Gearslutz.com - View Single Post - Above the Schroeder Frequency. Diffraction.

SAC · 1st February 2010

('We' means knowledgeable folks who have taken the time to study the concepts as well as to do EXTENSIVE applied research and experimentation in the process of validating theory with respect to practical objectively verifiable results! )

The waterfall is not truly a time domain measurement. Let me try to explain…The cumulative spectral decay uses a left-slope window to successively remove the early part of the impulse response at a user determined time interval.

In other words, each slice of the CSD shows the spectrum of the portion of the impulse response that remains after application of the window. It is useful for observing portions of the spectral response that persist–namely resonances. Since the end of the time window is fixed, frequency resolution is lost with each successive sectioning (in other words, the window gets smaller as the leading edge is moved.) This reduction in resolution is apparent at low frequencies as the resolution seems to be cut off in an arc curving outward and increasing with time – with the low frequencies appearing to be cut off or tapered as the persistence in time increases, deleting initial time data.

And it makes no sense to try to determine the frequency at a point in time due to the inherent resolution limits imposed on the reciprocal time/frequency relationship (f=1/t).

Bottomline, the waterfall does not show the frequency response of a particular reflection. And the persistence or resonance at a particular frequency is not a reflection.

We are talking 2 different things here, for two different purposes. You cannot combine them as if they are the same.

The waterfall tells you nothing about the specular reflections. Just as the specular reflections ‘in’ the ETC tell you nothing about modes.

Essentially a reflection is a redirection of the acoustic energy identical to the source. Now, the acoustic impedance of a reflecting surface can alter this somewhat, but for all intents and purposes we are employing broadband reflectors and absorbing surfaces.

The reflections are rays – vectors. They have magnitude and direction.

Each peak in an ETC represents the sum of the imaginary and real energies in the system from both the impulse and doublet response. (For more on this we will need to explore the Heyser spiral – the rotating phasor diagram that contains the full information of the system under review – as imaginary by no means implies unreal! – a truly unfortunate and misleading choice of terms).

As each peak represents energy arriving at some point in time relative to the direct signal, we are presented with an overview of all of the arriving reflections from all surfaces at any specified point in space. Thus we are necessarily dealing with vector quantities relative to the measuring position.

In any case, the magnitude is specified as a measured gain relative either to some time or distance. We also know that each vector/ray exhibits an orientation angle indicative of the reflection point(s) of incidence and likewise to the signal source.

Identifying a particular reflection is easy. It appears as a discrete spike in the ETC response. Depending upon the measurement platform you are using, various information is automatically generated by moving the cursor to correspond with the peak of each reflection. At its most basic, time is displayed. A simple augmentation displays its analog travel distance as well as corresponding gain indicated by the y-axis.

Given this information, as well as knowing the location of the source and the signals reception, it becomes a trivial exercise to extrapolate the travel path, via whatever means is determined to be easiest by the user – depending in large measure on how sophisticated the platform is in automatically calculating such information and also with regards to the physical layout and accessibility of the physical space one is in.

If you are using REW or Fuzz, your choices are reduced, as they apparently only display the time, and you must perform the trivial calculation to determine the normalized time of travel (relative to the direct signal arrival time without any machine or system propagation delays). Many big words meaning that you may have to subtract the direct signal arrival time from the reflection arrival time for each reflection of interest. And then you will have to calculate the distance traveled by each reflection by multiplying the time of travel with the speed of sound. Both trivial ‘busywork’ calculations. This would be a very simple and nice feature to see added to both programs.

Typically, there will be one unique path for each reflection that is incident to boundary surfaces within the room between source and microphone receiver.

