[VisceriousZERO]

I do believe that is one of the best things with the Abyss, being able to rotate the pads for a good seal.

 

[x838nwy]

What's "group delay" btw?

 

[eric65]

Responses abyss: yellow = pad pressed against measurement frame, blue = less sealed​, pink = even less sealed

 

The bass response (70-80 Hz) of the Abyss objectively varies by nearly 10 dB, depending on the seal or not of the pads.


According to several people who have tested the Abyss, sealing pads seems impossible.


Can we talk of true fidelity about this headphone with a response bass heightened by nearly 10 dB if sealing pads impossible ?
(as if we were to increase the bass of +10 dB with the bass potentiometer of an Hi-Fi amplifier to have more spectacular bass).

 

   
That's interesting. I've had half a dozen people come over to my place and from what I am aware, none of them had a problem with getting a good seal, if they so wished. It was simply a matter of rotating the cushions so the thickest part rested on the nape of the neck instead of directly behind the ears.
 
Moderate seal:

 
Good seal:

 
Hello purrin,
 
An obvious point, the measurements made by the French forumeur differ quite significantly from yours on the frequency response (function of the sealing pad) (ie you can find: bass roll off with Lack of seal:-5db at 20Hz, the French found: bass roll off with Lack of seal: -20 at -30 dB at 20Hz with an +10dB peak at 70-80 Hz with a lack of sealing pads) .
 
Also the same difficulty on the "group delay" substantially different according to the studied headphone
Unless it is related to the headphone model tested ? (series model in France)
 
On the issue of the Abyss port for a perfect seal, even turning the pads, you acknowledge yourself that there may be a real problem for some people (depending on the anatomy of the head) making sealing sometimes impossible.
 
You said : “ I didn't have a problem with seal with the thick part of the pad toward the nape of my neck, though I can definitely see it as an issue with some people. With the above tweaks in place, I do prefer to run the Abyss fully sealed.

 

Even if the seal is not perfect, you only lose performance (in terms of distortion) below ~35Hz where there's not a lot of energy in most recordings to begin with. The more notable aspect is that there's no serious bass roll off with lack of seal: -5db at 20Hz in relation to 500Hz; and no change / 0db at 25Hz in relation to 500Hz. Lose the seal on a STAX, and the already light hitting sub-low bass becomes almost non-existent, and the mid-bass hump gets a significant hump around 120Hz. “
http://www.head-fi.org/t/694483/electrodynamic-and-electrostatic-headphones-now-obsolete/60#post_10067705
 
 
Pierre Paya (Shiva on Head-Fi), Abyss dealer in France said; Superfred21 (Member Head-Fi) also admits, and Arnaud also (about the sealing pad problem, at least in some people).
 
Links concerning recognized sealing pad problem.
http://www.homecinema-fr.com/forum/post177890488.html#p177890488
"Abyss was used without sealing total pad / ear simply because it is not possible.
Even plating headphones against the skull, the seal is not complete.
We found an intermediate setting between a fully open headphones and maximum contact with high temples and jaw.
In this case, the sealing was nonexistent."
 
http://www.homecinema-fr.com/forum/post177891911.html#p177891911
"The adjustment on the head is at first not very easy, there is no real seal between the pad and the pinna, the made ​​to rotate pads ovoid allows the slightly adjust the height of the headphone."

 

[nk77]

   
That's interesting. I've had half a dozen people come over to my place and from what I am aware, none of them had a problem with getting a good seal, if they so wished. It was simply a matter of rotating the cushions so the thickest part rested on the nape of the neck instead of directly behind the ears.

 
Seems with a good seal the bass smoothes out flatter along with very slight increase in treble energy around the 8-9khz region. But what could be the reason for -5dB decrease in FR Mag with a good seal between the 5-6kHz mark? Is something being cancelled out?

 

[x838nwy]

That group delay looks funky. Here is mine:

Okay - found out what it means. Still looks strangely bad for the abyss, no?

