Early Reflections on Home Audio Performance & Solutions
The human ear cannot distinguish between direct and reflected sound unless the reflected sound is delayed at least 8 milliseconds from the original source (some studies say 10mS and others 12mS). This means that any sound the has been reflected off a wall, ceiling, floor or other object that arrive with-in 8mS will be perceived as being part of the source signal. Therefore, a tick on a hi-hat that is 10mS long will sound like it is as much a 18mS (or longer, up to 22mS, if you rely on data from other studies).
The majority of perceptual studies suggest this phenomenon does not take away from the perception of the instruments an music as far as recognizing the sounds and locating them, but they do suggest the illusion of reality is lost due to these sub-8mS reflections. Studies by speaker manufacturers and acoustic engineers have also found that these reflections are detrimental to stereo imaging as the perceived point source for the sound is smeared away from the actual location of the speaker by the arrival of reflections off surfaces that are several feet away from the original source.
In the 1980s, there was a popular trend in room acoustics to build Live End/ Dead End (LEDE) room whereby all sound above 200Hz at the front of the room where the stereo speakers were located was absorbed by acoustical treatment and all sound as the rear of the room was 100% reflected, usually diffused with RPG diffusion panels. I helped build about a dozen LEDE rooms and the results were stunning, as all early reflections up to 20mS were eliminated, yet the reverberant sound was still present to satisfy the mind’s need for open sound without dead isolation.
Then along came multi-channel movies and the LEDE concept lost favor. There was no practical way to eliminate all early reflections for 5 speakers while still having a mostly live room. A new trend popped up to address the worse sources of early reflections for stereo listening while maintaining an open sound in the room.
This new approach involved using spot treatment of acoustic panel absorbers precisely located where the initial early reflections would occur. This meant small, usually 2 foot square sound panels placed on the walls right where the sound bounced off toward the listener. At first these panels were very large, but then got smaller over time.
Here is a diagram showing the primary early reflection locations in a small, closed, rectangular room:
As you can see, there are five simple reflection points. In reality there are many more due to complex reflections that bounce of more than one surface before reaching your ears.
To treat these reflections, you can place absorbing panels or acoustic foam at the reflection point. These locations can be found be having one person sit in the primary listening spot and have another person place a small mirror on the wall. By moving the mirror around on the wall until the person in the listening position can see a speaker you can find the reflection spots. Simply placing any absorbing material at that location will reduce or eliminate the reflection (depending on the quality and bandwidth of the absorber).
The ideal would be to find any reflection, complex or simple, where the sound travels less than 8.5ft to 12ft greater distance to the listener than the source (sound travels about 1ft 1/8in per mS) and apply a small amount of absorbing material there. If you want to treat the room for multiple listeners, simply re-evaluate the ideal reflection points from each listening position and either apply more absorbing panels or larger absorbing panels that cover all reflection points.
A more common, less time consuming solution is to apply absorption behind and next to the speakers and then smaller amounts on the side walls, rear walls and ceiling only where reflections occur. This usually yields excellent results, especially if nice broadband absorbing panels are used having smooth absorption to frequency responses.
Let’s look at the nature of reflected sound.
Sound will bounce of most surfaces, but the shape of the surface will cause some sound to be reflected and some to be absorbed. Sometimes the reflected sound will be diffused to various degrees. Diffused sound is reflected in many directions instead of in an angular trajectory based on the source angle. A good comparison is with light which can be perfectly reflected in a mirror or completely diffused as when the mirror is fogged over. All of the light hitting the foggy mirror is reflected, but the reflections are in every direction and the image is not recognizable.
If we assume the walls are perfect acoustic reflectors, we can imagine a room where there is one speaker and one listener with one direct line for the sound and 4 first reflection points:
I listed the distances between the direct sound and the reflections. The direct sound travels 7 feet, or 6.2mS, before reaching the listener. The next sound arrives from the reflection off wall #2 after traveling 11 feet, or 9.8mS.
Distance traveled and time from source to listener:
Direct = 7 feet = 6.2mS
Wall #1 = 14 feet = 12.4mS
Wall #2 = 11 feet = 9.8mS
Wall #3 = 16.5 feet = 14.6mS
Wall #4 = 20.5 feet = 18.2mS
The key to using this information is the differences in arrival time and distance.
Difference between direct and reflections:
Direct = 0 feet = 0mS
Wall #1 = 7 feet = 6.2mS
Wall #2 = 4 feet = 3.6mS
Wall #3 = 9.5 feet = 8.4mS
Wall #4 = 13.5 feet = 12mS
Now, since sound loses intensity as it travels from the source to the listener (Inverse Square Law), let’s look at how the SPL changes due to the different distances the sound must travel:
Drop in SPL in dB as compared to the direct sound:
Direct = 0dB SPL
Wall #1 = -6.0dB SPL
Wall #2 = -3.9dB SPL
Wall #3 = -7.4dB SPL
Wall#4 = -9.3dB SPL
Sound reflected from wall #2 arrives at the listener’s ear 3.6mS after the direct sound as is 3.9dB quieter than the direct sound. That means that if the direct sound is 80dB SPL, the reflected sound will be 81.2dB SPL - assuming the two signals are perfectly in phase.
