Headphone CSD waterfall plots

[cappellano]

Dear Purrin!
 
Do you have the possibility to mesaure the Tesla T1 with low oputput and high output impedance amplifier? 
It would be really interesting to understand if the T1 is designed with the IEC standard or not!
 
Thanks,
 
Marco

 

[Reticuli2]

[size=10pt]Wow, aside from its upper midrange emphasis, the 701 looks good on Tyll's data, but I suppose it's tough to compare yours and his.[/size]
 
[size=10pt]Is there an overhead of the ESP950 waterfall? [/size]
 
[size=10pt]Square waves and THD across frequency seem to be easier to "get" when you look at them for some things. You know what I mean? It's hard to compare, say, the LCD-2 to the k701 looking at the waterfalls because there's so much variation and intricacy to the visual information. It's a little overwhelming to look at and juxtapose.[/size]
 
[size=10pt]The big swaths of low frequency resonant or delayed energy on a lot of these headphones... Are we assuming that's testing methodology? Have any of Tylls data sets for a headphone compared to the same model tested by Purrin shown the latter erroneously plotting more gunk down there? I know you said that was a possibility, but has it been established? [/size]

 

[purrin]

Quote:

[size=10pt]Wow, aside from its upper midrange emphasis, the 701 looks good on Tyll's data, but I suppose it's tough to compare yours and his.[/size]
 
...
 
[size=10pt]The big swaths of low frequency resonant or delayed energy on a lot of these headphones... Are we assuming that's testing methodology? Have any of Tylls data sets for a headphone compared to the same model tested by Purrin shown the latter erroneously plotting more gunk down there? I know you said that was a possibility, but has it been established? [/size]

 
Again, Tyll's data and mine should not be compared, especially at this time. Different measurement methods, different software, etc. Arnaud is still trying to work out some data collection issues so Tyll will be able to generate better data going forward. Without going into too many details, we believe there is a software decay function applied to Tyll's data to make it look neater than it should be.
 
Also be aware of the frequency scales on the graphs.

 

[purrin]

Quote:

Is there an overhead of the ESP950 waterfall?

 
A good suggestion. I'm privy to the data so I know what's going on. An overhead plot better illustrates the decay times. Try to stretch all three images out on your browser.......
 
Let's play a game:
 

 

• Which one is the 'stat?
• Which one has the most open enclosure?
• Which one may have a little bit of hardness or glare?
• Which one has the most restrictive enclosure?
• Which one has a lot of stored energy in the treble?
• Which one is still one of the fastest dynamics despite being produced in the 80s?

 
Answers are at bottom of post #87, after my rant.
 
Quote:

It would be interesting to see more measurements because it could be a great example of how simple changes in FR affect the perception of every other quality- speed, accuracy, detail, distortion, etc. 

 
Uplifted treble is mathematically similar to the excessive sharpening algorithms used in photos straight out of most P&S cameras or badly processed by Photoshop sharpening filters. True resolution is much more subtle.
 

 

[rhythmdevils]

 
Quote:

 
Quote:

It would be interesting to see more measurements because it could be a great example of how simple changes in FR affect the perception of every other quality- speed, accuracy, detail, distortion, etc. 

 
Uplifted treble is mathematically similar to the excessive sharpening algorithms used in photos straight out of most P&S cameras or badly processed by Photoshop sharpening filters. True resolution is much more subtle.
 

Good analogy! 

 

[Armaegis]

Quote:

 
A good suggestion. I'm privy to the data so I know what's going on. An overhead plot better illustrates the decay times. Try to stretch all three images out on your browser.......
 
Let's play a game:

• Which one is the 'stat?
• Which one has the most open enclosure?
• Which one may have a little bit of hardness or glare?
• Which one has the most restrictive enclosure?
• Which one has a lot of stored energy in the treble?
• Which one is still one of the fastest dynamics despite being produced in the 80s?

 

 

 
I don't have much experience with Stats so it's a not-really-educated guess for me...
1. #3
2. #1
3. #1
4. #2
5. #2
6. #1

 

[Reticuli2]

Quote:

 
Uplifted treble is mathematically similar to the excessive sharpening algorithms used in photos straight out of most P&S cameras or badly processed by Photoshop sharpening filters. True resolution is much more subtle.
 

Yep.  Edge enhancement is sooooo bad on a 1080p picture.  People don't realize that it destroys subtlety and real resolution at the expense of boosting higher-level macro detail.  I'm not sure that's exactly the same as a bright headphone, but sharpness controls are definitely something to be avoided with high resolution.  The tricky part is when screen makers combine the edge enhancement function with a softening function and you're left to figure out which spot turns both off... sometimes there isn't a spot.  Dumb.

 

[Reticuli2]

[size=10pt]2 is the ESP950[/size]
 
[size=10pt]1 looks like the HD800[/size]
 
[size=10pt]3 looks like one of the Grados, probably the HP1000[/size]
 
[size=10pt]The "one is still one of the fastest dynamics despite being produced in the 80s" comment is well illustrated with the (I think) HP1000 overhead. It's hard to see just how well that decays in the upper mids and treble with the 3D/isometric view.[/size]
 
[size=10pt]With all that mush in the lows and mids on the Koss, I'm surprised people rave about the mids as much as they do. Maybe the nice mids sound is because the distortion bleeding into that band is low end harmonics shifting up in frequency and with a significant enough delay that it doesn't "blend" into what you perceive as the midrange.  Kind of like if you put your speakers in the middle of a very live room far away from the walls, you can get rid of much of the live sound and hear the speakers… like a poor man’s anechoic chamber.  Only problem is you lose bass reinforcement.[/size]
 
[size=10pt]The Koss plots also make me think there's something about that bass reflections/distortion timing or phasing that must be obscuring your perception of the bass, because superficially it appears (from the plots) to be a slightly and evenly dark, warm sounding headphone that otherwise should not sound bass lite to anyone... even them crazy Beats fans.[/size]

 

[Reticuli2]

And by "your perception" I meant in general terms, not that you specifically, Purrin.  You said the Koss sounds like that back leading edge looks, right?
 
