basic theory behind what we do - Page 4 - Home Theater Forum and Systems - HomeTheaterShack.com

Reply
 
Thread Tools
post #31 of 44 Old 07-27-08, 04:12 PM
mas
Inactive
 
Posts: n/a
Re: basic theory behind what we do

Quote:
lcaillo wrote: View Post
Seems like you are long on condescension and short on practical advice. You make lots of assumptions about what people do, know, believe, and understand.
Really?
I started off by expressing concern about the perceived use of EQ, especially since the thread began with the question that presumed a use for which EQ has traditionally been used and fo which it fails.

Except for sparse regions, the sound field of a room with multiple real and/or virtual superposed signals is not minimum phase and cannot be EQ'd.

{Of course, instead of acoustics principles, you can always choose to discuss me, as if that changes the validity of the concept to which I refer... But I would suggest that that is a rather boring subject as compared to the acoustical physics issue at hand.}

So, one might ask, if they are not familiar, what superposition is, or what real or virtual sources are, or why destructive superposition is not equalizable...but so far no one has. One of the moderators has referred to minimum phase conditions. Do you or others understand what is meant by minimum or non-minimum phase? As the term was used, should I assume that no one understands it, or that this has been covered in depth and that it is part of the assumed baseline knowledge here? And if it is, why are we having this discussion? And since we are, is it valid for me to assume that the term is not widely understood, as its implications for EQ should be understood as well! Hmmm, I understand the term and its use and implications...and I am both chastised for assuming some understand it and also for others not knowing it. LOL!


Then I made a comment to the effect that many folks who consider themselves 'audiophiles' still have the notion that room problems (other than the direct Ld signal) can be equalized and that room issues consist of tuning the frequency response of the room in the frequency domain, (of which DrWho can verify at least one prominent instance of this!) as well as the additional evidence of many of the increasingly common 'room EQ' functions built into many devices - many of which simply measure the frequency response of the mic position and invert and feedback the signal in an attempt to flatten the frequency response - if ONLY such a method were so simple or effective!

Then others who expressed confusion over the limitations I had quickly mentioned stated that others were quite happy with using EQ. As I was concerned with limitations regarding its use for the general use in altering room response other than in the limited situations originally stated (LF modal peaks and minimum phase regions) I am not sure what their point was (as the link did not work for me).

Or you can waste your time debating the semantics of just what constitutes an "audiophile'! If some what to debate a definition of the term 'audiophile', they are more than welcome to do so. If they want to debate whether the aggregate to whom I refer are 'audiophiles', I suggest that this is not only a separate issue, but it is spurious to the basic substance of the acoustical issue stated regarding the valid and invalid uses of equalization which are very limited in addressing full range acoustical room phenomena.

I also suggested that others could query another member (DrWho) for verification that at least several prominent persons who consider themselves "audiophiles" have expressed quite the opposite opinion. If anyone doubts the facticity of this claim, PM me and I will provide the links!

I then provided several links which address some of the legitimate uses and the very limitations to the use of EQ to which I asserted and was then told that acoustic references from pro-audio sources, because they are considered from 'pro' sources, are somehow invalid in 'amateur' acoustics - a distinction which is curious to me as within the limits of similar atmospheric conditions, large acoustical space and small acoustical space acoustics are the same - be it for pro or amateurs!! A very curious distinction!



As to how one EQ a room other than for LF modal response which has already addressed at length in the forum?

Its simple, yet not so simple.

Measurement tools such as the TEF EZ-Tune software or the use of SMAART are able to identify the minimum phase passbands (if any indeed exist) where the response is minimum phase and for which parametric EQ settings are calculated and specified, and can then be applied and the response verified, Other than these methods, I am not aware of a simple efficient manner to to do it. And such attempts, without the aid of specialized equipment are generally more destructive than useful


But even if one has the ability to identify the minimum phase passbands of a room response capable of being EQ'd, one is focusing on the very last step of a long process and missing the larger more productive steps in tuning the room and putting the cart before the horse!

Instead, after addressing room modes, one should be focusing upon the envelope time curve (the ETC response).
Here you need a ETC display capable of displaying the discrete specular reflections in terms of their arrival times and their amplitudes. From this can be surgically resolved the specific paths and points of incidence whereby the surgical (minimum) application of absorption is used to damp the early first order reflections sufficient to define the Initial Signal Delay (ISD) gap - a period of time in which the listening experience is essentially anechoic and the predominate source is the direct signal Ld.

After this defined period where the early arriving specular reflections that interfere with intelligibility (Haas effect), we address the discrete focused specular reflections. Rather than absorbing these reflections (a technique which dominated the period of time prior to Peter D'Antonio's application of Manfred Schroeder's reflective phase gratings and which this technology has largely rendered this technique obsolete since the late 80s), the discrete display of specular reflections via the ETC and diffusive technology allow us to resolve each focused specular reflection with regards to its path and reflective points of incidence. At these points we employ diffusive treatments which not only diffuses the signal reducing the concentration (amplitude) of acoustic energy, but in so doing distributes the energy in time, producing a more diffuse sound field. The overall later arriving specular reflections are adjusted as necessary using diffusive treatments (and only rarely the additional surgical use of absorption for very exceptional issues) until the energy that constitutes the later arriving specular reflections is reduced in amplitude and diffused in time in a manner that it approaches a decreasing logarithmic slope decay.