 

[eric65]

In view of these measures is it possible to find the "secret" of the Abyss to explain his great listening pleasure compared to other headphones ?
 
The lack sealing of pads of Abyss and the resonance frequency particulary low in free air (70-80 Hz) explain perhaps the dynamism and the presence of these headphone in the low (middle bass) giving plenty of rythm and life at these headphone and also influence the sound stage (which depends in part on the bass reproduction).
 
Distortions noted at different frequencies (20 Hz, 300-400 Hz, 6-10 kHz), higher than other headphones (HE-500 and SR-009) could cause listening pleasure "artificial" by harmonics created, where some euphony (like some tube amplifiers that illuminate the sound with superimposed harmonics (due to distortions)).
 
In short, the Abyss give a very pleasant sound, fine, vibrant and dynamic, but not very faithful because of its relatively high distortions (compared to other headphones) and extra bass quite advantageous frequency (at 70-80 Hz).
 
In contrast, one can consider the Stax SR-009 headphone (just as fast and accurate as Abyss) as more faithful to the strict meaning of the Hi-Fi (less distortion, more linear frequency response), even if it is the apparent price of a renderer in the middle bass less dynamic and present, and euphony (listening pleasure) less.

 

[purrin]

Okay - found out what it means. Still looks strangely bad for the abyss, no?

 
I purposely provided a graph which was more of an apples to apples comparison. The question really is: how much group delay (in terms of ms) can one hear to notice a difference? So in that respect one needs to be very careful in the presentation of data. Also, there's a little bit of roughness in the group delay measurements which are of no consequence, so smoothing the graph a little bit doesn't hurt. "Consensus" among speaker builders seems to indicate that the audible threshold is about 25-30ms. Even then, there is a lot of debate concerning this. So in that respect, the graph should be visualized more like this: 
 

 
Finally, one needs to understand what context the group delay measurement is used in by speaker designers: bass reflex boxes and crossover design. For open headphones, this measurement doesn't provide anything meaningful which can be correlated with subjective experience. If the measurement has little or no correlation with subjective experience, then it's useless. In such cases, to present it as a differentiator (and in such a way as to intentionally or unintentionally highlight differences) only adds to confusion of the uninitiated. Because people who are unfamiliar with interpreting measurements have a tendency of makings all sorts of stuff up.

 

[purrin]

 

An obvious point, the measurements made by the French forumeur differ quite significantly from yours on the frequency response (function of the sealing pad) (ie you can find: bass roll off with Lack of seal:-5db at 20Hz, the French found: bass roll off with Lack of seal: -20 at -30 dB at 20Hz with an +10dB peak at 70-80 Hz with a lack of sealing pads) .

 
Yes. Probably a result of different methods. Getting a good seal with headphone measurements is extremely difficult and something both Tyll and I have struggled with, but have managed to work out with "tricks" over time.
 
My prior version 1 measurement rig had issues with obtaining a good seal with the STAXen. I've since built an improved version 2 rig which does a much more consistent job on bass measurements. (FYI: The photos of my measurement rig in Jude's signature line are totally out of date and do not reflect the state of my current rig.) I am much happier with the newer measurement rig because the bass measurements correlates well with my subjective experience. This new rig also includes "hair" and "nape" simulator attachments to account for less than prefect seal. My "good seal" measurements only utilize the hair simulator.
 
Finally I would ask you this: even if a less than perfect seal, do you actually hear a +10db increase at 70-80Hz with the Abyss. I certainly don't. +10db at 70-80Hz would be extremely noticeable and annoying with most of the music I listen to. If the Abyss did truly exhibit such behavior, I would most certainly throw it away in the garbage can.
 

 
On the issue of the Abyss port for a perfect seal, even turning the pads, you acknowledge yourself that there may be a real problem for some people (depending on the anatomy of the head) making sealing sometimes impossible.
 
You said : “ I didn't have a problem with seal with the thick part of the pad toward the nape of my neck, though I can definitely see it as an issue with some people. With the above tweaks in place, I do prefer to run the Abyss fully sealed.