Combining the SPL of the direct sound (80dB) to the SPL of all the reflected sound (74dB, 76.1dB, 72.6dB and 70.7dB) results in an SPL of 82.9dB - when the frequencies are all in phase.
So, the reinforcement of the reflected sound actually increases the SPL of the speaker.
However, in real life the sound waves are not perfectly in phase after being reflected and delayed. Due to the delay, some frequencies are perfectly out of phase and will be completely cancelled out by the combination of the reflected sound to the direct sound. These root cancellation frequency can be determined simply by calculating from the difference in distance between the direct sound and the reflected sound.
For wall #1, the difference is 7 feet, which is the wavelength of 161Hz. At that frequency the reflected sound is exactly 360 degrees out phase, or basically in phase. However, at ‘ that frequency the reflected sound is exactly 180 degrees out of phase, which is the point where cancellation occurs. So, at 80.5Hz there will be no perceivable SPL level as compared to the other frequencies due to perfect cancellation. The perfect cancellation frequencies for the four reflections are:
Wall #1 = 80.5Hz
Wall #2 = 141Hz
Wall #3 = 118.7Hz
Wall #4 = 83.5Hz
Keep in mind, however, that the drop in SPL will reduce the intensity of the cancellation. After all subtracting dB is similar to combining dB. Since the SPL of the sound reflected from Wall #4 is 9.3dB lower than the direct sound (80dB), the reduction in output at 83.5Hz is actually 79.5dB. Whereas, the cancellation intensity of the reflection of Wall #2, which is a mere 3.9dB less than the direct sound, is 77.7dB SPL.
See how confusing this can all be? It is not simple math, either.
Now, let’s take a closer look at the nature of the reflected sound.
Speakers do not produce the same output in all directions. The response of the output off axis is different from the on axis sound. This is caused by many factors, but most often it is the nature of most transducers to start narrowing their off axis dispersion as the wavelength of the audio being reproduced gets close to the width of the diaphragm. For my speakers with 5’ inch mid/woofers and ‘ inch tweeters, the sound starts slightly beaming at approximately 850Hz and then returns to on axis levels at the crossover to the tweeters up to the frequency where the tweeter starts beaming at about 9 kHz.
Now, the angle of the reflected sound that is radiating from the speaker will affect the response changes due to off axis irregularities. For instance, the sound reflecting off Wall #1 is limited, for the most part, to the frequencies above the directionality of the speaker baffle. My speakers are 8’ wide, so the frequency where the sound stops traveling behind the speaker is about 423Hz. That means the sound reflected off Wall #1 is somewhat limited to frequencies below 423Hz (this does not take into account diffraction off the edges of the speaker baffle).
Here is a response chart that shows the approximate difference from the direct sound to the frequency response of the various wall reflections:
When doing the calculations on how the sound at your ear is affected by reflected sound, you have to take into account the amount of energy at a given frequency that is being reflected. Clearly, the sound reflecting off Wall #1 will not contribute to the levels of sound heard at your ear at 5kHz since the levels of reflected sound at 5kHz is so low it will not contribute to what you hear.
If we combine the outputs of all these reflected sounds with the direct sound (assuming all the reflections are perfectly in phase), we get this frequency response:
The brown line shows the combination of all the acoustical SPL levels combined if the signals are in phase. Of course, the reflected sound and the direct sound will not be in phase at every frequency, but I cannot possibly cover all of that in a thread like this.
Keep in mind we have only been discussing the reflections off of four walls. I have not tried to include the ceiling or the floor. Nor have I attempted to discuss the absorptive and diffusive properties of the wall surface.
Now, let’s apply both of the concepts we have discussed, reflection cancellations and variation in response, to floor reflections.
I have on many occasions told everyone that hardwood floors are better than carpet for Home Theaters and Studios. People have been confused by this, but let me put that into perspective with some science and logic.
Here is a diagram of my room from the side with the distances listed for the ceiling and floor reflections:
As in the first post above, here is the pertinent data:
Direct = 7 feet = 0mS = 0dB
Floor = 10 feet = 2.7mS = -3.1dB SPL = 188Hz (cancellation)
Ceiling = 17.2 feet = 9.0mS = -7.8dB SPL = 55.3Hz (cancellation)
If we chart the responses of just the floor and direct, they look like this:
However, if we take into account the effect of the out of phase reflection at 188Hz and the comb filter effect of the carpet’s absorption coefficients, then the contribution of the floor reflection is much more off the wall. The chart below is an example (though not perfectly accurate) of what I am talking about and how the combined output will look:
As you can see, the comb effect of the carpet and the cancellation at 188Hz makes for a very ragged response from the main speaker. If the floor were 100% reflective, then the response in the midrange will be much smoother, while not perfectly flat due to the off axis issues I mention above.
All this time I have been using diagrams of my new HT with the speaker and listener placed as they are in real life. Now let me get into what I did in my room to deal with all the issues I have described.