The finite available later arriving acoustical energy in the form of specular reflections is not, nor need it be, absorbed. Rather it is diffused in a well behaved manner such that it it is devoid of more intense specular reflections that are detrimental to the perceived intelligibility while at the same time imparting an increased, psycho-acoustically pleasant, sense of space. In doing so, a region dominated by sparse focused specular reflections is transformed into a more greatly diffused semi-reverberant space.


Only after this is accomplished is EQ to be attempted!

At this point, if one has the means, minimum phase regions of the passband at a particular listening position are identified and precise parametric equalization applied to the degree possible. Note, that there do not always exist minimum phase regions capable of being equalized. And lacking the ability to effectively identify the minimum phase passbands capable of being equalized, the majority of the most significant tuning which constitutes the MAJOR improvement in intelligibility has already been accomplished. It is the fundamental addressing of the acoustical field with respect to time that resolves the vast majority of issues, including those that may may be observed in the alternative perspective of the frequency domain.


In lieu of the ability to identify minimum phase passbands, the application of EQ merely shifts the phase of the direct signal which then recombines with the various additional real and virtual sources (reflections) via superposition and the resultant comb filtering and polar anomalies are shifted slightly with regards to their frequency and orientation about the axis, as well as severe phase wrap (group delay) issues.

So, I would suggest that there are much more fundamental issues to be addressed before EQ is even considered in regions other than for the mitigation of LF modal peaks.

And to identify minimum phase regions requires a few advanced tools.


And it was only several days ago that I was asked just what such things such as TEF, EASERA, SMAART, and other advanced industry standard tools are. So, should I assume that ll are familiar with them, or should I assume that they are not?

You are worried about what I assume, when I quite frankly can only assume that many are not familiar with the equipment, or the theory as evidenced by some of the questions and assumptions posited in the nature of the thread itself!! And all you offer is a challenge to what you perceive as my assumption, despite a few here claiming to be familiar with the concept of minimum phase and others not having any idea as to the limitations to which I refer!!! Again we can relate to Dorothy's dilemma! "Which way to OZ, Scarecrow!"

So, as in logical truth tables, a negative in an either/or case renders the results negative, I have chosen to assume that many here do not understand the limitations of the use of EQ. And such an understanding is important. But in order to understand the full reasons, it requires a broad understanding of acoustical behavior in the time domain as well as superposition. And we are very quickly back at a very large change in the entire perceptual paradigm that has occurred during the last 35 years in acoustics - a veritable quantum leap if you will. So where do you want us to start? Again, I would begin by pointing you to Sound Sytem Engineering, 3rd Ed., by Davis and Patronis for a baseline introduction to the subject. (Do you denote a pattern here? ;-)

I can refer you to a few quality sources that attempt to introduce many of the fundamental changes in the paradigm, but it is just a bit beyond the scope for me to present the full explanation of the changes! And how many have bothered to query the sources such as Davis and Patronis' Sound System Engineering, 3rd Ed. that attempt to address this issue? How many have read Heysers works which are collected in the AES Anthology of his works? I can make reference to them, but if you are expecting a 25 word or less explanation of them, aside from summary statements regarding what can be done and what cannot, you what what to my knowledge cannot be done.

If you want to attribute that to a personal limitation of mine, have at it. But I must suggest that it also reflects on those who refuse to seek out the readily available sources published in various texts of D'Antonio and Cox, the AES Journal, ASA Journal , Davis and Patronis, Manfred Schroeder, various SynAudCon publications, ProSoundWeb articles and any number of other sources!

I can make comments and references. Do I have to do all of your reading for you too? The irony is that I have not only read, but been involved with a lot of the research over the years. If someone has a question that they want to ask me, PM me and I will be glad to chat or even call if you live in North America and discuss it and try to narrowly refine whatever you might like to know that I can assist with. But if all you have is an issue with your presumptions of my assumptions, you are going to have to do a bit better than that - as evidently you are not familiar with the topic to which I refer.

And folks are evidently still wondering about the desirability or, even more fundamentally, the possibility to simply EQ a room response to achieve a "ruler flat" response! The fact is, without it being minimum phase, you can't. You are going to have comb filtering and the accompanying polar anomalies.

Needless to say, my assumption is that many are not aware of the fundamental changes in the acoustical paradigm. And they have profound implications for the tuning of rooms, as well as system design! And acoustics does not vary between amateur or audiophile or pro applications! The only variant that one might legitimately make between these groups might, generally but not necessarily, be the nature of the acoustical environment, be it a large or small acoustical space as defined by Schroeder (for which you can read the pertinent chapters in Davis and Patronis' Sound System Engineering) and perhaps a variation in the purpose for the various spaces being utilized. (I wonder if I should make reference to SSE once again??? ;-)

Bottomline...one cannot EQ a room to resolve issues resulting from the superposition of multiple phase variant (non-minimum phase) signals. If you already understand this, great. For those who do not, may I suggest discovering why by referencing some of the sources listed, and then asking further questions. But if you do not know, and simply want to complain about my assumptions, I would suggest that your eforts are misplaced.

