It's completely lost on me as well. I was pretty sure that we understood how sound waves propagate - it hasn't been a mystery for along time.
I suppose something that oscillates isn't going to have any difference in inertia with a changing signal. Maybe there is some rationale for a powerful magnet and underhung voice coil, but idk.
Could something vibrate at a certain frequency but also accelerate and decelerate faster, iow complete the same number of cycles per second but have a different velocity at certain points?
It sounds like for the most part the research shows that everything revolves around keeping and maintaining a near flat FR as much as possible. I would imagine even energy from the cabinet reflecting through the cone could also change FR, and this effect might change with real music.
Still its hard to explain why speakers sound so different from each other.
Last edited:
Imagine sine waves drawn with dotted lines Vs solid lines...
Sorry guys,
My description lacks clarity... I am trying to present an alternative (more accurate?) way to look at sound generation and therefore an alternative to how we design loudspeaker drivers... Hope this helps.
"Create" Vs "propagate".
If one accepts that all airborne sounds are simple point source APE's (Air Pressure Events) we can then break down every sound into a series of on / off pulses which radiate out in a 360 degree bubble.
A fleeting percussive drum strike Vs a gently bowed Cello with rich decay... They are both APE's, only the duration and the level of the air pressure change vary.
Analogy - A single shot (pulse) fired from a machine gun Vs a continuous stream of 100 bullets (wave) fired from the same gun ... Both are composed of identical sonic events but the 100 bullet event simply has a longer time duration.
Does this matter and if it does, how does this affect loudspeaker driver design?
Sorry guys,
My description lacks clarity... I am trying to present an alternative (more accurate?) way to look at sound generation and therefore an alternative to how we design loudspeaker drivers... Hope this helps.
"Create" Vs "propagate".
If one accepts that all airborne sounds are simple point source APE's (Air Pressure Events) we can then break down every sound into a series of on / off pulses which radiate out in a 360 degree bubble.
A fleeting percussive drum strike Vs a gently bowed Cello with rich decay... They are both APE's, only the duration and the level of the air pressure change vary.
Analogy - A single shot (pulse) fired from a machine gun Vs a continuous stream of 100 bullets (wave) fired from the same gun ... Both are composed of identical sonic events but the 100 bullet event simply has a longer time duration.
Does this matter and if it does, how does this affect loudspeaker driver design?
It's still an analogue wave, what you are saying implies digital to me
Assume perfect cone /dome rigidity with zero break up...
YES!
Lets get down and dirty with specific driver design details...
In order to focus on the driver motor & suspension lets assume we have designed a perfectly rigid cone or dome with no measurable cone break up / bending modes or cone flexing within the 20Hz to 20KHz band.
Now with the perfect cone all we need to do is ensure that the driver motor and suspension have perfect control over our cone.... This is the key point and where the time domain is vital.
By definition we must design a mechanical system which can reproduce the rise time and decay of the electrical signal input (music or any recorded sound).
The problem is our electrical input is composed of a series of millions of individual impulses (recordings of sounds) with rise times approaching instantaneous and decays in the low Micro second bracket...
For the moment, lets assume our loudspeaker driver motor / suspension can approach the rise time of the input ( a BIG assumption!) and lets focus on the (energy) decay time of current drivers.
Using a quality reference, an ATC dome midrange, and staying within its bandwidth limits of approx 400Hz to 4KHz it has a decay time (as shown in a CSD plot) of around 1 to 5 milli seconds depending on frequency and SPL .... Many orders of magnitude slower than the input!!
This means that our loudspeaker driver is incapable of reproducing a single impulse accurately.... In fact it adds gross distortion in the form of "ghost echos" as the cone / dome bounces around on its suspension.
Vitally, the ratio of the time spent bouncing around on the suspension (decay time) after every single impulse is many times greater than the impulse duration.... Post ringing to dwarf the worst DSP artifacts!!
And this is using a state of the art ATC dome driver as our reference....
Drivers with rubber surrounds are even worse, and as the energy required increases 8 fold (following the 8 fold driver excursion curve) for each octave lower in frequency the problem gets significantly worse in the low midrange and bass.
YES!
Lets get down and dirty with specific driver design details...
In order to focus on the driver motor & suspension lets assume we have designed a perfectly rigid cone or dome with no measurable cone break up / bending modes or cone flexing within the 20Hz to 20KHz band.
Now with the perfect cone all we need to do is ensure that the driver motor and suspension have perfect control over our cone.... This is the key point and where the time domain is vital.
By definition we must design a mechanical system which can reproduce the rise time and decay of the electrical signal input (music or any recorded sound).
The problem is our electrical input is composed of a series of millions of individual impulses (recordings of sounds) with rise times approaching instantaneous and decays in the low Micro second bracket...
For the moment, lets assume our loudspeaker driver motor / suspension can approach the rise time of the input ( a BIG assumption!) and lets focus on the (energy) decay time of current drivers.
Using a quality reference, an ATC dome midrange, and staying within its bandwidth limits of approx 400Hz to 4KHz it has a decay time (as shown in a CSD plot) of around 1 to 5 milli seconds depending on frequency and SPL .... Many orders of magnitude slower than the input!!
This means that our loudspeaker driver is incapable of reproducing a single impulse accurately.... In fact it adds gross distortion in the form of "ghost echos" as the cone / dome bounces around on its suspension.
Vitally, the ratio of the time spent bouncing around on the suspension (decay time) after every single impulse is many times greater than the impulse duration.... Post ringing to dwarf the worst DSP artifacts!!
