Measuring a reflection

[Matthew J Poes]

I was rereading an article from Floyd Toole and noticed a picture he included of a measurement of the reflected sound off a wall treated by an acoustic panel. It was meant as a cautionary tail that treating first reflections may be worse than no treatment at all. His point being that the response of the reflected sound was far less even than a normal reflection.

This really caught my attention as an interesting idea for testing acoustic panels in situ. I’ve been playing quite a bit with different membrane materials. Given how expensive actual testing would be I mostly rely on models and simulations. I thought the idea of tuning a panels response shape to provide an even reflection of reduced amplitude would be preferable if possible.

It seems like measuring this is possible right? If I place a panel on a wall. Precisely place a speaker a set distance from the panel such that it’s distance to the panel doesn’t match its distance to any other barrier, and then take a measurement with and without the panel, the impulse response would contain the reflection. I could use gating of the impulse at that distance to look at just the response of the reflection correct? Anything I’m missing?

 

[John Mulcahy]

Sounds plausible. Ideally the reflection would be sufficiently later than the direct sound that the gating isn't so tight as to provide very poor frequency resolution and the environment damped enough that the response of the room doesn't obscure too much of the reflection.

 

[Matthew J Poes]

Sounds plausible. Ideally the reflection would be sufficiently later than the direct sound that the gating isn't so tight as to provide very poor frequency resolution and the environment damped enough that the response of the room doesn't obscure too much of the reflection.

I had some similar thinking. Initially I thought that the speaker could be placed at say 30 degrees off center from the panel and the mic 60 dregrees from that (30 degrees in the other direction. Distance for direct sound and reflected sound established such that more than 1 foot difference exists.

However I noticed an article looking to do in situ absorption measurements that noted an approach by which the speaker is say 2 meters from the panel and dead on. The Mic is 3 inches from the panel and dead on, in line with the speaker. You then subtract the impulse of the direct sound from the impulse response by copying just the first maybe 2-3 ms, inverting it, and adding it to the impulse response. This cancels the initial impulse and leaves only the reflections.

I really wanted to measure at an angle so I may try different approaches and see what gives me the most reasonable results. This also makes me question if what Toole did made sense. Maybe whatever response differences he measured were also a result of having the direct sound signal and reflection too close together and thus not allowing for a high resolution response.

I'm doing this in situ and the room is very well damped. The RT60 down to 125hz is under .4 seconds and under .2 over most of that range. I can add additional absorption temporarily as well.

 

[Silverprout]

His point being that the response of the reflected sound was far less even than a normal reflection.

In a diffuse field, if you damp a peculiar frequency, it will make a dip in the FR.

 

[Matthew J Poes]

In a diffuse field, if you damp a peculiar frequency, it will make a dip in the FR.

Hi Silverprout thanks for stopping by. I’m not sure we are talking about the same thing here.

A diffuse field does not exist in a domestic space. It does exist in a reverberation chamber used for testing panels. However the Sabin absorption coefficient is based on a change in decay of a given frequency, it’s not based on a notch in the response. It would not be possible to measure that in a reverberant chamber.

What I’m trying to do relies on opposite conditions, a minimum of additional reflections. I’m simply trying to measure the response of the reflection off the panel. As Floyd Toole points out, if a panel has a flat absorption response, you would expect the response or the reflection to equal the response of the direct signal, but reduced in level. This isn’t what happens, instead the reflection has a very uneven response. I’m trying to measure it the same way he did, by gating around the reflection.

Based on my research into this idea it seems like an in situ measurement standard for absorption is actually my best approach for getting sufficient resolution.

Here is an example:

https://www.prosoundtraining.com/2010/03/17/in-situ-absorption-coefficients-part-2/

The idea behind this is that it gives you an alternative method of characterizing the panel behavior. You can use that to tune the panel and create a better behaved absorber. The primary cause of the uneven absorption is that the absorption changes with the change in incident angle. As such I suspect that a series of measurements are needed and that a panels shape and surface treatment will matter.

There is also an argument by some that an uneven reflection is a good thing. That the reflection becomes decorrelated. I’m not so sure of that idea myself, but again with a measurement of the reflection you would have the data to characterize the behavior and test.