And I suspect that my assumption that many still do not understand the fundamental limitations of employing EQ in a non-minimum phase acoustical environment is a valid one. Otherwise we would not be ssing such good questions as the the one that leads this thread. With luck, it just may lead to a better understanding of the problems associated with such an assumption.

And I would suggest a better use of your time would be to discover why this is the case, rather than simply complaining that I have not reproduced the source material here for which I would be glad to do were it not for EULA of the site!

You will just have to pardon me if I fail to respond in a manner that targets and accommodating every question or misunderstanding that everyone of varying levels of understanding might have! LOL!

But for those who have made an effort and who are still confused, my offer of trying to provide a more targeted explanation remains open if you simply PM me. Of course, you can always complain instead about what you assume my assumptions to be...if you think that adds to your understanding of acoustics...



Stylized ETC 'regional' diagram (note: the Haas kicker generally corresponds to the direct signal Ld):




Example of an ETC showing before and after snapshots:

Last edited by mas; 07-31-08 at 02:49 PM.
Sponsored Links
Advertisement
 
post #32 of 44 Old 07-28-08, 12:53 AM
Elite Shackster
Maverick
 
Join Date: Mar 2007
Location: Sandnes, Norway
Posts: 1,384
Re: basic theory behind what we do

I'm sure I'm not alone here, I sure like to have someone 'in the industry' on this board.

Could you please, in layman's terms if possible try to explain what a minimum phase system/region is?

And what is a polar anomaly, and what does one sound like? I'm sure it's not another environmental issue?

I have read a few books on acoustics, but many are not written for a 'selftaught' layman like myself.

I have used EQ myself, in a subwoofer application, with great effect. I currently use Audyssey, which is not 'EQ' in a traditional sense. I would love a deeper understanding of the subject, to better identify problem areas in my own room, so I can address them 'physically', not electronically.


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.



To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!

To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Done


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!
atledreier is offline  
post #33 of 44 Old 07-28-08, 06:22 AM
mas
Inactive
 
Posts: n/a
Re: basic theory behind what we do

Superposition is a fundamental characteristic of wave behavior. Derived from ‘to superimpose’, superposition is the way two or more waveforms sum based u[on their phase offset. If the waveforms are in phase they simply add and the amplitude is greater by 6dB. If they are 180 degrees out of phase, they destructively sum to zero amplitude, effectively canceling each other – a behavior that is more commonly known as active noise cancellation. When the phase is other than unity or 180 degrees, the sum is a complex combination that combines the characteristics of both addition and subtraction.

If two or more signals originate from two or more sources, be they real or virtual (such as a diffractive or a reflective source), superposition is what determines their combined response.

While most may be used to seeing such behavior manifest in waveforms on a device like an oscilloscope or in waves in a string, in 3 space, sound waves exhibit this behavior in two ways of great concern to us.
These two characteristics are comb filtering and polar anomalies. Depending upon the time offset (which can also be seen from the perspective a separation by distance – with the distance corresponding to a time offset), there is a fundamental cancellation that is repeated at integer harmonic intervals in the frequency domain. The frequency response looks like a comb with a repeated series of notches, and hence its name.

The polar response refers to the spatial dissipation of the energy, both in a vertical and horizontal aspect in 3 space. Again, with the in phase summing increasing the gain by 6 dB and the 180 degree out of phase intersections resulting in complete cancellation of acoustic energy – with an infinite number of in between states – especially as we deal with non-sinusoidal waveforms.

Below is posted a picture of the effect of 2 speakers modeled in EASE, and you see both the comb filtering in the frequency domain as well as the spatial polar response – how the summed waveform propagates in 3 space. I have also included a few diagrams from a wave tank which display a 2space dissipation of the polar response of an in phase real and virtual source (wall and speaker) illustrating space loading, real and virtual out of phase sources that combine destructively, and two speakers which also sum destructively.

What does this mean? It means that depending upon your location relative to the two (or more) sources, that certain frequency multiples will be notched – creating the equivalent of a highly equalized signal with certain frequencies missing entirely, others accentuated unnaturally, and still others in between emphasized and deemphasized based upon the characteristics determined by the wave itself and the time/distance (phase) offset.

If you have ever been to a concert with multiple arrays or spaced stacks, the effect of this can be dramatic. Next time, you might want to walk slowly from one side of the hall to the other side – from stage right to stage left about half way back from the stage. You will note various spots where certain frequencies seem to literally disappear, and after moving a few feet more, other frequencies being exaggerated and still other frequencies disappearing. In other words, as you move about, it is as if someone is playing a radical game with an EQ. What is happening is that you are physically moving in and out of polar lobbing nulls and sums, and also experiencing the frequency anomalies that ‘go with’ the spatial polar distribution. And depending upon the physical and time offsets, the distribution will vary in both the horizontal and vertical planes. But as we usually aren’t going up and down stairs while listening, we are usually only concerned with the horizontal plane, But in larger venues where you have balconies, etc., the distribution in the vertical plane can also come into play.

