The Beyerdynamic DT48 Arrives...

[shamu144]

Beyerdynamic just increased the MRSP of the DT48, at least here in Europe, where currency exchange will not come into consideration.
 
From 279 euros previously on Thomann, up to 319 euros. That is a 15% price increase, with a very unfortunate timing due to severe economic recesion hitting Europe. There is no official comunication yet from Beyer, but I asked some details.
 
I have the feeling however this could be a marketing tactical move, to differentiate the DT48 from the DT1350 (MRSP 269 euros in Europe). It also makes very clear that the DT1350 is no substitue for the DT48. Thank Godness! Price gap in Europe between both got larger, and sincerely, that makes a lot of sense. Beyer feel strong enough about the DT48 to make such a bold move. To me, it says a lot of the value of the DT48. Of course, I would have prefered to see a [much] lower MRSP for the DT1350, but that is a different story.

 

[joelpearce]

Or else they are sick of manufacturing them and are now trying to push people to the new Tesla line instead.

 

[shamu144]

Mmm...I wouldn't bet on that one: the DT1350 (portable, on the go, studio, musicians and DJ),  and DT48 (full size, ENG/EFP, field recordist, sound film recordist, lab, audiometry)  are targeted at 2 completely different markets as stated by Beyer themselves.

 

[Deep Funk]

If so that explaines a lot... 

 

[shamu144]

In my opinion, and of course I am a little biased towards the DT48

, the DT1350 don't hold a candle when compared to its grand father.
 
The tonal balance of DT1350 is overall nice, with fantastic and remarkably accurate bass (that I can't say about the DT48

 ). Highs intensity is spot on, but not their timbres. In fact, the DT1350 suffers to my ears from similar weaknesses as the T1, with what appears to be glossy and shiny metallic highs, and a slightly dry midrange (not as bad as T1 though). Timbres are really inaccurate (or colored) and not convincing, with what seem to be a lack of micro dynamic (miss texture and real low level detail in articulations, sense of air). OTOH, they handle macro dynamic very well and have a good instrument separation. Pity that the soundstage is so collapsed and 2D, flat. I really miss air around instruments. Overall, it tends to be a very analytical listen, one that I feel is perfect for monitoring frequency unbalance in a mix, but completely unatural for the ears and overall not engaging for casual music listening...

 

[shamu144]

Following up on this frequency vs time domain separation, and willing to dig the question a little further, I inevitably end up reading on the Theory of signals... This is way too complex and technical for me, but there are some great learnings I feel, and they talk about measurments that we very often overlook.
 
Basically, all signals - please correct me if I am wrong this is rather advanced abstract theory for us mere mortals - can be decomposed of its frequency response (amplitude domain) and phase response (time domain)... And then comes Fourier transform which tells us that any impulse response (i.e. transient response) can be calculated (not easy) from both the frequency response and phase response.
 
As I understand it, proper phase response here is key for the linearity in the time domain of the transducers. Phase response need not to be perfectly flat, but rather perfectly smooth (no bump), so that Group delay (for a group of frequencies) is indeed a constant. Phase response is usually measured in degrees.
 
One thing is certain: phase response is at least equally important as frequency response to understand the nature of a transducer.
 
Well, I don't know if it is a coincidence or not, but the DT48E phase response when driven by a 120 ohms output impedance as measured by Tyll at InnerFidelity is.... perfectly flat at 0º, only matched by the best orthos out there (LCD2, HE500). Not even T1 or HD800 come close to the phase response of the DT48. I have always felt that the DT48 has outstanding transient response and is able to follow accurately amplitude changes in the signal, giving it a very linear response in the time domain and it's absence of colorations. Could it be the reason why... I don't have the answer, but thought this was an interesting question to share...

 

[Soaa-]

I've been studying the Fourier transform along with other frequency-domain analysis techniques for the past while for a personal project of mine.
 
I don't think the phase response graphs fro Tyll represent the acoustic phase response. I think they have to do with the electrical properties, but I'm not sure what. It may have to do with headphone-amplifier interactions.
 
A certain Japanese measurements site provides acoustic phase measurement graphs, and they look quite different from the ones from InnerFidelity.
 