 

[Silverprout]

Hi Silverprout thanks for stopping by. I’m not sure we are talking about the same thing here

We are in an environement that is between the free field and the diffuse field (everything is relative to another) and we consider that the loudspeakers are in the far field (everything is relative to another, in order to be sure of that we must put the loudspeaker in the center of the room or far from the walls.)

So , you are looking for "neutral" absorbtion panels able to return an undistorded signal... amplitude ?

 

[Matthew J Poes]

We are in an environement that is between the free field and the diffuse field (everything is relative to another) and we consider that the loudspeakers are in the far field (everything is relative to another, in order to be sure of that we must put the loudspeaker in the center of the room or far from the walls.)

So , you are looking for "neutral" absorbtion panels able to return an undistorded signal... amplitude ?

Ok so a few comments. I don’t mean to come off pedantic but want to be sure we are talking about the same things. The terms you used aren’t part of a continuum with each other.

The continuum for these terms is as follows:
Near field (particle phase and particularly velocity out of phase)-far field (both are in phase and follow the inverse square law)

Diffuse field/reverberant field (there exist several reflections for a given point in space that are statistically uniform, energy density uniform)-direct field (sound substantially attributed to the direct radiation of the speaker, reflections are mostly under 50ms)

Related to the above
Free field (space in which anechoic conditions are met)-reverberant field (space in which multiple reflections exist)

A small room such as any domestic space can’t ever achieve a diffuse field by definition, it would never have energy density uniformity, but it wouldn’t be anechoic either. Small rooms don’t have acoustic field names but in practice it’s semi-anechoic in nature typically and it’s largely direct field above the Schroeder frequency (or at least outside the transition zone). Most of these terms have standard IEC definitions developed for large spaces like studios and performance spaces. These folks didn’t really care about domestic spaces when defining the terms. My experience is that it’s just referred to as small room acoustics.
http://www.acoustic-glossary.co.uk/sound-fields.htm#diffuse

In terms of this concept, see Floyd Toole’s article addressing this and the image of an acoustic panel reflection.
http://www.audioholics.com/room-acoustics/room-reflections-human-adaptation

My thought is to figure out how he made that measurement. Once I can reliably get a measurement of the reflection, I can work to create a neutral reflection response. Yes I’m looking for a smooth/flat amplitude response for the reflection.

 

[Silverprout]

Near field (particle phase and particularly velocity out of phase)-far field (both are in phase and follow the inverse square law)

My thought is to figure out how he made that measurement. Once I can reliably get a measurement of the reflection, I can work to create a neutral reflection response. Yes I’m looking for a smooth/flat amplitude response for the reflection.

I'm a technician, people are often expecting short, synthetic and simple answers from me.
Near field (particle phase and particularly velocity out of phase) : predicting what happen in the near field is a tad presumptuous IMHO.

So, to be simple:

Above the transition frequency the reflected waves are not reflected as the direct waves and usual reflective surfaces are difracting of refracting indirect waves... it blurs the sound image ?

Or, the spectra of the direct and reflected sounds are significantly different mean that the reflected soudwaves are deformed... the reflected sounds are not at the same frequencies of the direct sounds ?

 

[Matthew J Poes]

I'm a technician, people are often expecting short, synthetic and simple answers from me.
Near field (particle phase and particularly velocity out of phase) : predicting what happen in the near field is a tad presumptuous IMHO.

So, to be simple:

Above the transition frequency the reflected waves are not reflected as the direct waves and usual reflective surfaces are difracting of refracting indirect waves... it blurs the sound image ?

Or, the spectra of the direct and reflected sounds are significantly different mean that the reflected soudwaves are deformed... the reflected sounds are not at the same frequencies of the direct sounds ?

The latter is correct. That is the issue at hand. The only argument made in favor of this is that the sound is decorrelated. However Harman tested this and found that a smooth off axis response which lead to smooth reflections was vital to perceived sound quality and imaging.

 

[Matthew J Poes]

I'm a technician, people are often expecting short, synthetic and simple answers from me.
Near field (particle phase and particularly velocity out of phase) : predicting what happen in the near field is a tad presumptuous IMHO.

