in case you for instance listen to the DT880 track on DT880 cans, you’ll hear those cans twice… Irregularities in the frequency range are doubled.
BTW: these are fairly large files
http://www.tollenique.nl/Head-fi/mamboDT880 1.mp3
http://www.tollenique.nl/Head-fi/mamboW1000 1.mp3
It works very well for me, but of course this is very personal, because it's recorded with the mikes in my ears. I am very curious if it works for any of you though.

This method could also possibly give answers to several questions that came about these last few years on Head-fi.
- Do some cans have real bass, or is it artificially introduced (a bump in mid-bass for example)
- Burn-in; does it really exist?
- Is there a way to compare cans more easily and leave out a bit of the human preference of others?
- Can you avoid the bad audio memory when trying to compare cans on different occasions?
I've already found some very interesting differences between cans.But that's another subject.
For some of the answers you’ll need calibrated mikes and a fixed procedure for levels, placement of cans etc.

Enough to keep me busy for a while. But first I’m going to enjoy my new cans. Because burnt-in or not, they already sound fantastic.

Completely off subject but still interesting: DT880’s sounded better to me with the mikes in my ears (subdued highs).
post #2 of 47
Sounds interesting... I was considering getting two pairs of k701's to see the difference between the "out of box" sound, and the "burn't in sound"... But that didnt happen... I would really like to A/B a new set Vs. a burnt in pair... cant wait to see some empirical results.
post #3 of 47
That's fascinating.

I listened to both the music samples and you can really hear the differences in the sound sigs. Kind of a cool way to try before you buy??

Anymore you can add.?
post #4 of 47
A very novel approach to the subject. When you're done, it would be interesting to see your "setup".

An interesting approach: instead of relying on fallible human hearing, this attempts to introduce some "numbers" into the debate. Graphical changes have been observed with standard speakers, as you know, but headphones is more difficult.
post #5 of 47
Heres the wave form of the pre burnt in w1000 pink noise.

post #6 of 47
Thread Starter 
Quote:
Originally Posted by adanac061 View Post
That's fascinating.

I listened to both the music samples and you can really hear the differences in the sound sigs. Kind of a cool way to try before you buy??
That's one of my goals.

I also saved some prints of the frequency ranges. I did the same with sound samples from 31 Hz to 20 kHz. This indicates the very even frequency distribution of the (not burnt-in) W1000's but also shows humps in the mid-lows and mid-highs and very highs of the DT880's. As stated, these are all RELATIVE measurements, because the mikes are not calibrated and probably don't produce a completely straight frequency graph.
post #7 of 47
A side note. From my PL750, the music clip with DT880 sounded good. W1000 sounded nasal and just plain wrong. Gotta try it with MS1 too and see if the results are same.
post #8 of 47
Thread Starter 
Quote:
Originally Posted by MaZa View Post
A side note. From my PL750, the music clip with DT880 sounded good. W1000 sounded nasal and just plain wrong. Gotta try it with MS1 too and see if the results are same.
I understand you can get that impression if you use DT880's are a reference. In real life, it's not nearly as pronounced. DT880's sound very dry when you take W1000's as reference.
I partly blame the mikes for this impression also.
post #9 of 47
Quote:
Originally Posted by thrillmetoo View Post
I understand you can get that impression if you use DT880's are a reference. In real life, it's not nearly as pronounced. DT880's sound very dry when you take W1000's as reference.
I partly blame the mikes for this impression also.
Yup. Pl750 are already quite distant sounding cans, so they do pronounce the effect.

With MS1 the W1000 clip isnt as muffled and nasal. Its actually quite good. But bass is wierd I think. DT880 clip was perhaps bit too sparkly and hissing sounding, but bass is deep and overall excellent.

I guess W1000 are not for me? LOL
post #10 of 47
Thread Starter 
Quote:
Originally Posted by MaZa View Post
Yup. Pl750 are already quite distant sounding cans, so they do pronounce the effect.

With MS1 the W1000 clip isnt as muffled and nasal. Its actually quite good. But bass is wierd I think. DT880 clip was perhaps bit too sparkly and hissing sounding, but bass is deep and overall excellent.

