Headphone CSD waterfall plots

[purrin]

@arnaud:
 

Black: "Rigid Plate" Left Channel
Blue: "Rigid Plate" Right Channel
Yellow: "Anechoic Plate" Left Channel 
Red: "Anechoic Plate" Right Channel 
 

• The rigid plate consists of a CD-ROM with a 2mm piece of non-porous kids construction foam adhered to the top of it. A hole of sufficient size is punched to allow the end of the microphone wand to fit flush with it.
• The headphone has "Anax mods" consisting of 2mm pieces of non-porous kids construction foam adhered to the metal ring and exposed plastic portion inside the headphone.
• The original plain HD800 impulse response file I sent you was of the left channel.
• Yup, that's right. Sennheiser's driver matching is nothing short of amazing.
• Now you understand why I don't like running rigid plate or dummy head tests. The results stray too far from perception. Although the calculations are mathematically fascinating, it's just too inconvenient to create specific HTRFs for each type of headphone enclosure; not to mention the combination of individuals' variables for each headphone. Of course the hard part would be to determine what the real "earcup" average is, if that is even possible. IMO, the real earcup average are actually the anecholic plate results - the modded HD800s sound exactly like that to me: warmish with valley in the high-mids and energetic throughout the entire treble. With the Anax mods, my HD800s have that 5-6k peak totally suppressed.

 
I will send you the impulse response files out-of-band. One of these days I'll get another Panasonic measurement mic and shove it in my ear.
 

 

[purrin]

LCD2 revision 1

Anax was a witness to these measurements. He also helped out. We re-measured to ensure results were consistent and repeatable. The left channel was 1.5 db louder than the other. The driver frequency response starts to vary significantly after 5kHz.
 
 
 
 
 

 

[rhythmdevils]

WOW.  Are these accurate?  Might that explain the resonant/reverb sound I hear in them? 

 

[LFF]

I'm stunned!

 

[purrin]

     Quote:

WOW.  Are these accurate?  Might that explain the resonant/reverb sound I hear in them? 

Anax and I shifted the cups around, performed several measurements, dis-assembled the measuring rig, re-assembled it, re-measured the LCD-2, and re-measured other already measured headphones to ensure the results we were getting were consistent and repeatable. I kind of felt we were giving this headphone some special treatment not afford to others during the measurement process.
 
For some odd reason, we did not hear the reverb effect. I think k3ct or someone said this effect seemed to be amp related. I know for sure I heard the reverb effect from the Peak/Volcano. However, the measurements do show this headphone has more of a semi-closed than open nature.
 
Quote:

I'm stunned!

 
The plots I produce are intentionally designed to show flaws as much as possible. The db range is 0 to -36. There is absolutely no smoothing applied in the frequency domain. The FR is purposely exaggerated (compared to other commonly seen plots) to highlight non-linearity. That being said, the plots here are no different and consistent with the other ones I've already produced. There was only minor adjustment I made with the more recent plots: I put in a windowing function to smooth out the progression of the plots over time.
 
All that being said - the LCD2 does play very well with the BA. I've never heard it sound better.

 

[Head Injury]

Now do the LCD-2 rev. 2 so I know what I'm listening to 

 
The response looks pretty impressive in the upper mid-range, not so good in the lower mid-range. It looks about as expected for a "warm" headphone. That 9kHz resonance is really strange, especially in the left cup where it looks razor-straight. Everything around it dies off in just over a millisecond 

 
Now I'm really interested in what the rev. 2 does to correct these things, if anything at all. Part of me hopes it does a lot so I can pat Audeze on the back, part of me hopes it doesn't so all the "creamy" talk in the LCD-2 thread dies down.
 
Looks like I need to try an HD800 some day, even if I'm afraid of its bright sound signature.

 

[rhythmdevils]

Quote:

Originally Posted by Head Injury /img/forum/go_quote.gif

 It looks about as expected for a "warm" headphone.