So, to be simple:

Above the transition frequency the reflected waves are not reflected as the direct waves and usual reflective surfaces are difracting of refracting indirect waves... it blurs the sound image ?

Or, the spectra of the direct and reflected sounds are significantly different mean that the reflected soudwaves are deformed... the reflected sounds are not at the same frequencies of the direct sounds ?

Also please don’t take offense to my clarifications. The nuance in these definitions becomes important in understanding the phenomena.

I am hoping to do a video or post on how speakers interact with rooms. I’ve started working on some images to show how a speakers radiation can be characterized as rays and how they reflect. However I want to include information on the speakers own off-axis response in explaining the issue and what impact first reflection absorption or diffusion has (diffusers couldn’t be measured in situ as far as I know but would still diffuse sound unevenly). While it seems some experts take an all or nothing view of the room/speaker system, and his reject speakers with poor off-axis response as flawed, I feel you can’t ignore their existence. For a speaker that has poor off-axis response I think absorption, even if not smooth, would be important. If a speaker has a very smooth and constant directory that falls off at extreme angles, then I can totally see how first reflection absorption would do more harm. I question the notion that a flat wall reflects evenly though. I would want to see a measurement of the reflection with a reflective surface and absorptive surface. I suspect neither will be even/smooth/flat.

 

[Silverprout]

Also please don’t take offense to my clarifications

I'm still dubitative on my level of comprehension the phenomena.
Mr Floyd Toole seems to insist on the fact that if we put absordbant materials on the walls in order to reduce reflexions, it simply absorb too much high frequencies and distort the reflexions of other frequencies above the transition one.

"absorbing materials change their absorption properties as a function of the angle of incidence"
If you take one point of the soundwaves impact on the surface and there is an absorbant material on this surface, incident rays will traverse more absorbant material than normal ones... no?

poor off-axis response

Perhaps, unadapted should be better than poor IMHO.

 

[Matthew J Poes]

I'm still dubitative on my level of comprehension the phenomena.
Mr Floyd Toole seems to insist on the fact that if we put absordbant materials on the walls in order to reduce reflexions, it simply absorb too much high frequencies and distort the reflexions of other frequencies above the transition one.

"absorbing materials change their absorption properties as a function of the angle of incidence"
If you take one point of the soundwaves impact on the surface and there is an absorbant material on this surface, incident rays will traverse more absorbant material than normal ones... no?



Perhaps, unadapted should be better than poor IMHO.

We are actually only concerned with incident rays. It’s not a comparison against a dead on strike (which would still be incident in the sense that it would bounce back the exact same direction, by his would likely never happen). He is just saying that the absorption property of a panel is different at different angles. Some of it is due to the change in depth and some of it is due to the change in surface features. Additionally he is pointing out that the commonly used fabric is actually reflective at high frequencies. He is pointing out that the method of measuring a panels absorption can’t take into account real world performance because it relies on an indirect measure. It is measuring the change in decay in the reverberant room based on multiple mics (need to ensure the measurement is that of a diffuse field). In a real room it is nether a diffuse field nor are our ears spread randomly around the room. He is noting the counterintuitive reality of panel absorption coefficients. That it absorbs 100% of the incident sound and yet a direct measure shows that is not true, that even with an absorption value of 1 (100% absorption of a hr incident wave) we can still measure a reflection.

What do you mean by unadapted? Adapted to what? I’m not sure I follow that comment.

 

[Silverprout]

We are actually only concerned with incident rays. It’s not a comparison against a dead on strike (which would still be incident in the sense that it would bounce back the exact same direction, by his would likely never happen). He is just saying that the absorption property of a panel is different at different angles. Some of it is due to the change in depth and some of it is due to the change in surface features. Additionally he is pointing out that the commonly used fabric is actually reflective at high frequencies. He is pointing out that the method of measuring a panels absorption can’t take into account real world performance because it relies on an indirect measure. It is measuring the change in decay in the reverberant room based on multiple mics (need to ensure the measurement is that of a diffuse field). In a real room it is nether a diffuse field nor are our ears spread randomly around the room. He is noting the counterintuitive reality of panel absorption coefficients. That it absorbs 100% of the incident sound and yet a direct measure shows that is not true, that even with an absorption value of 1 (100% absorption of a hr incident wave) we can still measure a reflection.