Here's what I know. Phase response is not perceived by the human ear for steady tones. Thus, phase response will not affect timbre. What's more, my impression is that the physics of driver movement would override any phase distortion in the event of a transient in the signal, and an impulse response measurement is more suited to determine the transient response of a driver. That's because phase response is a property in the frequency domain, and the frequency domain is an accurate model only for steady signals, and transients are time domain signals by nature.
 
In the case of the DT 48, the impulse response shows a good attack, but a long decay, which would contribute to the audibility of transients and details, but would be, in the end, an inaccurate representation of the signal. That means what we perceive as detail on the DT 48 is in fact enhanced by the headphones themselves, and is thus a coloration of the signal.

 

[shamu144]

Now we are getting serious

Mmm, why don't you want to call a cat a cat. I haven't heard of any other definition of the phase response for a transducer and I indeed asume Tyll measurment of the phase response are just that, phase response. If you have seen different phase response measurments for the DT48, please share them.
 
Quote:

 
I don't think the phase response graphs fro Tyll represent the acoustic phase response. I think they have to do with the electrical properties, but I'm not sure what. It may have to do with headphone-amplifier interactions.
 
A certain Japanese measurements site provides acoustic phase measurement graphs, and they look quite different from the ones from InnerFidelity.

 
 
I think this statement is not completely accurate. The phase response is basically a time domain measurement. It is represented as a function of the frequency, but correspond to the amount of time delay (measured in degrees but can be traduced to seconds) between the signal input and output for a pure sin wave (steady state, fixed frequency).
 
Imagine that lows have a time delay of 1 hour (exagerated of course) as compared to mid/highs... Well, that would definitely affect the presentation of the musical signal 

... Now what is the threesold of audibility in phase response distorsion in another very interesting question.
 
 

 
Quote:

 
That's because phase response is a property in the frequency domain.

 
 
There are three ways apparently to evaluate the transient response of a drivers: impulse response, square wave, tone burst.
 
But, the impulse response does seem to present some issues as mentionned as follows: "The problem with testing using impulses is that the amount of energy that you can pack into an impulse of a height that is small enough to be processed with no clipping or other nonlienar distoriton is limited. Because the energy is limited, there is a problem with noise in the measured response. " Is transient response a problem with headphones [Hydrogenaudio].
 
Quote:

 
an impulse response measurement is more suited to determine the transient response of a driver.

 
 
More on the subject here [hydrogenaudio]
 
"First, note a fundamental property of linear time-invariant systems (like headphones neglecting THD): the impulse response and the complex transfer function are interchangeable, between the time and frequency domains. And the transfer function can be decomposed into a magnitude function (the usual "frequency response" measured in dB) and a phase function (measured in degrees).
 
The "ideal" phase response is zero degrees at all frequencies, and non-ideal phase response can be analyzed in rather clear and understandable ways. It's a lot harder to compare non-ideal transient responses than to compare phase responses."

So it seems to me that some caution should apply when looking at the impulse response or square waves graphs as represented in Tyll's measurments and before drawing any quick conclusions. Actually, a combination of the frequency response and phase response seem to be a more reliable method to judge the transient behaviour of a transducer.
 
Quote:

 
In the case of the DT 48, the impulse response shows a good attack, but a long decay, which would contribute to the audibility of transients and details, but would be, in the end, an inaccurate representation of the signal. That means what we perceive as detail on the DT 48 is in fact enhanced by the headphones themselves, and is thus a coloration of the signal.

 

[shamu144]

A very interesting thread discussed on those boards a few years ago helping to understand the influence of the phase response for proper transient response, and how they correlate to the frequency response in amplitude: Leading edge transient response.
 
Edit: crucial finding as well in this paper from the AES by Lipchitz, On The Audibility of Midrange Phase Distorsion in Audio Systems, where they state the following:
 
2) Audibility [of Midrange Phase Distorsion] is far greater on headphones than on loudspeakers.... To be reflected...

 

[Deep Funk]

Would the slower decay be a clue as to why effects and echoes some times seem to be more audible? The CD900 ST had the same thing for effects and echoes. 