Initially you may want to use a string long enough to accommodate the farthest reflection distance (and marked in feet and perhaps in inched in the segments where you know you have reflections) until you gain a bit of pattern recognition and confidence to do this reliably, and thereafter you will quickly acquire sufficient pattern recognition to be able to carefully repeat the sweeps such that intercepting the path with a small piece of absorber intercepts the path causing the reflection to be damped in the measurement. A few auto-repeated sweeps will then allow you to quickly and accurately ‘walk’ the absorbent material along the incident pathway back to the precise spot on the boundary. In the process you move from clocking a relatively large region when near the mic to a progressively smaller and more fine tuned location the further away one moves from the mic. This is actually a generally quick and easy process.

Most platforms allow for the auto-repetition of the measurements to be performed. If REW and Fuzz do not, you will simply have to press the button a few more times.

{If one is truly serious, and especially if you generally encounter more complex spaces where access to ceiling and boundary surfaces can be difficult, you might want to investigate the polar ETC capabilities of various platforms whereby the software package can generate the 3 space coordinates suitable for one to replace the mic with a laser pointer in a transit suitable for identifying the precise points on the boundary surfaces. Additionally, you may want to investigate the means by which you can do this yourself. But this process is currently FAR beyond the scope of this thread and this forum at this point. I mention this simply to provide an example of what is possible, but admittedly overkill for what one might be doing here. But I will also say that once you become used to it, it is quite easy to become very spoiled.}

In this manner you are able to not only specifically identify each reflection of interest in the display based upon gain and/or arrival time, but you are able to specifically identify the precise path it traverses. And this ability will pay great dividends in treating the room – as it will afford you a great deal more leverage than simply identifying early reflections to be absorbed.

It is going to afford you much knowledge, understanding, and leverage over the manipulation of the entire sound field, which is a significantly important process that is generally overlooked here. It is THIS aspect of tuning a room where the intelligibility, imaging, and tonal qualities of the signal response will be adjusted. And the ETC response will form a invaluable tool for doing this.

So, with this you have the basic tools to investigate, acquire and manipulate the data.

But this is not enough. You then need to understand what needs to be done with the tools -You need a defined goal that will determine what criterion need to be met in order to turn a space into a comprehensive tuned space according to current best practices relative to the particular acoustic model you choose.