For the primary wall reflections, I applied sound absorbing panels that are effective down to below 200Hz. The absorption material on Wall #1 is effective down to below 150Hz. Here’s a diagram:
For the ceiling I applied a different solution in the form of a reflector that reflects most sound down to about 300Hz and absorbs some sound in the 100Hz to 300Hz range while absorbing lots of sound below 100Hz. I did this to prevent the room from becoming too dead, which results in an unnatural boominess. Here’s a diagram of that treatment:
Notice that no sound reflected off the ceiling treatment will reach the listener’s ears until it has bounced off the rear wall.
To improve the response of the floor and ceiling reflections, I designed my speakers using a MTM arrangement with the midrange units spaced according to D’Appolito’s technique which greatly narrows the vertical dispersion, thus reducing the output levels that reach the floor and the ceiling in the range from about 280Hz up to the crossover to the tweeter. To get a similar effect from the tweeter, I added a felt ring to the baffle to absorb off axis sound off the tweeter. The shape of the felt was carefully designed to create a tweeter off axis response which is nearly identical to the mid/woofers in both the vertical and horizontal planes.
Finally, to demonstrate visually what the effects of the treatments have done, I created another response chart that shows the response characteristics of the reflections that are arriving from all six primary reflection points. The responses of each reflected sound has been tailored to reflect the treatment used AND the controlled off axis characteristics of the speaker. I also removed the notches in the bass response since the bass speaker in my speaker setup is close to the floor and not likely to have the out of phase effects I discuss above and calculated based on my tweeter location being the source of all sound originating at the speaker.
Here is the example chart:
Notice that the ripple effect from the carpet comb filter issue has been removed. This is due to the controlled dispersion of the D’Appolito arrangement and the tweeter felt ring.
So far I have only discussed the primary, first order reflections. Maybe another day I will create a part three of this series discussing reverb and higher order reflections.
One of the things I completely failed to mention is that below about 150Hz, the boundary effect will reinforce the bass from the speaker just because it is near a reflective surface. When frequency’s 1/4 wavelength is more than the distance between the speaker and a reflective surface, the combined output of the reflection and the source is 6dB louder than the source alone. So, if a speaker is within 3 feet of a wall, the output below about 94Hz will be 6dB louder than if the speaker is not by the wall. Since the speaker is already sitting on the ground, there is a 6dB gain for the floor reflection as well. So, placing a speaker in a corner will create an 18dB boost in the SPL over the same speaker being in a true anechoic chamber.
This detail means that my charts above are only useful above about 150Hz since I did not include the boundary reinforcement in my simulations.
As a fun side note, the effect I just described also applies to listeners. If you are sitting in a corner, the bass SPL could be increased by 18dB.
FAQ:
What methods are you using to determine your reflections?
I use a mirror and a lamp. I place the lamp bulb right where my head would be aimed at the wall about where I think the reflection should be. Then I place a simple handheld makeup mirror flat against the wall and move it around until the reflected light hits the tweeter of the speaker. I then mark that spot on the wall. After that I move mirror until I have the reflection point for each driver in the speaker and mark the wall appropriately.
I’m using a variation of your design for the one inch OC panels for absorption do you also have plans for diffusion/diffusers?
If you check out my HT Construction Thread (LINK), you''ll notice I have more diffusors than I do absorbers. I plan on adding some more to the ceiling over the head of the listeners. While I was writing this thread and creating all the charts and diagrams, I was planning on creating a part three where are I go into late reflections (as opposed to early reflections), RT60 (room decay) and reverb. That thread would discuss the purpose and need for diffusors. The lack of activity on this thread is not encouraging, so I may never work on a part three.
What methods do you use to determine the placements of diffusers?
I spend lots and lots of time finding all the dual and triple reflection points on the walls and ceiling and place the diffusors at those points to convert what would be late echoes hitting my head into smooth reverb. It works well.
I was personally going to just use the mirror method for reflections; I don't have much faith in that system.
The mirror system is the finest available because you can quickly get the perfect location for each speaker reflection on every wall and ceiling.
SECONDARY REFLECTIONS
Secondary reflections are the reflections that could bounce off two or more walls and still arrive at the listener's ear pretty much sounding the same as the direct sound, just delayed and lower in SPL. Since all of this happens in a three dimensional environment, I can only diagram the most basic forms of secondary reflections in my two dimensional models.
Here is a diagram of the four most important simple secondary reflections using only the walls as reflectors:
Here is a diagram of the only simple secondary reflection utilizing the ceiling and rear wall:
I could go through the math to calculate the delay in mS and the reduction in SPL based on distance traveled, but I won't at this time.
The point is, these sounds will reach the listener's ears as distinct echoes, not as smooth decay or such.
I have two sample audio files demonstrating the difference between a clap in my HT with lots of echo and the same room after being partially acoustically treated presenting reverb instead of echo.
Part of the treatment process is to diffuse the secondary reflections (as well as all non-absorbed reflections) to prevent the sound from fully and completely reflecting directly back at the listener. If the sound is diffused, what would originally reflect right to the listener will be split up and reflect off all the room's surfaces and eventually hit the listener's ears at every possible delay interval imaginable. This is reverb.