I guess W1000 are not for me? LOL
I think you're right right on the mark. The bass of W1000's slowly decline below 70 Hz, whyle at that point DT880's have another small peak (in comparison with the in-ear method). W1000's whyle having bass down till -6dB at 35 Hz, are no bass monsters.

This could be exiting. You probably now already know W1000's are not for you, whyle you never had them on your head...
post #11 of 47
Quote:
Originally Posted by spinali View Post
A very novel approach to the subject. When you're done, it would be interesting to see your "setup".

An interesting approach: instead of relying on fallible human hearing, this attempts to introduce some "numbers" into the debate. Graphical changes have been observed with standard speakers, as you know, but headphones is more difficult.
Since humans beings are the ones doing the listening you can hardly take them out of the debate. It doesn't matter what the numbers are if you can't establish that people can't detect or don't care about the difference in numbers.

That said, getting objective measurements is a good start and it is in my opinion a safe bet that if there are numerical differences that these will matter to at least some listeners with some equipment.
post #12 of 47
Quote:
Originally Posted by edstrelow View Post
Since humans beings are the ones doing the listening you can hardly take them out of the debate. It doesn't matter what the numbers are if you can't establish that people can't detect or don't care about the difference in numbers.

That said, getting objective measurements is a good start and it is in my opinion a safe bet that if there are numerical differences that these will matter to at least some listeners with some equipment.
I see your point, but the OP isn't trying to discern the differences between a fully burned in and non burned W1000, rather, he's trying to prove a numerical difference exists in the first place. Technically, if the frequency graphs of both the non and fully burned in W1000 match identically, than the concept of burn-in is flawed and thus any perceived "changes" in sound character must be made in the listener's head and not by the headphones themselves.
post #13 of 47
Quote:
Originally Posted by M0T0XGUY View Post
...Technically, if the frequency graphs of both the non and fully burned in W1000 match identically, than the concept of burn-in is flawed and thus any perceived "changes" in sound character must be made in the listener's head and not by the headphones themselves.
Not necessarily. Look at the frequency responses of different amps: They're virtually identical, yet the amps sound (sometimes fundamentally) different.

I predict that the pink-noise method won't work. You'll get identical results, simply because -- as already HeadRoom has stated -- their FR measurements haven't shown significant differences between new and used headphones. Moreover I suspect that microphone placement is so crucial that you might get just as great deviations between measurements done at the same time as between those done at different phases of burn-in.

My guess is that the main measurable difference resulting from break-in is in the harmonic-distortion pattern; and maybe -- to a lesser degree -- in the transient response. For both criteria pink noise is unserviceable. You should try tone bursts and frequency sweeps instead.
.
post #14 of 47
Thread Starter 
Quote:
Originally Posted by JaZZ View Post
Not necessarily. Look at the frequency responses of different amps: They're virtually identical, yet the amps sound (sometimes fundamentally) different.

I predict that the pink-noise method won't work. You'll get identical results, simply because -- as already HeadRoom has stated -- their FR measurements haven't shown significant differences between new and used headphones. Moreover I suspect that microphone placement is so crucial that you might get just as great deviations between measurements done at the same time as between those done at different phases of burn-in.

My guess is that the main measurable difference resulting from break-in is in the harmonic-distortion pattern; and maybe -- to a lesser degree -- in the transient response. For both criteria pink noise is unserviceable. You should try tone bursts and frequency sweeps instead.
.
At no point I tried to be absolute with graphs and figures and values. I simply started with pink noise just to check after say 100 hours BY EAR if burn-in exists in the first place. If you get used to the sound of pink noise it shows slight differences in character of different cans immediately. A rudimental frequency response graph will indeed probably show and/or prove nothing.
However as I already stated a few posts back, I also recorded tone scales (boring stuff to listen to) from 31 Hz till 20 kHz, so I can really check SHOULD changes in the frequency respons occur.
post #15 of 47
You're tests look interesting. I wonder what effect the ear canal has on the sound from headphones? Since you're blocking you're ear canals with the mikes, this might skew the results a little.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Headphones (full-size)
Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › How to proof burn-in (and other experiments)