Warm has nothing to do with it.  Treble peaks don't make a headphone faster, they just make it seem faster.  That narrow band of ringing is independent of both planar technology and it's "warm" (ie lack of treble peaks) sound sig. (afaik)
 
FWIW, very few people seem to hear it, including some really good ears so it's not something to worry about.  But it does make me feel a bit less crazy.  I couldn't stand it personally, and sold my rev 1's for this reason alone. 

 
 
Can someone send their Rev2's to Purrin? 

 

[Anaxilus]

Quote:

 

Can someone send their Rev2's to Purrin? 

Hopefully one shows up at the local meet in Oct.  We might have another r.1 show up that day as well so we could verify the results.
 

 

[purrin]

Quote:

Looks like I need to try an HD800 some day, even if I'm afraid of its bright sound signature.

I've only tried the r2 very briefly on my own rig, it sounded peakier than the r1. One thing I must say is that I've heard a lot of variation among the LCD2r1s.
 
Quote:

Looks like I need to try an HD800 some day, even if I'm afraid of its bright sound signature.

 
However much I love the HD800s, I have to say: be afraid, be very afraid. I've applied "Anax mods" to mine to drop the excess treble energy.

 

[Head Injury]

Quote:

Warm has nothing to do with it.  Treble peaks don't make a headphone faster, they just make it seem faster.  That narrow band of ringing is totally independent of both planar technology and it's "warm" (ie lack of treble peaks) sound sig. AFAIK

I was talking more about the ringing in the lower mid-range. Either that causes some of the "warm" sound, or it's caused by the louder lower mid-range (and thus caused by the "warm" sound).
 
I wouldn't say warmth is attributed to a lack of treble peaks, either. I think it's caused by loud lower mid-range, around 300-500Hz or higher. I'd say the LCD-2 is "warm" because it's flat up to 1kHz.

 

[purrin]

Quote:

Originally Posted by rhythmdevils /img/forum/go_quote.gif
 
FWIW, very few people seem to hear it, including some really good ears so it's not something to worry about.  But it does make me feel a bit less crazy.  I couldn't stand it personally, and sold my rev 1's for this reason alone. 

 

There is so much variation among the LCD2s that I've heard (including this pair with a peak at 7k on the left channel and a peak at 9k on the right.) I think this must be the 9th pair I've listened to. Remember at the Bay Area meet we at four LCD2s on my table at one point and we were hearing some slight differences among them? From my experience, some LCD2s (I'm talking r1) tend to have that peak more than others. The LCD2's slow rolloff helps with hiding this peak. The magnitude of the peak with this pair isn't that bad, but still, it is very audible to me. Anax can to attest to this: I winced at some sections (harmonics of David Byrne's voice) of one of the reference tracks that we use (Talking Heads - This Must Be the Place).
 

 

[arnaud]

Hi Marv,
 
You're continuing with the great work, very nice . Looking at the data from InnerFidelity, I can't see this ringing in the LCD2 (rev2) pair that Tyll had on his hands. Setup differences? Large variability across samples? 
 

 

[purrin]

^ Zoom in carefully at the tail end of the HF2 impulse response data (both Tyll and I have our separate measurements) in the last 1 ms (by this I mean from 2ms to 3ms). I believe Tyll's hardware/software does some awfully wierd unnatural smoothing to the waveform.

 

[arnaud]

Quote:

^ Zoom in carefully at the tail end of the HF2 impulse response data (both Tyll and I have our separate measurements) in the last 1 ms (by this I mean from 2ms to 3ms). I believe Tyll's hardware/software does some awfully wierd unnatural smoothing to the waveform.

Yeah, this is not all clear yet as, somehow, the impulse response exported from his MLS setup is not in the sample sample rate / resolution as the raw MLS data. There's some kind of decimation going on and possibly artificial windowing in the process but none is mentioned in the user manual (at least I couldn't find any reference to that). Hopefully Tyll can revisit this at some point by changing the export settings but he's pretty busy with all the ES measurements at the moment though .

 

[purrin]

I thought I saw some setting in the software screens - or maybe it was in the manual. Gimmie a second to post a graph or two. It's not smoothing, but more like a high-frequency filter.