Here are a couple of sample diagrams of how I diffused the basic secondary reflections in my HT:
Note that the sample audio files were recorded before I applied all of the diffusion to the walls, so the final sound is actually more diffuse that the so called "treated" clap file suggests.
Now that I have explained the nature of the reflections - their properties in SPL, phase shift, delay, and such - let's look at how we measure them in a way that can be translated in what we hear.
The standard method for measuring room reflection is the RT60 chart. This chart measures the echo decay of the room in time and SPL.
All of the data I have used above suggests a RT60 type chart such as this:
I limited the chart to the real reflections we have already diagramed and a 12dB window. In a proper chart, the SPL range would extend down about 30dB and the time axis would last for about 60mS.
Notice that the initial sound is followed by 5 reflections that are relatively loud within 10mS. Our brains will interpret those first 5 reflections as part of the original signal and give the impression that a tick on a ride cymbal in a recording is actually nearly 9 mS long. All of the adjectives I used above apply to the sound represented in this chart.
Now, let's look at a chart of the sound we could expect in the treated room:
Notice that aside from the floor reflection, the entire reflection pool has been eliminated and instead of echoes happening after 12mS, there is a smooth reverb arriving at the listener's ear from the reflections being diffused and reflected off other surfaces and diffused again.
The treated room will wound more natural and detailed to the listener and present less distortion to the mind.
THIS VERY THING IS WHY I CLAIM THAT ACOUSTICS PLAY AS BIG A ROLE IN SOUND QUALITY AS THE SPEAKERS.
The best speakers in the world placed in an untreated room will be perceived as being fat and blurred when compared to average speakers in a well-designed listening environment.
As a side note, in my own room, the floor reflection is actually much less loud and limited to lower frequencies than we perceive as detail due to my use of two midrange drivers in a D'Appolito arrangement and the felt rings around the tweeters. So, the reality of that floor echo is much different for me.
Here is a chart that more accurately represents how my RT60 histogram might look:
FAQ:
1) A question about carpet vs. hardwood floors. I always thought carpet was better too. But won''t hardwood be more of a reflective surface? How would you treat the primary reflections on the floor? Wouldn''t a rug have the same effect as carpet?
I choose not to treat the floor because it is impractical. Instead, to get the most natural sound, a fully reflective floor is better than a semi-reflective floor, especially with the comb filtering that carpet brings to the table. Different carpets and rugs have different effects, so results will vary. In some of the very high-end mastering/mising studios I have contracted on, we did treat the floor reflections the same way we treated the wall reflections with large acoustic panels - either absorbers or diffusors. This is possible, and it can look nice, but it makes that part of the floor useless for any other purpose.
2) With my speaker's waveguide and my mains using dual woofers, does that help reduce the comb effect the carpet has?
If both of those woofers are operating in the midrange, and assuming they spaced them to take advantage of the D'Appolito principle of controlled dispersion, then you make have less of a problem with the midrange. However, the wave guide for the tweeter is too shallow to truly eliminate vertical transmission of sound. The design of the wave guide on your speakers is merely a less effective way to cut down on edge diffraction. Sound from the tweeter will still reflect off the floor.
3) What about sound coming from the surround speakers. How should one treat a room for these? Even though most movies and surround music dosen't have much that comes from these speakers some SACD/DVDA''s do have A LOT of music and sound coming from them (Pink Floyd DSOTM, Dr. Chesky's 5.1 surround sound show, David Sanborn "Time Again", to name a few).
My article on acoustics out at http://www.soundenvironments.com thoroughly covers that subject. I discuss the principles of a "LEDE," or live end dead end, listening room and how it can be modified to work with a surround sound setup.
4) Since the panels I'm building are portable could a possible room arrangement be to have only 1 on the left and right walls at the primary reflection point (btw... the right wall has an walk way. It may not be in the primary reflection point but if it is what advantage do I have to treating it?) And use the other two panels for absorption in the front of the room? My speaker's baffle is 9.5" wide for the mains and rears. 7 7/8" for the center channel.
I would experiment. The width of the baffle suggests that most sound above 350Hz will not emanate behind the speaker, so unless you have significant edge diffraction, the reflections off the front wall will all be below about 350Hz. The only way to know is to stick a baffle there and see what happens. As for the side walls, you need to place a baffle on both sides of the room in a mirrored arrangement, even if there is a walkway. For two-channel listening it is essential that the sound from the left & right be identical. The second place I would experiment with the two extra panels is on the cross room reflection points for the opposite speaker. In other words, the left speaker’s reflection point off the right wall.
Here are the absorption coefficients for 2" thick 703:
Type in. (mm) 125 250 500 1000 2000 4000 NRC
703 Plain (51) 0.17 0.86 1.14 1.07 1.02 0.98 1.00
703, FRK (51) 0.63 0.56 0.95 0.79 0.60 0.35 0.75
How low does the effectiveness go for the 2" thick OC-703? It looks like 250 Hz as well with some absorption down to 125 Hz.
