Hello,
A moving mass in a resilient suspension system that continues to move when any external force is removed sounds like stored energy to me.
For speaker drivers the one that responds more quickly, least latency and hang over, is more accurate than the others that are slower in response. Toss in graphics that show retained energy that is out of phase with the external force.
That sounds like sour grapes that you did not identify and publish first.
DT
All just for fun!
A moving mass in a resilient suspension system that continues to move when any external force is removed sounds like stored energy to me.
For speaker drivers the one that responds more quickly, least latency and hang over, is more accurate than the others that are slower in response. Toss in graphics that show retained energy that is out of phase with the external force.
That sounds like sour grapes that you did not identify and publish first.
DT
All just for fun!
After request from Mr.Stoll i will remeasure the Scan Speak wideband in a longer and more rigid tube. Mr.Stoll thought that the rather soft mineral water bottle could have some mechanical problems that affected the measurements.
DT, do you still like me to get you 2 x 3Liter spheres ? The one i have shown have 6 liter but i can get you some smaller ones. They come in all kind of diameters. Sorry that i did not respond faster. I could send them to you at the end of next week.
DT, do you still like me to get you 2 x 3Liter spheres ? The one i have shown have 6 liter but i can get you some smaller ones. They come in all kind of diameters. Sorry that i did not respond faster. I could send them to you at the end of next week.
Your statement that distortion rises if eq is applied is equivalent to saying distrotion rises if you trun up (down) the level of the input signal. That may be, but it is not the result of eq. It is the characteristic of the driver's distrotion as a function of input level and it is because after eqing a notch/peak the driver is receiving a different input level than w/o eq.
Eq is a linear function and can not introduce distortion of its own (assuming good electronics). If you introduce a 5dB boost or cut at frequency X then all the driver knows is that it is receiving a input signal that is 5 dB greater of less that it would receive w/o eq. That is, you should measure the same distortion just by increasing/decreasing the level. If you measure differently then you should look to your electronics (the eq circuit) as the source of additional distrotion. Olny then could you attribute the distortion to the eq.
Perhaps a little light can be shined on linear distortion and seperating it from nonlinear distortion. Obviously a system is either linear or nonlinear. It can not be both at the same time. If we subject a system to an input, I(t), then we can express its output in a power series as
O(t) = a0 + a1 x I(t) + a2 x (I(t))^2 + a3 x (I(t))^3 + .....
a0 is the DC aspect of the output which we can ignore for a AC input or just take it as the DC offset in the case where O(t) is an AC signal. A1 then is the coefficient of the linear term. a2 and so on are the coefficients of the nonlinear terms. If a2 and so on are of the same order ("size") as a1, then the system is highly nonlinear. However, if a2 and so on are relatively small compared to a1 we see that the system behaves predominantly as a linear system and the higher order terms are referred to as "nonlinear distortion". However, the linear term should follow some predetermined behavior. For example, if the response is to be that of a high pass filter then we would expect a1 to be defined by the transfer function of that high pass filter. So a1 as a function of frequency would be given by the amplitude vs f of the HP filter, to at least a constant scale factor. Looking a little deeper, the HP filter has both amplitude and phase so that a1 must actually have both amplitude and phase. In any event, linear distortion can then be considered as the difference between the actual value of a1 and the ideal value. Thus, for any system linear distortion is the deviation of the linear component of the output of the system from the target.
Stored energy is the ability of a system to retain energy is some form of reactive element, capaciatnce, inductance, or by exchanging it between two reactive elements. Since any causal system which is not perfectly flat form DC to infinity will have at least one reactive component, any causal system which is not perfectly flat with zero phase has stored energy. Thus, a perfectly linear system with a band pass response will still have stored energy. The perfect linear system would have flat response and zero phase from DC to infinity so in a sence, when measured against the perfect linear system, any causal system with finite band width has linear distortion and stored energy. Thus in a sense, linear distrotion and stored energy are the same thing. when we design systems composed of finite bandwidth components, like a 2 or 3 way sspeaker we can not avoid stored energy. What we strive to do is to reduct it to a minimum. EQing a response flat tends to minimize amplitude related stored energy and eqing phase to linear further reduces stored energy in the system.