Oh, and to make this even more complex, this superposition occurs in a number of orders of scale – of a variety of magnitudes where we have two or more sources reproducing the same signals. Examples are multi-driver speakers – ESPECIALLY if they employ non-signal aligned passive crossovers where adjacent non-coincident drivers – or a driver and a virtual propagation source such as a cabinet diffraction - reproduce an overlapping passband, and then to the next order of magnitude where we have two speaker assemblies reproducing the same signals (especially as in a concert where multiple speakers are summed for additional gain), to the combination of a speaker assembly and their summation with virtual sources such as their reflected signals off room surfaces. (Note, the origins of the D'Appolito MTM configuration was designed to compensate for exactly such a vertical polar tilt utilizing a 3rd order LR filter.)

Thus the situation is reminiscent of a Russian Matryoshka, or nested doll, set. We have a situation where this phenomenon manifests itself in a variety of orders of magnitude.

So we have big trouble in River City! What do we do? Good question!

First, the superposition of signals that are out of phase are known as “non-minimum phase”. While mentioning this, we are not going to dwell much in the technical definition which involves the H(s) in what is commonly referred to as the ‘S plane’. In this model, In a minimum phase system, each pole in the transfer function must lie in the negative half of the complex S plane. While this is quickly getting beyond the scope of this discussion, it is also jumping smack dab in the middle of one of the most exciting developments that has arisen from the development of the TEF – the ability to not only measure, but to display in 3 space, the complete complex impedance of a system - and the system assumes the form of a rotating phasor existing in the real(X) and imaginary(Y) planes rotating about the time or frequency (Z) axes. The imaginary(Y) axis corresponds to kinetic energy and the real(X) axis corresponds to potential energy. And the rotation of the phasor provides a signature of the complex impedance (reactance) at any time or frequency. These displays are known as the Heyser spirals and Nyquist display and are profound in their implications. From these diagrams, the projections of the total response upon each plane - the polar, imaginary and real planes, provide all of the perspectives available for the various topological mappings of the behavior of the total acoustic (and electrical) phenomenon. (I have included a few VERY simplified examples in the diagrams that are easily generated for a component or a system with the TEF) Note that this response captures the entire complex impedance of the system response.

Additionally, from the phase response, we are able to ascertain regions of minimum and non-minimum phase behavior. But this aspect is a bit beyond the scope of what we are addressing at this juncture…

OK, I realize that I have probably lost a few of you, but I hope that a few will realize that they have just seen a diagram of what imaginary number represent! And they are anything but imaginary!

Rather they are a view of the phenomena that we normally view in the real domain whose projections onto their respective planes represent the impulse(real) and doublet(imaginary) responses each related by a shift of 90 degrees by the Hilbert transform to provide the complimentary point of view, whereas the view with respect to the frequency domain affords the projection onto the respective planes related as well by the Hilbert transform in the form of coincident(real) and quadrature(imaginary) responses.
The significance is that one discovers that amplitude and phase are simply two different viewpoints of the same event. And as should become apparent, the Domain Map provides a relational diagram of the various perspectives e now have available to us of an acoustical event.

No longer are we limited to the flatland of the frequency response! And we also quickly discover that the time domain exhibits a fundamental causal relationship relative to the frequency domain, meaning that events in the time domain are determinate over much of what is manifest in the frequency response perspective.



And this is true as well with minimum and non-minimum phase environments. We quickly discover that manipulation of the frequency domain is well-behaved only in a minimum phase relationship.

In the room, we ONLY have control over the direct signal emanating from the speaker (which is, as you will recall, also subject to the superposition of the various drivers, especially if they utilize passive crossovers which lack substantial ability to align the acoustic origins – the ability to make align the two sources, at least in one plane, in the time domain. A feature often referred to incorrectly as ‘time alignment’. But note, we can only align signals within the time domain! We cannot align time!

So, ideally, only the direct signal can be equalized (cognizant that non-coincident drivers and passive crossovers as well as diffractive factors impart their own degree of superposition and comb and polar anomalies. But we are going to selectively ignore this aspect for now! We have bigger fish to fry here!!

In our example, we can only equalize the direct signal (Ld). But this is not the only signal that we experience in the space of the room! We not only have multiple real sources of a signal, but we have the reflected focused specular reflections in the form or first order early arriving reflections which arrive within what is commonly referred to as the Haas interval ( a subset of the Henry Precedence Effect) – a period of time where our ear/brain lacks the acuity to effectively resolve each arriving reflection into separate signals, with the practical impact being that the signal is smeared and the intelligibility suffers. A classic example of thi that everyone has experienced is in a gymnasium or airport, where the distributed PA system exhibits plenty of gain, but which the resultant sound is utterly incomprehensible. This is precisely due to the lack of controlled distribution with many sources arriving within the Haas time interval subverting our ears ability to resolve each signal separately.

We experience the superposition (summation) of the direct and reflected signals arriving at slightly different times in a variety of ways. This affects us in the small acoustical space in the form of reduced intelligibility, localization and image shift, as well as comb filtering and polar lobing anomalies.
And equalization cannot correct for superposed signals! It may adjust the frequency content of the direct signal, and by virtue of the general use of (R)LC filters, vary the phase a bit, but the reflected signal is still offset in time from the direct signal, and the resulting superposition results in comb filtering and polar lobing anomalies simply shifted a bit relative to the changed direct signal. We cannot alleviate the destructive effect. We can only ‘move the problem around’.