There is significant absorption down to about 225Hz with very little below that. If you can place the fiberglass up to 2" from the wall, the low frequency effectiveness can be extended down to about 180Hz. This is what I did with my panels and the results are pretty amazing.
The formula for determining the frequency at which a baffle will not direct sound forward anymore is:
C = Speed of sound = 1128ft/S
W = Width in Feet = 0.79166666, or 9.5 inches
C / W = F’
1128 / 0.7916666 = 1424.84Hz
That is the same frequency as the width of the baffle. Above that frequency all sound will radiate to the front of the baffle (except of diffracted and bend sound). At 1/4 wavelength all sound will bend around the baffle.
1424.84 / 4 = 356.21Hz
So, below 356Hz, all sound will disperse in all directions from the baffle. Between the root frequency (1.4 kHz) and the quarter wavelength (350Hz) the sound will start bending around the baffle in a curve shape pattern. So at 500Hz more sound will come around the baffle than at 1 kHz.
See how that works?
As for lower frequency absorption, the best tools are "bass" traps. Ethan Winer has plans for simple lower midrange traps that are more effective tha most. If you want a faster solution, using FRK compressed fiberglass with the foil on the outside, facing into the room, can be pretty effective. But then you get reflection at the higher frequencies, so ther are trade-offs.
I have mentioned the issue of our brains not being able to tell a direct sound from an echo if the echo arrives less than 10mS after the direct sound. In other places I have suggested that difference needs to be greater, like 20mS (or more), in order to assure that you can inherently hear the difference between a direct sound and an echo.
The reality, like nearly everything, is more variable.
When the echo comes from the side or rear in relation to the direct sound being in front of you, the typical line in the sand between our perception of direct and echo it about 10mS - and it varies from person to person.
However, if the echo is arriving from the same area as the direct sound, it takes a longer delay before our brains can tell any difference between it and the direct sound. In the situation that the echo is coming from the same area as the direct sound, the typical delay requirement to tell a difference is generally accepted as being 20mS.
So, the cool looking decay chart I posted above with the various arrival times of echoes from the various walls is not truly sufficient when tuning a room properly - especially small rooms. The direction of the reflections in relation to the direct sound needs to be known in order to know the impact they will have on the perceived accuracy of the direct sound.
1) What about diaphragmatic absorbers? I have a question, why fill the air gap with fiberglass? (I know there should be a .25 or .5'' gap between the plywood and fiberglass but why have fiberglass inside the panel? Simply designing a diaphragm panel of .25'' plywood (.74 lb/sq ft) and placing it on 2 by 4s would create an are gap of 3.75'' and a frequency of resonance of 102Hz. I realize the fiberglass helps absorb more of the sound wave at lower freqs but normally glass fiber loses its effectiveness when the wavelength of the sound is much larger than the thickness of the fiberglass (i.e. At 102Hz the wavelength is around 11 feet.) How does the fiberglass help if it in itself can't do a lot of absorbing at such low freqs.
- The fiberglass slows the speed of sound and thus lowers the tuning frequency of the trap. It also dampens the sound by absorbing a portion of the pressure wave. Finally, it widens the Q of the total panel. If you leave the fiberglass out, the panel isn't really absorbing much other that the specific frequency that makes the diaphragm vibrate.
2) Polycylindrical Absorbers. Are these practical for home use?
- Not in most cases.
3) Membrane absorbers. The master handbook of acoustics gives a simple example of a membrane being buliding insulation with kraft paper toward the sound. They also say that building insulation has not caught on as an inexpensive absorbing material due to cosmetic and protective covers required. Hmm.... I might have to think about using that as a bass absorber one day. Or is it even worth considering? I also have a feeling you aren't referring to this by Membrane absorbers.
- That is one solution. When I said membrane, I was refering to what the Handbook refers to as "Diagphramic" absorbers.
4) Helmholtz Resonators. I could use a lot of 10oz coca cola bottles to absorb 5.9 Sabins at 185 Hz each!! Unfortunately they have a bandwidth of .67 Hz so only 1 tangential mode (185.5 Hz) and 1 oblique mode (184.6Hz) would be affected by adding a coca cola bottle. I realize bottles and pots are not good Helmholtz resonators but I found the above amusing. However reading about these makes it seem hard to build properly (need a high Q factor (and this seems to have advantages, being higher Q''s have less losses, and disadvantages, being higher Q''s have reverb problems) and need precise placement to work.
- The two solutions below are variations of a simple Helmholtz resonator.
5) Perforated Panel Absorbers. These look very promising. Could I build a free standing model?
Yes you could. In fact there are hundreds of samples of these in the commercial industry and you see them all over the place. My ceiling reflectors are a variation on this design with an open back, which increases the tuning frequency.
6) Slat Absorbers hmm... the book doesn't mention much about these
- These are the most common version of Helmholtz resonators used in recording studios, especially drum room.
- You didn't mention hanging panels or huge masses of absorbing material in corners.
Which are generally the best routes to go with in a home to absorb low-mid freqs? Are any of these practical to make as a freestanding panel/resonator?
- I like my variation on that uses a monocylindrical front diffusor and an open back filled with fiberglass. In my less than perfect tests, my panels absorb in the 200Hz to the 600Hz range quite well, considering their size.