O(t) = a0 + a1 x I(t) + a2 x (I(t))^2 + a3 x (I(t))^3 + .....
a0 is the DC aspect of the output which we can ignore for a AC input or just take it as the DC offset in the case where O(t) is an AC signal. A1 then is the coefficient of the linear term. a2 and so on are the coefficients of the nonlinear terms. If a2 and so on are of the same order ("size") as a1, then the system is highly nonlinear. However, if a2 and so on are relatively small compared to a1 we see that the system behaves predominantly as a linear system and the higher order terms are referred to as "nonlinear distortion". However, the linear term should follow some predetermined behavior. For example, if the response is to be that of a high pass filter then we would expect a1 to be defined by the transfer function of that high pass filter. So a1 as a function of frequency would be given by the amplitude vs f of the HP filter, to at least a constant scale factor. Looking a little deeper, the HP filter has both amplitude and phase so that a1 must actually have both amplitude and phase. In any event, linear distortion can then be considered as the difference between the actual value of a1 and the ideal value. Thus, for any system linear distortion is the deviation of the linear component of the output of the system from the target.
Stored energy is the ability of a system to retain energy is some form of reactive element, capaciatnce, inductance, or by exchanging it between two reactive elements. Since any causal system which is not perfectly flat form DC to infinity will have at least one reactive component, any causal system which is not perfectly flat with zero phase has stored energy. Thus, a perfectly linear system with a band pass response will still have stored energy. The perfect linear system would have flat response and zero phase from DC to infinity so in a sence, when measured against the perfect linear system, any causal system with finite band width has linear distortion and stored energy. Thus in a sense, linear distrotion and stored energy are the same thing. when we design systems composed of finite bandwidth components, like a 2 or 3 way sspeaker we can not avoid stored energy. What we strive to do is to reduct it to a minimum. EQing a response flat tends to minimize amplitude related stored energy and eqing phase to linear further reduces stored energy in the system.
You need to read the full report by Linkwitz carefully. The cause of the "stored energy" is an external " diffusor cap" mounted over the diaphragm. The stored or in this case - more appropriately termed, "latent" energy is not that much different from the latency introduced by horn loaded tweeters. The motion of the diaphragm itself is not producing the latency and resonance - the external structure is. So Michael does appear to have a point - not exactly sour grapes. Linkwitz used terminology that is to some extent a little misleading in that most people consider the term "stored energy" to be a reference to energy stored in the movement of a diaphragm with mass.
Let's separate the phenomina of stored energy from the location of stored energy. Otherwise, the discussion can get confusing and misleading.No SL's "stored energy" term isn't interesting at all - its mere B*S*T - and anybody is jumping at it because it applies to each and everything that makes up for the slightest wiggle in FR and because it was coined by an audio icon - sorry Siegfried ...
Would he have had investigated any deeper instead of being superficial (in this respect) - he immediately would have come across the fact that there are fundamental differences in so called "stored energy" - and - to lump it together in a phony term will not make understanding of whats actually going on any easier.
Michael
This is why the measurement point becomes very critical because the mic receives a combination of waves.
Thanks for the explanation. I think it's easy to understand when put that way.
From this explanation, it seems that equing for each driver then equing the entire system would probably be best, which would make it a multistep process.
I see it that way : ones the sound has left the menbrane it is hard to control. That is one of the reasons i try to watch the radiation pattern in my design and use drivers that do not need a lot of EQ and use them in the optimum frequency range where not much breakup and bundeling happens. I agree with John though. I perfectly digitally linearised loudspeaker behaves like a band pass filter of the same amplitude and phase response ON THE DESIGN AXIS when it is not over or UNDER driven. Yes, some drivers show rising distortion when the driving level is too low so there is a resolution limit.
Hello,
With a quick look this Midrange distortion test is the earliest i see a reference to stored energy. I do not see a external cap. It appears that all the drivers were tested under the same conditions.
Facts Please? If this is a hypothetical construct show us.
DT
All just for fun!