Let me digress and explain the fundamental problem with the RTA, essentially a fancy multi-band sound pressure level meter. They do have a few uses, but they are extremely limited with regards to what we want to accomplish. The problem is that we need to be able to resolve the component reflections into the pieces and parts that comprise the superposed/summed signals. We need to be able to deal with each component piece in order to adjust and maximize intelligibility. And the RTA and SPL meter are unable to do so. Thus we simply get a picture of the entire mixed stew.

So, we have a mess.

We need tools that allow use to atomistically identify and focus on each component part of the stew – each reflection in terms of amplitude, arrival time and corresponding distance of travel and locations of their points of incidence as they reflect about the space.


That is where the TEF, EASRA, SMAART and other time domain ‘aware’ tools prove invaluable and make possible what the RTA simply cannot do.. And while we can look at this via many perspectives, the most useful view is the Envelope Time Curve (ETC) which shows the direct signal as well as each specular reflection in terms of their amplitude and time of arrival – and from there, depending upon the tools you use, or the additional busy work you must perform based upon the power of the analytical platform employed, we are able to isolate the component reflections and all of their characteristics.

(Ignoring the use of the cumulative spectral decay response – commonly referred to as the waterfall plot, and the issue of standing waves were the size of the wavelengths are large compared to the room dimensions, we will move ahead to the behavior of acoustical energy where the wavelengths are small relative to the room dimensions – where sound waves behave more like light waves. )

Based upon this, we proceed to define an initial signal delay gap, where we essentially experience the direct signal anechoically, observing the limitations of the Haas interval and allowing us to maximize our ability to localize the signal as well as the intelligibility of the signal. We establish this initial signal delay gap by judiciously and surgically applying absorption.

Thereafter, we want to establish a well-behaved diffuse semi-reverberant sound field from the reflections arriving outside of the of the Hass interval, which imparts a sense of space as opposed to an overly damped dead zone. Here we must identify. This is where we address the focused specular reflections and reduce their amplitude and reduce their focused characteristics by increasing their randomized diffusion in both space and time with diffusers. The energy in the small acoustical space is finite, and we do not want to overly absorb this acoustical energy as its decay in the overwhelming majority of cases is not sufficient to cause concern. In fact, it is an all too frequently encountered problem that the room exhibits a far to absorbent nature! A significant problem specifically acknowledged in the publications of such notables as D’Antonio, Burger, Davis, Janssen, Cary a list far too long to post here. Instead we want to ‘redistribute’ it in a well behaved manner by manipulating it to create a more diffuse lower energy distribution over time.

Well, I guess this enough for now. This is akin to a 25 word overly reductionist simplification of the many profound implications afforded by the tools and perspectives afforded to us. And I dare it should be readily apparent that the flatland perspective limited to the traditional frequency domain is a very limited point of view! In the immortal words of Dorothy “I've a feeling we're not in Kansas anymore!”


Comb filtering and superposition resulting from 2 spaced sources

Wave tank examples of polar lobing anomalies for a real and virtual (reflected) source in phase; out of phase, and two spaced sources (with non-reflective damped lateral and far end surfaces)



Domain map



Heyser spirals with respect to time and frequency

Last edited by mas; 07-31-08 at 02:50 PM. Reason: This editor stinks! as it deletes significant material in the body and attachements when simple typos are corrected
 
post #34 of 44 Old 07-28-08, 08:32 AM
REW Author
 
JohnM's Avatar
John
 
Join Date: Apr 2006
Location: UK
Posts: 6,710
Re: basic theory behind what we do

Thanks for taking the time to assemble that Mark, very well covered.

Quote:
mas wrote: View Post
First, the superposition of signals that are out of phase are known as “non-minimum phase”.
That is perhaps over-stated. Superposition only really becomes non-minimum phase when the delay exceeds the period at the frequency being considered, so whilst any system with a time delayed reflection will become non-minimum phase at a high enough frequency it remains minimum phase below that, which is particularly important in small rooms with correspondingly shorter delays.

Quote:
mas wrote: View Post
While mentioning this, we are not going to dwell much in the technical definition which involves the H(s) in what is commonly referred to as the ‘S plane’. In this model, In a minimum phase system, each pole in the transfer function must lie in the negative half of the complex S plane.
That is not correct, a system with right half plane poles would be unstable which, feedback aside , is not what we are generally dealing with. Non-minimum phase systems have right half plane zeroes (or zeroes outside the unit circle if we deal in the z plane, which is probably more approriate for the sampled data systems we generally deal with). Reflecting those zeroes back into the left half plane/inside the unit circle gives us the equivalent minimum phase system with the same magnitude response.

Quote:
mas wrote: View Post
That is where the TEF, EASRA, SMAART and other time domain ‘aware’ tools prove invaluable and make possible what the RTA simply cannot do.
The focus on RTAs is probably a bit unfair, particularly on the support forum for an acoustic measurement tool which is far from an RTA . Most acoustic measurement software is directed at determining the impulse response and using that as the source of subsequent analysis views, that has been the case for a long time now, paticularly since MLS-based systems appeared and right through the transition to the now more commonly encountered logarithmic sweep based systems we find around us today - even TrueRTA offers a log sweep based frequency response measurement!