 
Because all humans beings are built slightly differently, in theory I can see some people not being able to get a good seal. In practice, this has not occurred with the half-dozen folks who have dropped by my house for a listen.

 

[purrin]

 
Seems with a good seal the bass smoothes out flatter along with very slight increase in treble energy around the 8-9khz region. But what could be the reason for -5dB decrease in FR Mag with a good seal between the 5-6kHz mark? Is something being cancelled out?

 
Probably - more effective wave cancellation. What's the full wavelength of 5500Hz? Around 6cm ~= inner diameter of ear cushions. 

 

[x838nwy]

   
I purposely provided a graph which was more of an apples to apples comparison. The question really is: how much group delay (in terms of ms) can one hear to notice a difference? So in that respect one needs to be very careful in the presentation of data. Also, there's a little bit of roughness in the group delay measurements which are of no consequence, so smoothing the graph a little bit doesn't hurt. "Consensus" among speaker builders seems to indicate that the audible threshold is about 25-30ms. Even then, there is a lot of debate concerning this. So in that respect, the graph should be visualized more like this: 
 

 
Finally, one needs to understand what context the group delay measurement is used in by speaker designers: bass reflex boxes and crossover design. For open headphones, this measurement doesn't provide anything meaningful which can be correlated with subjective experience. If the measurement has little or no correlation with subjective experience, then it's useless. In such cases, to present it as a differentiator (and in such a way as to intentionally or unintentionally highlight differences) only adds to confusion of the uninitiated. Because people who are unfamiliar with interpreting measurements have a tendency of makings all sorts of stuff up.

 
 
From what I read (which is a great deal less than you have admittedly), this really isn't just related to bass reflex boxes and cross-overs but also other things including positioning of speakers/transducers. So I take it to mean that (in the case of headphones) you throw a signal which consists of >=2 frequencies with a known phase relationship between the two at a driver, then look at what comes out?
 
I am wondering how the system gets this result? Does it just send out an impulse input, fft it and look at the phase against freq? I recall an earlier plot where the abyss at about 350kHz kindda went 'advance' with -ve group delay which is a little weird and looks like some kind of resonance was crossed or something which is even stranger as it doesn't have a cross-over to do that sort of thing. I admit i'm no expert (did study vibrations and control back in my years, but not audio specific), but saying tens of milliseconds is inaudible is perhaps too dismissive? Don't get me wrong, all I'm saying is that many, many experienced ears have not mentioned much wrong with the Abyss so a glaringly poor result as the above must have manifested itself to someone - either that or there's something not quite accurate about the measurement/processing of the data. May be it's to do with the fact that the drivers are much closer to our ears than with speakers? May be the group delay results is an analysis of a square wave input? This would be an odd choice cos you'd need lots of square wave frequencies to get an accurate result?
 
I don't really know to be honest, I've never had the equipment/computing power you guys have these days so I'm not sure what's currently possible/done. But I am curious about the fact that a seemingly relevant measurement does not show itself to be audible. More importantly, the bass is an area which nearly all says the abyss excels yet seem to come out worst on these plots...

 

[arnaud]

Finally, one needs to understand what context the group delay measurement is used in by speaker designers: bass reflex boxes and crossover design. For open headphones, this measurement doesn't provide anything meaningful which can be correlated with subjective experience. If the measurement has little or no correlation with subjective experience, then it's useless. In such cases, to present it as a differentiator (and in such a way as to intentionally or unintentionally highlight differences) only adds to confusion of the uninitiated. Because people who are unfamiliar with interpreting measurements have a tendency of makings all sorts of stuff up.

 
I agree that the concept of group delay is a little confusing when applied to headphones. I used to look into this for crossovers where significant delay can occur with steep filters else enclosure LF tuning but was not aware people looked into this for headphones. After seeing the post from Agnostic1er on HCFR, I looked it up a bit and it turns out we can see a diaphragm resonance (1 guess then 2nd concentric mode) at 4-500Hz (visible in CSD from golden ears as well as the FRFs from the various sites).
 
