Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › Headphone CSD waterfall plots
New Posts  All Forums:Forum Nav:

Headphone CSD waterfall plots - Page 5

post #61 of 937
Thread Starter 
Quote:
Originally Posted by tdockweiler View Post

Sorry, I feel like a complete idiot trying to decipher these graphs. So, I take it the DJ100 measured very poorly..oh well..

I did expect a lot of spikes near the upper mids and my ears hear this too. I kind of like it though. Certainly it's no studio monitor biggrin.gif

 

 


LOL. I'm going to re-measure them at the request of the owner. I could have messed up on something. They don't sound as bad as the graphs suggest.

 

The CSD plots measure energy content over time across all frequencies based on a "sound burst" that excites all frequencies. They are sort of like a FR over time (technically not really as they measure energy throughout the band in successively smaller windows over time.) It is a good tool to "see" driver ringing (seen as ridges), which typically sounds bad, and to see how fast the driver decays. See this post for what a perfect waterfall plot might look similar to. For a perfect response, the decay should be immediate. You see a wall (meaning all frequencies are excited) at the beginning. The next few fractions of a millisecond, you see the wall fall to absolutely nothing. It's very evident that headphone drivers are far from ideal. Some much less ideal than others.


Edited by purrin - 9/4/11 at 9:56pm
post #62 of 937
Quote:
Originally Posted by purrin View Post




LOL. I'm going to re-measure them at the request of the owner. I could have messed up on something. They don't sound as bad as the graphs suggest.

 

The CSD plots measure energy content over time across all frequencies based on a "sound burst" that excites all frequencies. They are sort of like a FR over time (technically not really as they measure energy throughout the band in successively smaller windows over time.) It is a good tool to "see" driver ringing (seen as ridges), which typically sounds bad, and to see how fast the driver decays. See this post for what a perfect waterfall plot might look similar to. For a perfect response, the decay should be immediate. You see a wall (meaning all frequencies are excited) at the beginning. The next few fractions of a millisecond, you see the wall fall to absolutely nothing. It's very evident that headphone drivers are far from ideal. Some much less ideal than others.

 

Thanks for this explanation. Yeah, they could have the worst graph ever and I'd still love my pair normal_smile%20.gif. I like them for vocal oriented music, but they're certainly not for everyone. I wonder if they would measure totally different with M50 pads. That's what I've been using with mine, but it gives me the impression of much more treble.
 

 

post #63 of 937

Yeah, those plots look identical on the ESP950 with the different amp.  Do the impulse, square, or THD tests look any different?  If they sound different, it's possible something will show up.

 

There is almost a closed headphone level of resonances on the ESP bass plots that make me think maybe we're wrong about the cheap driver housing being entirely to blame.  Perhaps the diaphram itself is interracting with those resonances.

 

Those latest t50rp stock plots were from a completely unmodded can?  They look like a completely different response from the first pair you posted or mine (A) and Tyll's (B).  The next pair of "mods" seemed to decrease resonances, increase mids, and decrease high frequency extension.  But it looks like the plots right above it are from some kind of modification, if less & different than the next.  ??

post #64 of 937

Really nice work Purrin.  It's all really great to see!

 

Why does the sample "ideal" graph curve upwards at the end?  Is that an artifact of some kind?

post #65 of 937
Thread Starter 


      Quote:

Originally Posted by Reticuli2 View Post

Yeah, those plots look identical on the ESP950 with the different amp.  Do the impulse, square, or THD tests look any different?  If they sound different, it's possible something will show up.

 

 

Will do for the impulse. Can't do square and THD yet since I've been too lazy to dig up a separate mic pre-amp (just moved and 50% of stuff is packed and will be packed for while.) For THD, instead of percentages, which are almost meaningless, my preference would be to run single and multiple sine tones and show what lights up in the entire spectrum with an RTA. If you can't tell by now, I like pictures.
 

 

Quote:
Originally Posted by Reticuli2 View Post

There is almost a closed headphone level of resonances on the ESP bass plots that make me think maybe we're wrong about the cheap driver housing being entirely to blame.  Perhaps the diaphram itself is interracting with those resonances.

 

The ESP950s are semi-open. They just don't breath like the STAX  -it's not even close. You blow on the grill on the ESP and air just bounces off it.
 