Quote:
mas wrote: View Post
And equalization cannot correct for superposed signals! It may adjust the frequency content of the direct signal, and by virtue of the general use of (R)LC filters, vary the phase a bit, but the reflected signal is still offset in time from the direct signal, and the resulting superposition results in comb filtering and polar lobing anomalies simply shifted a bit relative to the changed direct signal. We cannot alleviate the destructive effect. We can only ‘move the problem around’.
That is simply untrue. Even if you decide to reserve the term "equaliser" solely for devices applying IIR-based filters to the signal passing through them equalisation is highly effective in countering the room's modal resonances (which are about as superposed as signals get) in the region where the response is largely minimum phase, below a couple of hundred Hz in domestic-sized rooms. Expanding the term "equaliser" to encompass the many FIR-based devices on offer nowadays that apply (partial) inversion of the impulse response (not the frequency response!) brings us into the realm of devices that can and do fully correct for the effects of delayed reflections, albeit with the important proviso that the inversion is only valid within a region of the measurement position and that region becomes ever smaller as frequency increases. I'm not personally a fan of such products, but that is no reason to deny their existence. The scope within which they can improve the system response is well explored in various papers.
JohnM is offline  
post #35 of 44 Old 07-28-08, 10:21 AM
Senior Shackster
 
Bob_99's Avatar
Bob_99
 
Join Date: May 2006
Location: New York
Posts: 649
Re: basic theory behind what we do

After reading through this thread, I feel the Shack owes me an Associate Degree in Acoustics as the very minimum. Seriously though, great information and reflects my thoughts that room treatment was always the primary way to approach the acoustics. Thank you Mark for taking the time to post all this information.

Bob

"There is always hope, even if it is just a fool's hope."
Bob_99 is offline  
post #36 of 44 Old 07-28-08, 11:24 AM
mas
Inactive
 
Posts: n/a
Re: basic theory behind what we do

Thanks for taking the time to assemble that Mark, very well covered.

Quote:
mas wrote: View Post
First, the superposition of signals that are out of phase are known as “non-minimum phase”.
That is perhaps over-stated. Superposition only really becomes non-minimum phase when the delay exceeds the period at the frequency being considered, so whilst any system with a time delayed reflection will become non-minimum phase at a high enough frequency it remains minimum phase below that, which is particularly important in small rooms with correspondingly shorter delays.

Quote:
mas wrote: View Post
While mentioning this, we are not going to dwell much in the technical definition which involves the H(s) in what is commonly referred to as the ‘S plane’. In this model, In a minimum phase system, each pole in the transfer function must lie in the negative half of the complex S plane.

That is not correct, a system with right half plane poles would be unstable which, feedback aside , is not what we are generally dealing with. Non-minimum phase systems have right half plane zeroes (or zeroes outside the unit circle if we deal in the z plane, which is probably more approriate for the sampled data systems we generally deal with). Reflecting those zeroes back into the left half plane/inside the unit circle gives us the equivalent minimum phase system with the same magnitude response.

Sorry, but it is correct. To quote: "A physically realizable system is one whose impulse response decays with time. In order for this to be true each pole in the transfer function must have a negative real part. Geometrically, this means thst all poles must be located in the left half of the S plane. Given that this is the case, are there similar locations of the zeros of transfer functions? The answer to this is in the negative. Zeros, whether real, imaginary or complex can be located anywhere without any influence on stable system realizability. However, zero locations do play a role with regards to a system's phase response."

An analysis of the transfer functions is provided on page 380 of Sound System Engineering, and equalization and filter types are explored relative to this throughout Chapter 14, Signal Processing.

It further goes into great detail regarding the impossibility of equalizing a system where the transfer function has a zero on the right (positive) side of the S plane, which it can have if it is non-minimum phase, it is impossible to cancel the zero by a superimposed pole. Equalizing minimum phase systems results in the simultaneous correction of phase and amplitude.

"Equalization should only be applied to minimum phase systems and even when it can be employed, it must be employed with care so as not to force any subsystem beyond its range of of linear operation.

Going back almost 19 years, I have several notebooks of notes and papers as well as discussion notes from multiple workshops and seminars with Gene Patronis (PHd, physics professor emeritus, Ga Tech), Don Keele, Don Davis, Bruce Howse and correspondence with Dr. Ahnert regarding exactly this issue. When this group agrees, I will defer to their combined intelligence.



Quote:
mas wrote: View Post
That is where the TEF, EASRA, SMAART and other time domain ‘aware’ tools prove invaluable and make possible what the RTA simply cannot do.

The focus on RTAs is probably a bit unfair, particularly on the support forum for an acoustic measurement tool which is far from an RTA . Most acoustic measurement software is directed at determining the impulse response and using that as the source of subsequent analysis views, that has been the case for a long time now, particularly since MLS-based systems appeared and right through the transition to the now more commonly encountered logarithmic sweep based systems we find around us today - even TrueRTA offers a log sweep based frequency response measurement!

The mention of RTAs and SPL meters is in regard to tools many commonly use to 'tune' their systems. There are a plethora of additional tools available, The fact is that far too many still use such tools to focus on the frequency response rather than the behavior in the time domain.

And this thread began with a member focused on the notion of the goal of attempting to equalize a system until it was "ruler flat"! Obviously the limitations of addressing fundamental issues in the frequency is not well understood here. As such, a focus on such limitations is in order.