More on topic, is there any way to diffuse bass? (under 100hz?)
- Not in a typically sized room
Your diffusion panels are monocylindrical correct? These work very well to diffuse sound if it hits at 0 degrees. Don't they create a comb filter effect if sound hits them at a 45 degree angle? Does this cause any audible effect?
- Not really - at least not audibly. You are looking at this in terms of a monocylindrical unit on a very large wall. My walls are not large, nor is my room. The size and shape of my diffusors vary (to reduce both similarities in low frequency cutoff, vary the dispersion pattern of the high frequencies, and change the tuning of the cavity so the absorption frequencies vary). Now, in the truest sense, there are frequencies where there will be some comb-filtering, but these are limited and not consistent between units. My primary goal was to diffuse the midrange frequencies - from about 800Hz to 7 kHz.
The Quad-Diffusors are better than what I did, but I cannot afford to buy proper quad-diffusors and building my own would be too heavy to mount easily and removably.
The 2D quad-diffusors (Skylines, Art Diffusors) are better than the 1D versions in smaller rooms, but they have a narrower "Q'' and thus limited response.
My diffusors are more simple. They create a non-parallel, non-linear surface which will diffuse sound:
For frequencies above the wavelength of the diameter of the monocylindrical units I built, the sound will be diffused. So, if the diameter of the unit is 36" the lowest frequency which is diffuses will be 376Hz, however the perfect diffusion starts one octave above that at 752 Hz.
Now, these same units are also basically Helmholtz dampened resonators in that air pressure that goes around the front panel will be resonated in the cavity behind the front. If the width of the unit s 24", that means frequencies below about 280Hz will enter behind the unit and be absorbed by the mass of fiberglass inside the unit. Also, the shape will provide some basic absorption at the diameter (564Hz). Then there is the sound that can travel through the front material which will be absorbed by the mass of fiberglass.
Here are my assumed and poorly calculated properties for the larger diffusors I made for the side walls (above the absorption panels):
I was able to measure some of this data based on ground plane measurements and a test speaker, but as much is assumed performance via mathematical models as it is measured.
While I have described two easy methods for quickly finding the reflection acoustic points on the walls (or ceiling & floor) between the listener and the speakers, I have received numerous emails and PMs asking me to explain it again. It seems the more I explain these methods, the harder they sound to the people asking for my help.
So, today I made a few charts showing how to use a mirror to find the first reflections of sound between a speaker and the listener.
First, here is a diagram showing a room with a couch and two speakers. I added rays to the diagram to represent the acoustic energy which needs to be absorbed (or diffused) in order to reduce or eliminate the first order reflections which this entire series of articles attempted to expose.
This next drawing shows a simple way a single person with a tradition lamp can fairly easily find the points on the wall where the first order reflections are location so he can address them with absorption materials:
By placing a table lamp or adjustable height floor lamp where the speaker should be with the bulb at the very spot where the tweeter on the speaker should be located, one can simply move a hand mirror around on the wall to find the spots on the couch where the light is perfectly reflected. By marking the wall where the mirror perfectly reflects the light one will know the center of the area which requires acoustic treatment.
The lamp method can be inverted where the lamp is placed where the listener's head would be located and moving the mirror around until the light is perfectly reflected on the tweeter.
Here is another diagram using a mirror and two people:
This method is by far the fastest and doesn't call for any extra tools nor moving the speaker out of the way. One person sits in the various listening positions and another person moves the mirror around on the wall until the seated person can perfectly see the tweeter.
In both methods is it imperative that the mirror be held perfectly flat against the wall so that the glass is perfectly parallel to the wall. Otherwise using a mirror is pointless.
It is also important to treat the reflection from the speaker on the opposite side from each wall:
It is also important to consider the vertical aspects of treatment.
While the most critical reflections to treat are those from the tweeter, it is very advantageous to treat the woofer reflections as well.
Here is a diagram showing what might be required if the speakers are tower-sized with several woofers:
Here's some historical research to explain the "blurring" phenomena of early reflections:
Haas Effect
Also called the precedence effect - describes the human psychoacoustic phenomena of correctly identifying the direction of a sound source heard in both ears but arriving at different times. Due to the head's geometry (two ears spaced apart, separated by a barrier) the direct sound from any source first enters the ear closest to the source, then the ear farthest away. The Haas Effect tells us that humans localize a sound source based upon the first arriving sound, if the subsequent arrivals are within 25-35 milliseconds. If the later arrivals are longer than this, then two distinct sounds are heard. The Haas Effect is true even when the second arrival is louder than the first (even by as much as 10 dB). In essence we do not "hear" the delayed sound. This is the hearing example of human sensory inhibition that applies to all our senses. Sensory inhibition describes the phenomena where the response to a first stimulus causes the response to a second stimulus to be inhibited, i.e., sound first entering one ear cause us to "not hear" the delayed sound entering into the other ear (within the 35 milliseconds time window). Sound arriving at both ears simultaneously is heard as coming from straight ahead, or behind, or within the head. The Haas Effect describes how full stereophonic reproduction from only two loudspeakers is possible.
