A true (pure) delay, meaning no waveform alteration except a time shift, should be defined by the leading edge of the waveform.
Many would say that the 50% amplitude point is where to measure it, but that doesn't account for the "integrator delay"
due to a finite system bandwidth.
Last edited:
John, I remember having this discussion years ago with some other people. It is an issue I have also been wrestling with. The illustration I used, for what it's worth, the example of a cap being charged by a square wave current.
What is the 'real' delay? For me that is zero (1st approximation).
Why? At any time, any change in current immediately (1st approximation) results in a change in voltage.
See attachment. RED is the current into the cap switching beween 0 and 1mA; GREEN is the resultant voltage across the cap. When the current is zero, cap voltage (of course) remains constant. When the current steps up, IMMEDIATELY the voltage on the cap starts to change.
You can clearly see that at any time that the current into the cap changes level, IMMEDIATELY the voltage across the cap starts to change - so there is NO 'real' delay.
Jan
Last edited:
If you listen to "Drive my car" (Rubber Soul, the Beatles), you can get a good idea of what is a REAL delay. 50 years long (since 1965, when this hit was produced).
At this time, we were using tape recorders as delay in studio, varying the length of the tape between the recording head and the playback one, or the speed of the dedicated machine. I remember to had used 8 tracks car cassettes glued on Revox to get rid of the need to rewind the magnetic tape endlessly.
Last edited:
Perhaps by 'REAL delay' you mean 'latency' which has an audio meaning as time delay associated with otherwise unaltered full spectrum programme material ?
http://en.wikipedia.org/wiki/Latency_(audio)
Group delay applies instead to time shift of a short pulse of a fixed frequency and envelope shape, and its effect becomes 'real' when the peak amplitude of the pulse becomes advanced or delayed in time by an amount known as 'group delay'. In audio terms, a shift in the arrival time of the peak of a short pulse of a fixed frequency. That is real enough, and is equally applicable at high frequency as low. That it is numerically smaller for hf is just an artefact of the way it's defined, and not necessarily its audibility.
As I said, Audax was generally used, hardly anyone mentioned the full name. I have no idea when it became Son Audax. It's possible the name stuck after a subsidiary, just to designate its location or market. I don't know.
Christophe, if memory serves, Audax was a very old company, dating back to the late 20ies, about on a par with Britain's Celestion (also knows as Rola-Celestion, not that anybody called it so, it was just Celestion). At some point in time, abouth half of Britain's small volume brands used their silk dome tweeter as a must, sometimes also the mid-bass units. I still have (somewhere) their general catalog of the units most often used, the big and really good units, such as my bass driver, were always shown on separate sheets. I knew about them only because the friend I made the speaker with was their general representative for Serbia and had just about everything on file. Dealer price then was €280, which would be about €410 today assuming a 3% annual inflation.
What is not geerally known, is that at the time, JBL speakers for Europe and Middle East were made in Denmark, and used Auday drivers throughout. Rebadged, of course, by a sticker over their Audax designations. My own Ti600 model is made like that. JBL only kept their professional series made in USA. It also has a titanium dome, but that model cost some €40 at the time.
When Harman International shifted speaker manufacturing away from Denmark (don't know where), they put a lock on Audax gates in 2006.
Yes - for example transverse vibration of cantilevers in vinyl cartridges. Cantilevers flex in normal playback, displacement of the stylus results in a flex which propagates along the cantilever, following laws of beam mechanics. Often the 1/4 or 1/2 wavelength resonance is in the hf audioband, devised to mesh with the cartridge/loading LCR resonance and provide hf extension. What's more, such propagation is dispersive, ie travelling velocity depends on frequency, and is not min phase. The idea that a cantilever is rigid and dynamically moves as one body isn't so, few structures are rigid in dynamics. Cantilevers make interesting candidates for study because all the classic maths is readily available, and they are necessarily light and so rigidity is compromised.
I still don't see the difference in what it means for audibility with phase shift. If you have phase shift, with for instance a sine wave, the top of a phase-shifted sine wave comes before or after the top of the original wave.
But still we would not call that a 'real' delay, would we?
Jan
Well yes, it is a real delay for a sinusoid waveform. If you observe the original and the phase shifted signals, they are identical except for a time shift. Only for a sinusoid.
Group delay is a cousin, has enough of the same genes to be similar, but is nevertheless different. To become 'real' in audio, it alters the pulse shape of a short AM 'pulse' of a sinusoid, such that the arrival time of the peak of the AM envelope is shifted forward or backward in time.
Phase response and group delay are close relatives, close enough to match DNA and obtain one from the other in most cases.