 

[Soaa-]

shamu, you're misunderstanding phase distortion. Phase distortion is a purely a frequency domain property that can, for any particular frequency, range from 0 to 2pi. It's not a time domain delay, but an angular offset in the reproduction of a sine wave. In fact, phase distortion is a property that applies only accurately to infinite length sine waves, and has little effect on the fast-changing nature of music.
 
Consider a sine wave of 440Hz. Each cycle (therefore, 1 turn, or 2pi) of the wave lasts 1/440 of a second, and the cycles repeat forever. Phase distortion is simply the difference in angle between the input signal and the reproduced signal. A delay of 1/440 of a second (one turn) becomes indistinguishable from the original signal. Because the cycles repeat forever, it's not possible to know, nor is it even relevant to know the number of turns a reproduced signal has been delayed. The only relevant (and measurable) data is the offset as a fraction of a turn. Keep in mind that frequency-domain properties always assume the ideal infinite length signal.
 
Now, regarding transients, their nature is that of a very short time-domain signal. Sure, there's some correlation between the frequency response, phase response and transient characteristics, but they are not necessarily representative of each other. Let me reiterate: transients are time-domain signals with such fast changing frequency characteristics that frequency domain properties of headphones no longer apply.

 

[shamu144]

I am enjoying this good discussion

 
I understand what you say Soaa, and you are correct that for a given fixed and infinite frequency (now that is a boring signal), a phase shift of 2pi or 360º in the sin wave will not be audible. I think this is because we are using too similar concept that are correlated: Phase shift and Delay. While phase shift is expressed in degree (angular offset you mentionned), delay is expressed in time. They are the same thing but expressed in different units.
 
A uniform phase shift of 90º across the full frequency spectrum can perfectly be traduced in a time delay for each fixed frequency (you know the wavelength for each frequency and speed of sound, easy to calculate). But of course, the time delay resulting will vary for each frequency, which will introduce severe and uneven distorsions in the original complex waveform. This is where enters the notion of constant Group Delay needed over the frequency range.
 
From the Wiki definition:
 
Group delay is a measure of the time delay of the amplitude envelopes of the various sinusoidal components of a signal through a device under test, and is a function of the frequency of each component. Phase delay is a similar measure of the time delay of the phase, instead of the delay of the amplitude envelope, of each sinusoidal component.

More information here: Time, Phase, Frequency and Delay
 
This is why I understand a flat (meaning 0º phase shift) phase response accross the frequency spectrum is ideal as it will not introduce time distorsion in the original complex waveform. And apparently, the DT48 (and orthos) do that and accurately reproduce the original signal with high fidelity in the time domain.
 
Quote:

shamu, you're misunderstanding phase distortion. Phase distortion is a purely a frequency domain property that can, for any particular frequency, range from 0 to 2pi. It's not a time domain delay, but an angular offset in the reproduction of a sine wave. In fact, phase distortion is a property that applies only accurately to infinite length sine waves, and has little effect on the fast-changing nature of music.
 
Consider a sine wave of 440Hz. Each cycle (therefore, 1 turn, or 2pi) of the wave lasts 1/440 of a second, and the cycles repeat forever. Phase distortion is simply the difference in angle between the input signal and the reproduced signal. A delay of 1/440 of a second (one turn) becomes indistinguishable from the original signal. Because the cycles repeat forever, it's not possible to know, nor is it even relevant to know the number of turns a reproduced signal has been delayed. The only relevant (and measurable) data is the offset as a fraction of a turn. Keep in mind that frequency-domain properties always assume the ideal infinite length signal. 

 
I don't think it is correct to asume that the nature of transient response is a time domain only signal, but rather a combination of time domain (phase response) and amplitude domain (frequency response)... From this post on Hydrogenaudio: the impulse response and the complex transfer function are interchangeable, between the time and frequency domains.
 
Quote:

Now, regarding transients, their nature is that of a very short time-domain signal. 

 
 

 

[Deep Funk]

DT48S in, I did not expect this headphone to sound as it does. This one is a keeper.

 

[joelpearce]

Quote:

DT48S in, I did not expect this headphone to sound as it does. This one is a keeper.

How does it differ?  That's interesting...
 

 

[shamu144]

Congrats! What year / cushion / headband style ?
 
Quote:

DT48S in, I did not expect this headphone to sound as it does. This one is a keeper.