The human ear cannot distinguish between direct and reflected sound unless the reflected sound is delayed at least 8 milliseconds from the original source (some studies say 10mS and others 12mS). This means that any sound the has been reflected off a wall, ceiling, floor or other object that arrive with-in 8mS will be perceived as being part of the source signal. Therefore, a tick on a hi-hat that is 10mS long will sound like it is as much a 18mS (or longer, up to 22mS, if you rely on data from other studies).
The majority of perceptual studies suggest this phenomenon does not take away from the perception of the instruments an music as far as recognizing the sounds and locating them, but they do suggest the illusion of reality is lost due to these sub-8mS reflections. Studies by speaker manufacturers and acoustic engineers have also found that these reflections are detrimental to stereo imaging as the perceived point source for the sound is smeared away from the actual location of the speaker by the arrival of reflections off surfaces that are several feet away from the original source.
In the 1980s, there was a popular trend in room acoustics to build Live End/ Dead End (LEDE) room whereby all sound above 200Hz at the front of the room where the stereo speakers were located was absorbed by acoustical treatment and all sound as the rear of the room was 100% reflected, usually diffused with RPG diffusion panels. I helped build about a dozen LEDE rooms and the results were stunning, as all early reflections up to 20mS were eliminated, yet the reverberant sound was still present to satisfy the mind’s need for open sound without dead isolation.
Then along came multi-channel movies and the LEDE concept lost favor. There was no practical way to eliminate all early reflections for 5 speakers while still having a mostly live room. A new trend popped up to address the worse sources of early reflections for stereo listening while maintaining an open sound in the room.
This new approach involved using spot treatment of acoustic panel absorbers precisely located where the initial early reflections would occur. This meant small, usually 2 foot square sound panels placed on the walls right where the sound bounced off toward the listener. At first these panels were very large, but then got smaller over time.
Here is a diagram showing the primary early reflection locations in a small, closed, rectangular room:
As you can see, there are five simple reflection points. In reality there are many more due to complex reflections that bounce of more than one surface before reaching your ears.
To treat these reflections, you can place absorbing panels or acoustic foam at the reflection point. These locations can be found be having one person sit in the primary listening spot and have another person place a small mirror on the wall. By moving the mirror around on the wall until the person in the listening position can see a speaker you can find the reflection spots. Simply placing any absorbing material at that location will reduce or eliminate the reflection (depending on the quality and bandwidth of the absorber).
The ideal would be to find any reflection, complex or simple, where the sound travels less than 8.5ft to 12ft greater distance to the listener than the source (sound travels about 1ft 1/8in per mS) and apply a small amount of absorbing material there. If you want to treat the room for multiple listeners, simply re-evaluate the ideal reflection points from each listening position and either apply more absorbing panels or larger absorbing panels that cover all reflection points.
A more common, less time consuming solution is to apply absorption behind and next to the speakers and then smaller amounts on the side walls, rear walls and ceiling only where reflections occur. This usually yields excellent results, especially if nice broadband absorbing panels are used having smooth absorption to frequency responses.
Let’s look at the nature of reflected sound.
Sound will bounce of most surfaces, but the shape of the surface will cause some sound to be reflected and some to be absorbed. Sometimes the reflected sound will be diffused to various degrees. Diffused sound is reflected in many directions instead of in an angular trajectory based on the source angle. A good comparison is with light which can be perfectly reflected in a mirror or completely diffused as when the mirror is fogged over. All of the light hitting the foggy mirror is reflected, but the reflections are in every direction and the image is not recognizable.
If we assume the walls are perfect acoustic reflectors, we can imagine a room where there is one speaker and one listener with one direct line for the sound and 4 first reflection points:
I listed the distances between the direct sound and the reflections. The direct sound travels 7 feet, or 6.2mS, before reaching the listener. The next sound arrives from the reflection off wall #2 after traveling 11 feet, or 9.8mS.
Distance traveled and time from source to listener:
Direct = 7 feet = 6.2mS
Wall #1 = 14 feet = 12.4mS
Wall #2 = 11 feet = 9.8mS
Wall #3 = 16.5 feet = 14.6mS
Wall #4 = 20.5 feet = 18.2mS
The key to using this information is the differences in arrival time and distance.
Difference between direct and reflections:
Direct = 0 feet = 0mS
Wall #1 = 7 feet = 6.2mS
Wall #2 = 4 feet = 3.6mS
Wall #3 = 9.5 feet = 8.4mS
Wall #4 = 13.5 feet = 12mS
Now, since sound loses intensity as it travels from the source to the listener (Inverse Square Law), let’s look at how the SPL changes due to the different distances the sound must travel:
Drop in SPL in dB as compared to the direct sound:
Direct = 0dB SPL
Wall #1 = -6.0dB SPL
Wall #2 = -3.9dB SPL
Wall #3 = -7.4dB SPL
Wall#4 = -9.3dB SPL
Sound reflected from wall #2 arrives at the listener’s ear 3.6mS after the direct sound as is 3.9dB quieter than the direct sound. That means that if the direct sound is 80dB SPL, the reflected sound will be 81.2dB SPL - assuming the two signals are perfectly in phase.