I'm also thrilled to see on the (I think) overhead Koss plot some consistency with the Rat Sound measurement.  His didn't have very high resolution up top to show us exactly how high the ESP950 goes (it was just missing most of the 20khz band), but yours is showing about 16khz before it does goes a little wanky.  Similar extension of the ER4S.

 

[kkl10]

Very interesting thread.
It's like food for thought!
 
I always felt that there had to be some kind of fundamental measurement that would really help to precognize the sound of a particular headphone (when evaluated at the same light as other measurements).
This measurement had to somehow be able to show at least a glimpse of the acoustical interplay between the earpads, drivers and enclosure.
I'm not sure if these waterfalls are such...
 
Is it correct to assume that these waterfall measurements/graphs are representative of an headphone acoustical behavior?
If not, what is it that these waterfalls do really represent?
What kind of information can one extract from these graphs?
purrin, arnaud could you answer these questions?
 
I always found that the acoustics of headphones were neglected measurements until I found this thead...
I look forward as to how these initial efforts and colaborations will develop in the future!
 
EDIT:
Ohps! Forgot to answer the quiz...
 
I'll answer almost blindly, but I'll take my chance anyway:
 
1 - no idea
2 - 1
3 - 1
4 - 2
5 - 1... or 2?(not sure if I understood this question in the opposite way)
6 - 3
 
And what is ringing?
Do the second and third graphs show more of it than the third?
 
Thanks!
 
 
 

 

[purrin]

Answers are at bottom of post #87, after my rant.
 
Quote:

And by "your perception" I meant in general terms, not that you specifically, Purrin.  You said the Koss sounds like that back leading edge looks, right?

I personally do not think the ESP950 is bass lite at all. Its does lack low end extension (you can't see this on the graphs - they only go down to ~200Hz), but other than that, it's got a slightly dark tilt with a hint of midbass. A lot of electrostatic guys think the ESP950 is forward sounding, but this is probably in comparison to the O2s which are laid-back.
 

 

[Reticuli2]

Oh, those plots only go down to 200hz?  Wow, now we really need some 30hz square waves or THD across frequencies to get any idea what the bass is doing on them.

 

[purrin]

      Quote:

Is it correct to assume that these waterfall measurements/graphs are representative of an headphone acoustical behavior?
If not, what is it that these waterfalls do really represent?
What kind of information can one extract from these graphs?
 
And what is ringing?
Do the second and third graphs show more of it than the third?

 
Cumulative Spectral Decays, as I've performed them, are representative of a headphone's energy throughout the frequency band over time.
They are useful for seeing any issues with stored energy. Detecting ringing, seen as ridges, and evaluating how fast the sound settles.
 
Drivers are not too different from drums. You hit them, and they are going to vibrate at certain frequencies. This is ringing; and obviously bad. Drums are going to have a long decay - this is intentional. Driver decay should be immediate. One the signal stops, the driver should stop. But this is not the case in real-life, we can only get close to it.
 
None of these three headphones ring that much. There is some broad based ringing for a short1/2ms  duration with #1 at 5k. #3 has some very minor ringing in the high treble. If you want to see ringing, John (not Joe) Grados ring like crazy. It's part of their characteristic sound. Click on the SR80i link in my sig. to see how it looks.

 

[Armaegis]

#5: I wasn't too sure how to interpret the "stored energy" part. I was thinking either it was the "reddest" one or the one that had the longest decay
#6: so "fastest" headphone means shortest decay in this case?

 

[kkl10]

Quote:

   
Cumulative Spectral Decays, as I've performed them, are representative of a headphone's energy throughout the frequency band over time.
They are useful for seeing any issues with stored energy. Detecting ringing, seen as ridges, and evaluating how fast the sound settles.
 
Drivers are not too different from drums. You hit them, and they are going to vibrate at certain frequencies. This is ringing; and obviously bad. Drums are going to have a long decay - this is intentional. Driver decay should be immediate. One the signal stops, the driver should stop. But this is not the case in real-life, we can only get close to it.
 
None of these three headphones ring that much. There is some broad based ringing for a short1/2ms  duration with #1 at 5k. #3 has some very minor ringing in the high treble. If you want to see ringing, John (not Joe) Grados ring like crazy. It's part of their characteristic sound. Click on the SR80i link in my sig. to see how it looks.

Hmm... I see.
 
So assuming that you have two headphones to measure - the first is stock the second one is the modified version of the first:
What are the variables that might influence the results of how differently these two headphones measure? The enclosures, the earpads, the materials used in the construction of the headphones, etc...?
Does ringing have any correlation with ressonant frequency?
Do you use some sort of dummy head or similar principle to make these measurements? How do you make these measurements if you don't mind me asking?
 
Sorry for so many questions.
Thanks!