1st February 2010	#78
SAC Registered User Joined: Dec 2009 Posts: 2,622	('We' means knowledgeable folks who have taken the time to study the concepts as well as to do EXTENSIVE applied research and experimentation in the process of validating theory with respect to practical objectively verifiable results! ) The waterfall is not truly a time domain measurement. Let me try to explain…The cumulative spectral decay uses a left-slope window to successively remove the early part of the impulse response at a user determined time interval. In other words, each slice of the CSD shows the spectrum of the portion of the impulse response that remains after application of the window. It is useful for observing portions of the spectral response that persist–namely resonances. Since the end of the time window is fixed, frequency resolution is lost with each successive sectioning (in other words, the window gets smaller as the leading edge is moved.) This reduction in resolution is apparent at low frequencies as the resolution seems to be cut off in an arc curving outward and increasing with time – with the low frequencies appearing to be cut off or tapered as the persistence in time increases, deleting initial time data. And it makes no sense to try to determine the frequency at a point in time due to the inherent resolution limits imposed on the reciprocal time/frequency relationship (f=1/t). Bottomline, the waterfall does not show the frequency response of a particular reflection. And the persistence or resonance at a particular frequency is not a reflection. We are talking 2 different things here, for two different purposes. You cannot combine them as if they are the same. The waterfall tells you nothing about the specular reflections. Just as the specular reflections ‘in’ the ETC tell you nothing about modes. Essentially a reflection is a redirection of the acoustic energy identical to the source. Now, the acoustic impedance of a reflecting surface can alter this somewhat, but for all intents and purposes we are employing broadband reflectors and absorbing surfaces. The reflections are rays – vectors. They have magnitude and direction. Each peak in an ETC represents the sum of the imaginary and real energies in the system from both the impulse and doublet response. (For more on this we will need to explore the Heyser spiral – the rotating phasor diagram that contains the full information of the system under review – as imaginary by no means implies unreal! – a truly unfortunate and misleading choice of terms). As each peak represents energy arriving at some point in time relative to the direct signal, we are presented with an overview of all of the arriving reflections from all surfaces at any specified point in space. Thus we are necessarily dealing with vector quantities relative to the measuring position. In any case, the magnitude is specified as a measured gain relative either to some time or distance. We also know that each vector/ray exhibits an orientation angle indicative of the reflection point(s) of incidence and likewise to the signal source. Identifying a particular reflection is easy. It appears as a discrete spike in the ETC response. Depending upon the measurement platform you are using, various information is automatically generated by moving the cursor to correspond with the peak of each reflection. At its most basic, time is displayed. A simple augmentation displays its analog travel distance as well as corresponding gain indicated by the y-axis. Given this information, as well as knowing the location of the source and the signals reception, it becomes a trivial exercise to extrapolate the travel path, via whatever means is determined to be easiest by the user – depending in large measure on how sophisticated the platform is in automatically calculating such information and also with regards to the physical layout and accessibility of the physical space one is in. If you are using REW or Fuzz, your choices are reduced, as they apparently only display the time, and you must perform the trivial calculation to determine the normalized time of travel (relative to the direct signal arrival time without any machine or system propagation delays). Many big words meaning that you may have to subtract the direct signal arrival time from the reflection arrival time for each reflection of interest. And then you will have to calculate the distance traveled by each reflection by multiplying the time of travel with the speed of sound. Both trivial ‘busywork’ calculations. This would be a very simple and nice feature to see added to both programs. Typically, there will be one unique path for each reflection that is incident to boundary surfaces within the room between source and microphone receiver. Initially you may want to use a string long enough to accommodate the farthest reflection distance (and marked in feet and perhaps in inched in the segments where you know you have reflections) until you gain a bit of pattern recognition and confidence to do this reliably, and thereafter you will quickly acquire sufficient pattern recognition to be able to carefully repeat the sweeps such that intercepting the path with a small piece of absorber intercepts the path causing the reflection to be damped in the measurement. A few auto-repeated sweeps will then allow you to quickly and accurately ‘walk’ the absorbent material along the incident pathway back to the precise spot on the boundary. In the process you move from clocking a relatively large region when near the mic to a progressively smaller and more fine tuned location the further away one moves from the mic. This is actually a generally quick and easy process. Most platforms allow for the auto-repetition of the measurements to be performed. If REW and Fuzz do not, you will simply have to press the button a few more times. {If one is truly serious, and especially if you generally encounter more complex spaces where access to ceiling and boundary surfaces can be difficult, you might want to investigate the polar ETC capabilities of various platforms whereby the software package can generate the 3 space coordinates suitable for one to replace the mic with a laser pointer in a transit suitable for identifying the precise points on the boundary surfaces. Additionally, you may want to investigate the means by which you can do this yourself. But this process is currently FAR beyond the scope of this thread and this forum at this point. I mention this simply to provide an example of what is possible, but admittedly overkill for what one might be doing here. But I will also say that once you become used to it, it is quite easy to become very spoiled.} In this manner you are able to not only specifically identify each reflection of interest in the display based upon gain and/or arrival time, but you are able to specifically identify the precise path it traverses. And this ability will pay great dividends in treating the room – as it will afford you a great deal more leverage than simply identifying early reflections to be absorbed. It is going to afford you much knowledge, understanding, and leverage over the manipulation of the entire sound field, which is a significantly important process that is generally overlooked here. It is THIS aspect of tuning a room where the intelligibility, imaging, and tonal qualities of the signal response will be adjusted. And the ETC response will form a invaluable tool for doing this. So, with this you have the basic tools to investigate, acquire and manipulate the data. But this is not enough. You then need to understand what needs to be done with the tools -You need a defined goal that will determine what criterion need to be met in order to turn a space into a comprehensive tuned space according to current best practices relative to the particular acoustic model you choose.