When the soundwaves hit the material surface, this surface is becoming emissive, therefore the proximity of these surfaces are in the near field domain (and a mic array is needed in order to capture the circulating and propagating waves)
IMO you shoud go out of the near field soup if you need to capture propagating waves.
Panel absorption coeficients are measured only in direct incidence conditions, these numbers are not supposed to work in random incidence conditions.

What do you mean by unadapted? Adapted to what? I’m not sure I follow that comment.

Sometimes, a narrow dispersion is not compatible with short listening distance.

 

[Allen Rumbaugh]

I think the basic problem is that "absorption panels" are non linear in their absorption ability. I have found that placing diffusion there (at the early reflections) actually improves the stereo image. Could that be because the diffuser, if large enough, is linear in it's diffusion? If you take this thinking to the extreme the Blackbird Studio C in Nashville is really something to see and hear. It is George Massenburg's mix room that took a solid year to build and has 2D diffusers all around!

 

[Matthew J Poes]

When the soundwaves hit the material surface, this surface is becoming emissive, therefore the proximity of these surfaces are in the near field domain (and a mic array is needed in order to capture the circulating and propagating waves)
IMO you shoud go out of the near field soup if you need to capture propagating waves.
Panel absorption coeficients are measured only in direct incidence conditions, these numbers are not supposed to work in random incidence conditions.



Sometimes, a narrow dispersion is not compatible with short listening distance.

The method I plan to use has been tested and standardizes as an accepted approach for in situ absorption testing. Having said that I'm fine experimenting, but I'm concerned that too much distance will cause other reflections to reduce resolution once I window. Since I'm specifically trying to measure an incident reflection and not the panels overall absorption I'm not sure I see the problem with the single mic.

As for narrow dispersion, not sure how poor off-axis response is related to narrow dispersion. A speaker can have wide dispersion with a ragged and uneven off axis response, it it can be wide and smooth. Same for narrow. If you mean that waveguides have narrow dispersion I wouldn't agree with that claim. 90 degrees is a common dispersion width for home waveguide based speakers and as I mentioned in another post, that makes the sweet spot 17 feet wide where the response is within a 6db window. Very few direct radiators could better that, most could not come close to equaling that.

Here's a peerless based direct radiator horizontal response. This is a well designed direct radiator with a smooth off axis response but it's not flat and the tilt starts at a high frequency of 3khz.

Here is a fusion 15 with good control down to 500hz or so. That sure seems preferable to me at any listening distance.

 

[Silverprout]

I think the basic problem is that "absorption panels" are non linear in their absorption ability. I have found that placing diffusion there (at the early reflections) actually improves the stereo image. Could that be because the diffuser, if large enough, is linear in it's diffusion? If you take this thinking to the extreme the Blackbird Studio C in Nashville is really something to see and hear. It is George Massenburg's mix room that took a solid year to build and has 2D diffusers all around!

I really like these diffusers, they are beautiful.

The method I plan to use has been tested and standardizes as an accepted approach for in situ absorption testing. Having said that I'm fine experimenting, but I'm concerned that too much distance will cause other reflections to reduce resolution once I window. Since I'm specifically trying to measure an incident reflection and not the panels overall absorption I'm not sure I see the problem with the single mic.

As for narrow dispersion, not sure how poor off-axis response is related to narrow dispersion. A speaker can have wide dispersion with a ragged and uneven off axis response, it it can be wide and smooth. Same for narrow. If you mean that waveguides have narrow dispersion I wouldn't agree with that claim. 90 degrees is a common dispersion width for home waveguide based speakers and as I mentioned in another post, that makes the sweet spot 17 feet wide where the response is within a 6db window. Very few direct radiators could better that, most could not come close to equaling that.

Here's a peerless based direct radiator horizontal response. This is a well designed direct radiator with a smooth off axis response but it's not flat and the tilt starts at a high frequency of 3khz. View attachment 4759


View attachment 4760
Here is a fusion 15 with good control down to 500hz or so. That sure seems preferable to me at any listening distance.