This is the link Joachim provided that was the subject of Michael's comment.
Issues in speaker design - 2
If you examine the section towards the end, it discusses the tweeter's "diffusor cap".
Not sure how you got onto the midrange test link.
The basic problem is that "linear distortion" and "stored energy" sub sums CMP effects too - which are a different animal altogether.
Michael
You are simply saying that radiation of sound from a source varies with position. By your definitions every source is CMP with the exception of a true zero diameter point source. To me, accepting CMP in your terms would be mean CMP in my terms: Compromising My Principles.
Listening position ?
Not exactly - but even *if* - whats wrong with that ? - cant see it - its real world reality for most sound radiating systems - having put it as that diffraction cant be avoided, just be aligned
The point regarding CMP is that we have to look at the systemic behaviour of pronounced delay first to understand the fundamental difference regarding its specific behaviour - then we can have a look further on smearing due to "non point source behaviour" (more / less pronounced CMP)
To a more or less extent, sure - so what - less interesting to investigate in CMP then ?
To me, accepting CMP in your terms would be mean CMP in my terms: Compromising My Principles.
I love you using this very emoticon - don't remember having seen you doing that before
Michael
Last edited:
There is nothing new in any this other than you applying your preferred name to what has been studied and understood for years. I have and still use DOS software from the late 80's (CALSOD) that was specifically written to model drivers using, in its own name, one of its many modules called MPE, "Minimum Phase Element". CMP adds nothing new other than a name.
I refer you to the AES paper submitted in 1987 and revised and published in 1988, "Simulation and Optimization of Multiway Loudspeaker Systems Using a Personal Computer" by Witold Waldman. You can find it in the AES anthology "Loudspeakers Volume 3". It extended the work of Peter Shuck and relies upon the then well-known minimum-phase nature of drivers within their useful operating ranges. My use of CALSOD for years showed that even the breakup is minimum-phase and certainly within intended driver usage is linear.
There's simply nothing new with CMP other than the name.
Dave
Last edited:
Dave, I don't enter any discussion about details in any papers I haven't read and most certainly never will, unless they are available for free - even then - if you feel like - you are invited to cross (s)words rather in the CMP related thread, as such discussion possibly may take a little bit longer ...
Best
Michael
Best
Michael
If you're going to try to claim that you've "discovered" something new, it's not only incumbent on you to research and read all relevant prior work, it's unethical not doing so. I see no point to join in your thread unless and until you do your due diligence in this regard.Dave, I don't enter any discussion about details in any papers I haven't read and most certainly never will, unless they are available for free - even then - if you feel like - you are invited to cross (s)words rather in the CMP related thread, as such discussion possibly may take a little bit longer ...
Best
Michael
Dave
...Dave, I don't enter any discussion about details in any papers I haven't read and most certainly never will, unless they are available for free - even then - if you feel like - you are invited to cross (s)words rather in the CMP related thread, as such discussion possibly may take a little bit longer ...
Best
Michael
AES papers are really not expensive if you subscibe to the library.
Well - I'd say as long as UE isn't out its not exactly a tool that can be used by Joachim - no ?
(the ver 17 implemented is only "so lala" for the matter, as fa as I've researched - still awaiting an upgrade offer 16>17 from Bohdan)
Besides that - PC based convolution XO and EQ (and also dedicated hardware based) are a nice tool in our box as shinobivan and others (me included) have raved about extensively.
What I was making aware of - those benefits of advanced (and less advanced) EQing are generally - and brick wall - limited to systems that do not show pronounced CMP - be it whatever make and brand (which - until now - no one has raved about ).
Michael
(the ver 17 implemented is only "so lala" for the matter, as fa as I've researched - still awaiting an upgrade offer 16>17 from Bohdan)
Besides that - PC based convolution XO and EQ (and also dedicated hardware based) are a nice tool in our box as shinobivan and others (me included) have raved about extensively.
What I was making aware of - those benefits of advanced (and less advanced) EQing are generally - and brick wall - limited to systems that do not show pronounced CMP - be it whatever make and brand (which - until now - no one has raved about
).Michael
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.