Quote:
Originally Posted by Reticuli2 View Post

Those latest t50rp stock plots were from a completely unmodded can?  They look like a completely different response from the first pair you posted or mine (A) and Tyll's (B).  The next pair of "mods" seemed to decrease resonances, increase mids, and decrease high frequency extension.  But it looks like the plots right above it are from some kind of modification, if less & different than the next.  ??

 

Yup, those are completely un-modded (T50rpL and T50rp R). They are actually the same pair LFF worked on. One important thing to note for Set 8: timescale, frequency scale, and FFT resolution are different on those plots. I wanted to close in on pertinent areas of interest.

post #66 of 937
Thread Starter 


     Quote:

Originally Posted by rhythmdevils View Post

Really nice work Purrin.  It's all really great to see!

 

Why does the sample "ideal" graph curve upwards at the end?  Is that an artifact of some kind?



The "perfect curve" was measured via a feedback loop. It curves up because the electronics in the path are "slower" at decay in the higher frequencies.

 

So instead of this (side view):

 

\

 \

  \

   \

    \

 

it goes more progressively like this at higher frequencies:

 

\_

   \_ 

      \ 

       \

 

for driver measurements, it won't matter because the drivers are several magnitudes slower than the electronics.      

post #67 of 937
Thread Starter 

Impulse Response for ESP950s from different amps.

Red - Koss Energizer

Black - STAX SRM-323

ESP IR.gifESP R IR.gif

No difference really.

post #68 of 937
Thread Starter 

     Quote:

Originally Posted by tdockweiler View Post

 So, I take it the DJ100 measured very poorly..oh well..

I did expect a lot of spikes near the upper mids and my ears hear this too. I kind of like it though. Certainly it's no studio monitor biggrin.gif

Yeah, they could have the worst graph ever and I'd still love my pair normal_smile%20.gif. I like them for vocal oriented music, but they're certainly not for everyone.

 

Here's a second measurement with the timescale extended from 5ms to 10ms. Little bit better than the first, but it still won't settle! Six to seven spiky ridges. This headphone deserves a special award. 

 

EDIT: There was a bug. This looks better than the original CSD. It still looks bad though.


DJ100.txt.jpg

 

 


Edited by purrin - 9/6/11 at 7:37pm
post #69 of 937

Thanks for the second graph. That is seriously weird. That mountain near the upper mids makes sense to me (and at 5k) but I never would have expected so many spikes in the treble. Those are with stock pads right?
Oh well, I guess for a $80 headphone it could be worse...maybe normal_smile%20.gif I wonder what could cause all those ridges towards the end..

Quote:
Originally Posted by purrin View Post

     Quote:

 

Here's a second measurement with the timescale extended from 5ms to 10ms. Little bit better than the first, but it still won't settle! Seven spiky ridges. This headphone deserves a special award. 


DJ100.txt.jpg

 

 



 

post #70 of 937
Quote:
Originally Posted by purrin View Post

What! Despite the measurements, they do sound better out of the 323 compared to the stock Koss box, but maybe it's just not enough for you.


Correct.  

 

post #71 of 937
Quote:
Originally Posted by purrin View Post

     Quote:

 

Here's a second measurement with the timescale extended from 5ms to 10ms. Little bit better than the first, but it still won't settle! Seven spiky ridges. This headphone deserves a special award. 


DJ100.txt.jpg

 

 

Based on what I learned today...that looks horrible.

 

What award shall you give it?
 

 

post #72 of 937
How about the medal of kindness for making every other headphone look good when osculted through the revealing eye of Marv's CSDs ! Even the stock SR80 is going back up in my esteem after seeing this ;-). Even more interesting is the fact that you could consider the peak is begnin if only looking at the usual frequency response graph.

Nice job Purrin!
post #73 of 937
purrin, is it safe to generalize that FR curves reflect the responsiveness of the driver, while the waterfall charts reflect the interplay between the driver and its enclosure?
post #74 of 937

Either we need to get you a Q701 to measure or Tyll an ESP950, or both.  I'm very curious how those two headphones' measurements compare.

 

Looking forward to seeing how clean the LCD-2 actually is in light of our high expectations of it after the low res plot.


Edited by Reticuli2 - 9/5/11 at 9:41am
post #75 of 937

Are you serious? 

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Headphones (full-size)
Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › Headphone CSD waterfall plots