And many of the systems mention specifically use MLS, including TEF (in addition to TDS), EASERA and SMAART! And we could extend that list much further to include additional systems such as Praxis, CLIO, SysID and others! Quite frankly, having the larger systems, I have not bothered to mess with many of the small systems that have become available except in a passing manner. While they may work just fine, they lack some of the flexibility and additional features that render the more powerful tools so valuable and which comprise a complete multifunction analytical platform. This is not a slight toward the small tools, but it is a tacit acknowledgment that they are more limited in their range of features and tools.

And TDS remains the prime environment of choice on TEF and EASERA, if for no other reason than its dramatic advantages with regards to noise immunity that NO other platform affords. After all, if one is shooting a large auditorium and addressing incident surfaces requiring the access to points greater than 150 feet in the air, especially when there is a very real potential for making far too many repeated trips to isolate and effect solutions, it sure is nice to be able to quickly and reliably resolve the 3space incident points from the measurement location on the ground by use of software and a laser pointer thus limiting trips up scaffolding and scissor jacks to only those absolutely necessary to affect a solution; as well as to be able to listen to music over the same system while generating literally hundreds or even thousands of otherwise mind numbing sweeps over the course of a day! Besides, in the real world where one must contend with construction deadlines and the noise that would render any other test format, including MLS, moot and unable to proceed, how often does one encounter a hall with no noise, concurrent construction or maintenance in progress?? Time is money, and efficiency, as well as the ability to interface with systems within the common usage constraints of the real world - not to mention the ability to preserve one's sanity while being subjected to a barrage of mind numbing sweeps all day long, can certainly reduce the stress of the task


Quote:
mas wrote: View Post
And equalization cannot correct for superposed signals! It may adjust the frequency content of the direct signal, and by virtue of the general use of (R)LC filters, vary the phase a bit, but the reflected signal is still offset in time from the direct signal, and the resulting superposition results in comb filtering and polar lobing anomalies simply shifted a bit relative to the changed direct signal. We cannot alleviate the destructive effect. We can only ‘move the problem around’.

That is simply untrue. Even if you decide to reserve the term "equaliser" solely for devices applying IIR-based filters to the signal passing through them equalisation is highly effective in countering the room's modal resonances (which are about as superposed as signals get) in the region where the response is largely minimum phase, below a couple of hundred Hz in domestic-sized rooms. Expanding the term "equaliser" to encompass the many FIR-based devices on offer nowadays that apply (partial) inversion of the impulse response (not the frequency response!) brings us into the realm of devices that can and do fully correct for the effects of delayed reflections, albeit with the important proviso that the inversion is only valid within a region of the measurement position and that region becomes ever smaller as frequency increases. I'm not personally a fan of such products, but that is no reason to deny their existence. The scope within which they can improve the system response is well explored in various papers.

Untrue? Really!?

If one had bothered to read the complete post, a specific exclusion of room modes and wavelengths large compared to room dimensions was explicitly stated.

And again, the equalization of systems with IIR or FIR filters are limited to minimum phase regions. And such minimum phase regions, however limited, as well as the precise parametric EQ settings, are quickly identifiable using the TEF - as was duly noted.

Nor was DSP referenced here, as to my knowledge no one is employing it with their consumer parametric or third octave EQ unit! As I attempted to respond to the query of the forum member who asked a question regarding the use of a parametric equalizer for such uses.

Rather the primary purpose was to present a simplified methodology for using the ETC to address the specular reflections that dominate the room for wavelengths that are small relative to the room dimensions...AS WAS SPECIFICALLY REQUESTED IN THE HT ACOUSTICS THREAD AND BY A PREVIOUS POSTER. In fact, despite the fact that none of what I presented in this post was aimed at addressing room modes, much of the objections focus on that which was specifically excluded from the presentation.
post #37 of 44 Old 07-28-08, 12:39 PM
Elite Shackster
Maverick
 
Join Date: Mar 2007
Location: Sandnes, Norway
Posts: 1,384
Re: basic theory behind what we do

I actually understood some of that!

What I don't understand is this;

If a program like REW show us the frequency response of the system before the room can have an impact (before the reflections mess it up), how can we se an impact from acoustical treatments?

I see some change with and without bass traps and 1st ref absorbtion, but much more in the waterfall plots, as expected.

If I understand it correctly the frequency response is the 'direct' sound, and the waterfall is frequency response as time passes. Is that a fair approximation?


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.



To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!

To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Done


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!
atledreier is offline  
post #38 of 44 Old 07-28-08, 12:54 PM
REW Author
 
JohnM's Avatar
John
 
Join Date: Apr 2006
Location: UK
Posts: 6,710
Re: basic theory behind what we do

Mark,

Your statement was
Quote:
In a minimum phase system, each pole in the transfer function must lie in the negative half of the complex S plane.
Pole location determines system stability ("physical realisability" if you prefer, though this is not strictly accurate as it is exceedingly simple to physically realise an unstable system complete with its right half plane poles, just introduce positive feedback -> hence the comment regarding feedback in my previous post). Both minimum phase and non-minimum phase (stable) systems have their poles in the left half plane, what distinguishes the two is the locations of their zeroes. Systems only become non-minimum phase if they have zeroes in the right half plane. I'm not trying to turn this into an argument, or to criticise you, merely correcting a statement that could mislead others reading the thread.