George, as always you perplex me with you ability to locate sources on pretty much anything discussed on this site. In this case, however, I doubt the validity of the assertions made by Daniel Queen. For those interested, this is the link to the patent George mentioned Patent Images
For those who lost track of the discussion, it is about the speed with which the rim of a loudspeaker cone reacts to movements of the voice coil. My assertion was that this is determined by the speed of sound in the cone material, Luckythedog introduced the concept of flexing waves as the coupling mechanism. And then George came up with this patent, which indeed shows two formulae which are relevent to the propagation of waves through a solid under different circumstances. First a comment to this, and then to the patent quoted by George in a broader sense.
1) the question of the propagation speed of the acceleration of the voice through the cone boils down to: is it a plate, or a solid. In both cases it will be determined by the speed of sound of the specific material, but in a plate with a bending wave as the mechanism, it is considerably slower.
One thing is certain, a cone is not a plate. A plate is not constrained in one dimension. However, a cone is curved, often compounded, and thereby constrained in all directions. Obviously, a cone is not a solid either. However, the aim is to approach as closely as possible behaviour as it it were. Cone design is all about achieving pistonic behaviour, that is, creating a shape/material combination which allows the cone to react to movements of the voice coil without deformation within the desired frequency range.
Again, measurements prove that the cone rim moves faster than it takes sound emanating from the dust cap to reach the cone rim.
2) As to the patent, I am at a loss. It describes a problem I am not aware of and then goes on to producing a solution to it. The problem, apparently, is lobing of a single driver. I have not seen this in measurements. Sound pressure radiated by a pistonic driver as a function of measuring angle typically displays a relatively smooth transition from high to low without clear evidence of directional lobes.
There are drivers on the market now which use bending wave propagation NXT unveils latest technology - 15 December 2005 - Daily Online News - LSi Online. It is quite obvious that this cone is not a cone but a plate.
Last edited:
This delay story make me remember a machine i invented around 1980.
It was called "Magic box" and used to make perfect combos of records for radio programs.
The machine was using 2 delays and a command to power on a turntable. One to fire the play function of a Revox, one to fire the record function. And a single push button to send the 3 commands.
We used-it this way. When a first tune was recorded on the tape, we marked the exact place on the tape with a white pencil where we wanted the second tune to be inserted. then, drive this mark back in front of a mark at the left of the record head on the Revox. Then find the place on the vinyl on the turntable, where we wanted the second tune exactly to start, and made a full back turn of the record on the turn table. Pushing the button, and the turn table was powered on. After a first delay, the Revox started to play, then to record at the exact moment when the turntable had made his full turn, and the two tunes were 'magically' chained on the tape without scissors with a smooth "drop in".
http://www.esperado.fr/fr/mambots/content/dewplayer.swf?son=/fr/images/stories/musiques/cld2.mp3
It was called "Magic box" and used to make perfect combos of records for radio programs.
The machine was using 2 delays and a command to power on a turntable. One to fire the play function of a Revox, one to fire the record function. And a single push button to send the 3 commands.
We used-it this way. When a first tune was recorded on the tape, we marked the exact place on the tape with a white pencil where we wanted the second tune to be inserted. then, drive this mark back in front of a mark at the left of the record head on the Revox. Then find the place on the vinyl on the turntable, where we wanted the second tune exactly to start, and made a full back turn of the record on the turn table. Pushing the button, and the turn table was powered on. After a first delay, the Revox started to play, then to record at the exact moment when the turntable had made his full turn, and the two tunes were 'magically' chained on the tape without scissors with a smooth "drop in".
http://www.esperado.fr/fr/mambots/content/dewplayer.swf?son=/fr/images/stories/musiques/cld2.mp3
Last edited:
See that is my problem. I can't see the delay, because as I noted, at any time you make a change in the original, the effect is seen IMMEDIATELY in the phase shifted wave. So how can there then be a delay?
Edit: maybe you wanted to say: phase shift is NOT a delay but for a sine wave, because the two have identical shape, it APPEARS AS IF there is a delay.
Jan
Last edited:
or at its funniest ?automation at its finest !
>
Long time after, I was asked by Akaï to participate to the conception and development of their DD1500 recording and editing 'virtual' machine on magneto-optical and hard disk medias
Last edited:
'But for' meaning 'except for'. Sort of, except personally I wouldn't actually say that because strictly phase only has a meaning for sinusoids anyway. For all purposes the time shift is then real, even philosophically !
Check out Scott's post #69529, and the link from B&W which demonstrates and discusses cone flex and wave propagation with the cone. Some neat videos of simulations.
The B&W link is here: Tools of the trade - laser measurements | Society of Sound
You are mistaken on this matter, vacuphile.
- Status
- Not open for further replies.