As we are in the hi-fi domain i consider that some strange dispersions as a technical choice... in a vast spectrum of parameters (everything is defendable if well argumented especially in small reverberant rooms)
A waveguide is not really narrowing the dispersion much, but it is a great solution for the physical time alignement because it avoids diffractions.
This type of alignement offers the better dispertion avaliable because it is impossible to covers the frequency range of 500 to 20K with a single direct radiator witout harmring seriously others critical parameters that are at least as important as radiation homogeneity.

 

[Matthew J Poes]

I really like these diffusers, they are beautiful.



As we are in the hi-fi domain i consider that some strange dispersions as a technical choice... in a vast spectrum of parameters (everything is defendable if well argumented especially in small reverberant rooms)
A waveguide is not really narrowing the dispersion much, but it is a great solution for the physical time alignement because it avoids diffractions.
This type of alignement offers the better dispertion avaliable because it is impossible to covers the frequency range of 500 to 20K with a single direct radiator witout harmring seriously others critical parameters that are at least as important as radiation homogeneity.

I feel like you are disagreeing what my point by making my point. I'm not sure what is meant here. The Fusion 15 is crossed over at over 1khz. The dispersion control down to 500hz is due to pattern matching of the directivity of the 15" driver and that of the waveguide at the crossover point. My speaker operates the same way.

I consider a speaker with a poor off-axis response to be an indefensible problem with the design.

 

[Silverprout]

I feel like you are disagreeing what my point by making my point. I'm not sure what is meant here. The Fusion 15 is crossed over at over 1khz. The dispersion control down to 500hz is due to pattern matching of the directivity of the 15" driver and that of the waveguide at the crossover point. My speaker operates the same way.
I consider a speaker with a poor off-axis response to be an indefensible problem with the design.

Personally i will consider every desing if the designer is able to defend his technical choices with bullet proof technical arguements and i often stay open because i can make an error (and i don' want)
I'm not against you and neither with you, you make your technical choices and if they perform as you expected you are a good designer IMO.
Directivity control is tricky and never perfect with multiway loudspeakers (in all directions)

I don't see any data about the "Fusion 15"...

 

[Matthew J Poes]

Personally i will consider every desing if the designer is able to defend his technical choices with bullet proof technical arguements and i often stay open because i can make an error (and i don' want)
I'm not against you and neither with you, you make your technical choices and if they perform as you expected you are a good designer IMO.
Directivity control is tricky and never perfect with multiway loudspeakers (in all directions)

I don't see any data about the "Fusion 15"...

I can see the image I posted.

Here it is again just in case.

I will agree that it's good to stay open minded. I too look to see what the designer says and can show about his or her design and evaluate bases on that and my ears. I've heard a lot of speakers that don't meet my standard design approach but for which I find sound great. I suppose this is the idea that there are lots of ways to skin a cat.

There are a few speakers that use a near omnidirectional patter approach or a very wide angle dispersion. They still have smooth off-axis response but it remains level outbclose to 180 degrees. In many ways it's the exact opposite of a waveguide. I think such a design could be really great as long as you are in the sweet spot and have good first reflection control.

I also wonder if diffussers at the first reflection point is a better and more linear option. I hope to test this myself (listening of course, can't measure this).

I really want to measure the first reflection point through an absorber to see if it's possible to make a panel a constant and linear absorber. We know it's not, that it's performance in a room is non-constant absorption creating linear distortions (which may also change with volume making them non-linear). It's just an idea. I know I can change a panels absorption and reflection properties with membranes, surface preperation, and curvature. That is my plan.

 

[Silverprout]

I can see the image I posted.

Here it is again just in case.
I also wonder if diffussers at the first reflection point is a better and more linear option. I hope to test this myself (listening of course, can't measure this).

I really want to measure the first reflection point through an absorber to see if it's possible to make a panel a constant and linear absorber. We know it's not, that it's performance in a room is non-constant absorption creating linear distortions (which may also change with volume making them non-linear). It's just an idea. I know I can change a panels absorption and reflection properties with membranes, surface preperation, and curvature. That is my plan.

Your idea is a fundamental physics challenge, you need a staff and a lab, but the future is coming : https://phys.org/news/2012-08-metamaterials-device-focuses-camera-lens.html

My paradigm about the quality of a desing is resumed by the number of parameters taken in account, so i would be interested by some CSD and filtrered IR.