Quote:
And this thread began with a member focused on the notion of the goal of attempting to equalize a system until it was "ruler flat"! Obviously the limitations of addressing fundamental issues in the frequency is not well understood here. As such, a focus on such limitations is in order.
That is rather unfair on the original poster. He did title the thread as being a question about theory, spoke about the impossibility of making the response flat and specifically asked what the effect of altering the frequency response might be on the impulse response and the time domain.

Quote:
And TDS remains the prime environment of choice on TEF and EASERA, if for no other reason than its dramatic advantages with regards to noise immunity that NO other platform affords.
That's not actually correct. TDS certainly has very good noise immunity, and for equivalent sweep length to sequence length can be as much as 10dB better than using MLS, but logarithmic sweeps also have excellent noise rejection, total rejection of harmonic distortion and do not have the extreme sensitivity to clock jitter and replay/record clock matching that demands dedicated hardware to make reliable TDS measurements. None of that is intended as a criticism of the packages you listed, whose analysis capabilities are indeed superb, but merely an objective comment on the measuring techniques. TDS, MLS and an array of other measurement methods are reviewed at length in this paper: Transfer Function Measurement Using Sweeps

Regarding equalisation and what you refer to as "DSP", we are perhaps divided by our common language and using the same terms to mean different things. Equalisation, as I use the term, is the application of filters to a system - IIR, FIR or both. Most products nowadays implement that filtering in the digital domain on a DSP device, even products performing basic parametric EQ, and "DSP" in perhaps the sense you are applying the term is becoming commonplace thanks to the efforts of companies like Audyssey and Lyngdorf (formerly Tact).

You seem to have taken my comments somewhat personally, and I apologise if I have caused you offense, I assure you none was intended. My aim is only to ensure our forum members get the best information we can provide and minimise the risk of misinterpretation. It is rare for someone of your knowledge and experience to take the time to contribute and it is very much welcomed, I hope you will continue to inform.

Regards,

John
JohnM is offline  
post #39 of 44 Old 07-29-08, 10:53 AM
REW Author
 
JohnM's Avatar
John
 
Join Date: Apr 2006
Location: UK
Posts: 6,710
Re: basic theory behind what we do

Quote:
atledreier wrote: View Post
If a program like REW show us the frequency response of the system before the room can have an impact (before the reflections mess it up), how can we se an impact from acoustical treatments?

I see some change with and without bass traps and 1st ref absorbtion, but much more in the waterfall plots, as expected.

If I understand it correctly the frequency response is the 'direct' sound, and the waterfall is frequency response as time passes. Is that a fair approximation?
The frequency response that REW shows, using the default window settings, includes the contributions of the room - of necessity, since at low frequencies in particular that is our interest. How much of the room's contribution is included in the frequency response is controlled by the setting of the Impulse Response window, as the window is narrowed so more of the reflections are excluded - the trade-off is that the narrower the window, the lower the frequency resolution.

A very useful plot when looking at the effects of absorption is the Energy-Time curve (ETC). This is derived from the impulse response, peaks in the ETC after the initial peak are caused by reflections from the room's surfaces, the time delay betwen the main peak and any particular later peak correspond to the additional distance the sound has travelled in bouncing off whatever surface caused it. You can measure that on the graph, use Ctrl+right click and drag the mouse to see the delta shown as both time and equivalent distance. Looking at the result your absorber panels have on the ETC shows you how effective they are and whether they have been placed on the correct surfaces.
JohnM is offline  
post #40 of 44 Old 07-29-08, 11:14 AM
Elite Shackster
Maverick
 
Join Date: Mar 2007
Location: Sandnes, Norway
Posts: 1,384
Re: basic theory behind what we do

Yes, John, I'm aware of the ETC graph and it's usefulness. And I understand about the window now. I realized I never quite 'got' that before. Thanks!


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.



To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!

To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Done


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!


To view links or images in signatures your post count must be 10 or greater. You currently have 0 signatures.
Status: Sold!
atledreier is offline  
Reply

Bookmarks

Tags
basic , theory

Quick Reply
Message:
Options

Register Now




PLEASE COMPLETE ALL REQUIRED FIELDS BELOW... THANKS!

REQUIRED FIELDS ON THIS PAGE
YOU MUST COMPLETE ALL OF THESE

Username
Password
Confirm Password
Email Address
Confirm Email Address
Random Question
Random Question #2




User Name:
Password
Please enter a password for your user account. Note that passwords are case-sensitive.

Password:


Confirm Password:
Email Address
PLEASE READ BELOW PRIOR TO ENTERING AN EMAIL ADDRESS!

ATTENTION!

YOU MUST ACTIVATE YOUR ACCOUNT!

Activation requires you reply to an email we will send you after you register... if you do not reply to this email, you will not be able to view certain areas of the forum or certain images... nor will you be able download software.

AN INVALID EMAIL ADDRESS WILL CAUSE YOUR ACCOUNT TO BE DELETED!

See our banned email list here: Banned Email List

We DO NOT respond to spamcop, boxtrapper and spamblocker emails... please add @hometheatershack DOT com to your whitelist prior to registering or you will get nowhere on your registration.


Email Address:
OR

Log-in










Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page



Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML is not allowed!
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

 


For the best viewing experience please update your browser to Google Chrome