And this is using a state of the art ATC dome driver as our reference....
Drivers with rubber surrounds are even worse, and as the energy required increases 8 fold (following the 8 fold driver excursion curve) for each octave lower in frequency the problem gets significantly worse in the low midrange and bass.
I don't get what you are saying. The signal to the speaker is a continuously varying voltage of sufficient bandwidth to reproduce the audio frequencies.
Faster than what, a pure sine wave? Of course, every musical instrument does this, that's what gives them their timbre
"A continuously varying voltage" which varies, very quickly!
Exactly!
The electrical input varies (lowest point being noise floor of electronics, highest point being max SPL) and therefore in order for the driver (electro - mechanical converter) to convert the electrical input into sound it must track the input signal as accurately as possible.
In conclusion and to bring it right back bang on topic....
The drivers with the fastest energy decay have the lowest distortion... Voila!
PS
The typical levels of THD / IMD and even power compression ( thermal distortion) pale to insignificance compared the gross time domain distortion discussed above.
Exactly!
The electrical input varies (lowest point being noise floor of electronics, highest point being max SPL) and therefore in order for the driver (electro - mechanical converter) to convert the electrical input into sound it must track the input signal as accurately as possible.
In conclusion and to bring it right back bang on topic....
The drivers with the fastest energy decay have the lowest distortion... Voila!
PS
The typical levels of THD / IMD and even power compression ( thermal distortion) pale to insignificance compared the gross time domain distortion discussed above.
Alex - I'm sorry to have to say that your argument doesn't make much sense and I just wanted to make it known to others here that what you are saying is not really accepted science in the art. As we have shown, the decay is intimately linked to the frequency response and impulse response, so knowing those things and we know all, CSD, energy decay, the whole shibang included.
Guilding the lilly or lipstick on a t**d?!
Thanks Dr Geddes,
I have great respect for your work and your objective science based approach.... But I think you have missed my point!
I am not disputing the accepted and obvious commonality between energy decay (shown in CSD plots) and frequency response.
My point is that conventional loudspeaker drivers are fundamentally flawed due to excessive ringing ( I think you prefer the name ringing Vs energy decay?) and this flawed behaviour is most easily understood and visible in CSD plots.... But of course the same information can be extrapolated from the impulse response and frequency response.
So for the avoidance of doubt, I have never disputed the accepted and obvious commonality between energy decay (shown in CSD plots) and frequency response.
But I am saying there is a huge "elephant in the room" that is being ignored by driver designers when they view the information available from modern test equipment ie CSD / frequency / impulse / step response, laser imaging, FEA etc....
Getting all hung up on diamond coatings, exotic metal deposition, graphene, DSP / Eq crossovers (which I love!) etc... All lip stick on a great big steaming pile of**** which is fundamentally unfit for purpose.
Thanks Dr Geddes,
I have great respect for your work and your objective science based approach.... But I think you have missed my point!
I am not disputing the accepted and obvious commonality between energy decay (shown in CSD plots) and frequency response.
My point is that conventional loudspeaker drivers are fundamentally flawed due to excessive ringing ( I think you prefer the name ringing Vs energy decay?) and this flawed behaviour is most easily understood and visible in CSD plots.... But of course the same information can be extrapolated from the impulse response and frequency response.
So for the avoidance of doubt, I have never disputed the accepted and obvious commonality between energy decay (shown in CSD plots) and frequency response.
But I am saying there is a huge "elephant in the room" that is being ignored by driver designers when they view the information available from modern test equipment ie CSD / frequency / impulse / step response, laser imaging, FEA etc....
Getting all hung up on diamond coatings, exotic metal deposition, graphene, DSP / Eq crossovers (which I love!) etc... All lip stick on a great big steaming pile of**** which is fundamentally unfit for purpose.
I thought from the first post above that you understood the connection?
Wrong connection...
I am assuming everyone understands the connection between energy decay (shown in CSD plots) and frequency response....
But there is no connection between the subject matter of my posts and phase... All my posts are regarding ringing of drivers not their phase behaviour.
The fact that phase (and amplitude) information can be used to help view the ringing issue is hardly addressing the issue...
I am assuming everyone understands the connection between energy decay (shown in CSD plots) and frequency response....
But there is no connection between the subject matter of my posts and phase... All my posts are regarding ringing of drivers not their phase behaviour.
The fact that phase (and amplitude) information can be used to help view the ringing issue is hardly addressing the issue...
Reverb or more precisely the room impulse response is just about everything, so it matters a lot. But this is far afield of the current discussions and perhaps should be in another thread somewhere. How the loudspeakers directivity interacts with the room impulse response is what my work and designs are all about, but its not "lowest distortion speaker drivers".
I can buy that many designers look at things that make better copy than they make better sound, and I have always designed for a "compact" impulse response, which is another word for "low ringing", but I just want it to be clear to people that CSD does not show anything that the impulse response does not. The impulse response and CSD of a well designed compression driver on a waveguide dies like a rock, so they should meet with your approval. Bad horns are the exact opposite.
Yes, if the drivers settling time is fast it will not ring!
You havn't missed anything, its a simple point I am making but I believe it is a point with huge significance.
Ie if it was as fast (able to instantly shed the excess energy) as the electrical it will not ring.
Then there would be nothing (no issues) to view on the CSD plot and if we were using.
The problem is this is impossible!
We need to totally re-evaluate driver design.