 

[Sonnie Parker]

I have read and heard reasoning (and supposedly they have tested it) for both no type of panels (bare wall) and diffusion panels being better than absorption panels for first reflections. I have never tried anything except absorption panels, and I think it sounds fine. Wayne has measured reflections in my room and I think we came to the conclusion there was nothing alarming hurting anything.

However, supply me with the reflection panels and I will be happy to swap them out.

These are 4" thick and the small panels between the larger panels are 2" thick. Perhaps the varying sizes already count as diffusion.

 

[Matthew J Poes]

I have read and heard reasoning (and supposedly they have tested it) for both no type of panels (bare wall) and diffusion panels being better than absorption panels for first reflections. I have never tried anything except absorption panels, and I think it sounds fine. Wayne has measured reflections in my room and I think we came to the conclusion there was nothing alarming hurting anything.

However, supply me with the reflection panels and I will be happy to swap them out.

These are 4" thick and the small panels between the larger panels are 2" thick. Perhaps the varying sizes already count as diffusion.

I think it's called scattering when the arrangement is periodic and non-random. At certain frequencies the panels depth shouldn't make a difference and the same amount is absorbers. However Floyd's finding suggest to me that may not be true.

When Wayne measured reflections I am wondering if we are talking about the same thing. So you mean the reverberations in the room? Did he setup a speaker to fire at the panel and place a mic at either the incident angle or in line for a dead on measurement, then compare the direct signal to that bouncing back from the panel?

My plan at the moment is to use this method
https://www.prosoundtraining.com/2010/03/17/in-situ-absorption-coefficients-part-2/

But switch to an incident angle placement. I will try dead on like they do simply because I want to prove the approach works as expected before drawing conclusions from the measurement. Then I will move to an arrangement where the direct sound path is considerably shorter than the incident angle reflection path and try again. I will then subtract the inverted direct impulse portion from the total impulse to create a cleaner higher resolution measurement of the reflection from the panel.

My suspicion is that it will show that infact a thicker panel just absorbs more completely down lower, that it's still reflective at high frequencies. This will then allow me to experiment with how to increase the higher frequency absorption and flatten the in situ absorption curve.

Maybe I should do a little imperfect proof of concept tonight and share that to at least show what I'm doing. The problem will be that I will need to spend time working out how to cut, invert, and subtract the impulse. I could at least show the measurement setup and window around the reflection impulse. It will likely have low resolution and corruption but should be "close".

 

[Sonnie Parker]

Wayne would have to tell you more about it... but I know he used REW to do the measurements, and it was at the listening position, where I think it would matter the most. I believe we were trying to determine what reflections were affecting the sound the most at the listening position.

 

[Matthew J Poes]

Your idea is a fundamental physics challenge, you need a staff and a lab, but the future is coming : https://phys.org/news/2012-08-metamaterials-device-focuses-camera-lens.html

My paradigm about the quality of a desing is resumed by the number of parameters taken in account, so i would be interested by some CSD and filtrered IR.

You will get both, but no lab. My company is me at the moment and my resources are minimal. I hope to find enough useful information to gain interest and get help from a lab. I have a few interested parties and access to Riverbank labs, but they require that I prove this idea isn't nonsense first. Riverbank is near me but they don't work for free and I have no resources for a hunch. If I can improve the quality of absorbers simply I think that would be great.

 

[Matthew J Poes]

Wayne would have to tell you more about it... but I know he used REW to do the measurements, and it was at the listening position, where I think it would matter the most. I believe we were trying to determine what reflections were affecting the sound the most at the listening position.

Hi Sonnie, we are talking about different things I think. You are right that for understanding the sound of your room, the reflections at your listening position are what matter.

I'm trying to collect data in a normal room of actual incident absorption to characterize the real world behavior of an acoustic panel.

Here is the image from Floyd Toole on what an acoustic absorbers actual in room absorption looks like.


Look at the first and second image and panel example to see the key difference between the lab test and what they and I are testing.

You capture the reflection by windowing the impulse response of an in situ measurement of that incident angle. You could do this at the seat but the other reflections would make it harder to interpret with certainty.



Here is a drawing of what the window would sort of look like.