Now, as you said, is this very audible, I don't know. What I observe is that the abyss (with proper seal) has similar bass / low mid-bass rendering as the Stax 007A / 007mk1 and, subjectively, although the SR009 bass is more "linear", the omega 2 certainly feels like it has more bass quantity, and rounder sounding too.
 
As for the seal, someone told me on this board I made the mistake to try to enforce the seal, hence my poor appreciation of the abyss when I tried it. Then, I was surprised by your comment later on how the abyss was rather insensitive to sealing (as opposed to staxens where lack of seal is very unpleasant with bass hump in 150Hz range). When I saw the curves on HCFR, it then made sense to me: it's not that the abyss is insensitive to sealing, it's just that the lack of seal (happened to me but I also may not have rotated the pads properly) is not as unpleasant as for stases because the free air resonance is at ~80Hz and that actually makes a nice physiological filter for low level listener (or bass heads, whichever camp you are lol ).
 
Arnaud

 

[arnaud]

  From what I read (which is a great deal less than you have admittedly), this really isn't just related to bass reflex boxes and cross-overs but also other things including positioning of speakers/transducers. So I take it to mean that (in the case of headphones) you throw a signal which consists of >=2 frequencies with a known phase relationship between the two at a driver, then look at what comes out?
 
I am wondering how the system gets this result? Does it just send out an impulse input, fft it and look at the phase against freq? I recall an earlier plot where the abyss at about 350kHz kindda went 'advance' with -ve group delay which is a little weird and looks like some kind of resonance was crossed or something which is even stranger as it doesn't have a cross-over to do that sort of thing. I admit i'm no expert (did study vibrations and control back in my years, but not audio specific), but saying tens of milliseconds is inaudible is perhaps too dismissive? Don't get me wrong, all I'm saying is that many, many experienced ears have not mentioned much wrong with the Abyss so a glaringly poor result as the above must have manifested itself to someone - either that or there's something not quite accurate about the measurement/processing of the data. May be it's to do with the fact that the drivers are much closer to our ears than with speakers? May be the group delay results is an analysis of a square wave input? This would be an odd choice cos you'd need lots of square wave frequencies to get an accurate result?
 
I don't really know to be honest, I've never had the equipment/computing power you guys have these days so I'm not sure what's currently possible/done. But I am curious about the fact that a seemingly relevant measurement does not show itself to be audible. More importantly, the bass is an area which nearly all says the abyss excels yet seem to come out worst on these plots...

 
Vibration and controls, I had such hard time getting a handle on state space domain, never understood it I think
 
My guess on the group delay was simply looking at the delay in cross-correlation function between input signal and measured SPL for some kind of band limited signal. Maybe arta documentation says more?
 
Arnaud

 

[x838nwy]

   
Vibration and controls, I had such hard time getting a handle on state space domain, never understood it I think
 
My guess on the group delay was simply looking at the delay in cross-correlation function between input signal and measured SPL for some kind of band limited signal. Maybe arta documentation says more?
 
Arnaud

 
The creepy/weird thing is, a post-grad I worked with was a French dude called Arnaud.

 

[arnaud]

   
The creepy/weird thing is, a post-grad I worked with was a French dude called Arnaud.

 
Eh! I was at VTech, did MS in Vibration & Acoustics / Active Noise Control of PC fan noise. I know a few french people who also did controls but no other arnaud

 

[x838nwy]

   
Eh! I was at VTech, did MS in Vibration & Acoustics / Active Noise Control of PC fan noise. I know a few french people who also did controls but no other arnaud

 
Lolz this could have turned out really weird. I read Engineering (MEng) at Cambridge. I worked on semi-active roll control of oil tanker trailers for my 3rd year project with Arnaud. I don't know if it ever went further than the drawing board cos the control strategy was a bit problematic - trying to work out what your average trucker is about to do is a lot harder than with cars, it seemed.
 
Anyway, if my PC fan gets too loud, I'll shoot you a PM.