Combining the SPL of the direct sound (80dB) to the SPL of all the reflected sound (74dB, 76.1dB, 72.6dB and 70.7dB) results in an SPL of 82.9dB - when the frequencies are all in phase.
So, the reinforcement of the reflected sound actually increases the SPL of the speaker.
However, in real life the sound waves are not perfectly in phase after being reflected and delayed. Due to the delay, some frequencies are perfectly out of phase and will be completely cancelled out by the combination of the reflected sound to the direct sound. These root cancellation frequency can be determined simply by calculating from the difference in distance between the direct sound and the reflected sound.
For wall #1, the difference is 7 feet, which is the wavelength of 161Hz. At that frequency the reflected sound is exactly 360 degrees out phase, or basically in phase. However, at ‘ that frequency the reflected sound is exactly 180 degrees out of phase, which is the point where cancellation occurs. So, at 80.5Hz there will be no perceivable SPL level as compared to the other frequencies due to perfect cancellation. The perfect cancellation frequencies for the four reflections are:
Wall #1 = 80.5Hz
Wall #2 = 141Hz
Wall #3 = 118.7Hz
Wall #4 = 83.5Hz
Keep in mind, however, that the drop in SPL will reduce the intensity of the cancellation. After all subtracting dB is similar to combining dB. Since the SPL of the sound reflected from Wall #4 is 9.3dB lower than the direct sound (80dB), the reduction in output at 83.5Hz is actually 79.5dB. Whereas, the cancellation intensity of the reflection of Wall #2, which is a mere 3.9dB less than the direct sound, is 77.7dB SPL.
See how confusing this can all be? It is not simple math, either.
Now, let’s take a closer look at the nature of the reflected sound.
Speakers do not produce the same output in all directions. The response of the output off axis is different from the on axis sound. This is caused by many factors, but most often it is the nature of most transducers to start narrowing their off axis dispersion as the wavelength of the audio being reproduced gets close to the width of the diaphragm. For my speakers with 5’ inch mid/woofers and ‘ inch tweeters, the sound starts slightly beaming at approximately 850Hz and then returns to on axis levels at the crossover to the tweeters up to the frequency where the tweeter starts beaming at about 9 kHz.
Now, the angle of the reflected sound that is radiating from the speaker will affect the response changes due to off axis irregularities. For instance, the sound reflecting off Wall #1 is limited, for the most part, to the frequencies above the directionality of the speaker baffle. My speakers are 8’ wide, so the frequency where the sound stops traveling behind the speaker is about 423Hz. That means the sound reflected off Wall #1 is somewhat limited to frequencies below 423Hz (this does not take into account diffraction off the edges of the speaker baffle).
Here is a response chart that shows the approximate difference from the direct sound to the frequency response of the various wall reflections:
When doing the calculations on how the sound at your ear is affected by reflected sound, you have to take into account the amount of energy at a given frequency that is being reflected. Clearly, the sound reflecting off Wall #1 will not contribute to the levels of sound heard at your ear at 5kHz since the levels of reflected sound at 5kHz is so low it will not contribute to what you hear.
If we combine the outputs of all these reflected sounds with the direct sound (assuming all the reflections are perfectly in phase), we get this frequency response:
The brown line shows the combination of all the acoustical SPL levels combined if the signals are in phase. Of course, the reflected sound and the direct sound will not be in phase at every frequency, but I cannot possibly cover all of that in a thread like this.
Keep in mind we have only been discussing the reflections off of four walls. I have not tried to include the ceiling or the floor. Nor have I attempted to discuss the absorptive and diffusive properties of the wall surface.
Now, let’s apply both of the concepts we have discussed, reflection cancellations and variation in response, to floor reflections.
I have on many occasions told everyone that hardwood floors are better than carpet for Home Theaters and Studios. People have been confused by this, but let me put that into perspective with some science and logic.
Here is a diagram of my room from the side with the distances listed for the ceiling and floor reflections:
As in the first post above, here is the pertinent data:
Direct = 7 feet = 0mS = 0dB
Floor = 10 feet = 2.7mS = -3.1dB SPL = 188Hz (cancellation)
Ceiling = 17.2 feet = 9.0mS = -7.8dB SPL = 55.3Hz (cancellation)
If we chart the responses of just the floor and direct, they look like this:
However, if we take into account the effect of the out of phase reflection at 188Hz and the comb filter effect of the carpet’s absorption coefficients, then the contribution of the floor reflection is much more off the wall. The chart below is an example (though not perfectly accurate) of what I am talking about and how the combined output will look:
As you can see, the comb effect of the carpet and the cancellation at 188Hz makes for a very ragged response from the main speaker. If the floor were 100% reflective, then the response in the midrange will be much smoother, while not perfectly flat due to the off axis issues I mention above.
All this time I have been using diagrams of my new HT with the speaker and listener placed as they are in real life. Now let me get into what I did in my room to deal with all the issues I have described.
For the primary wall reflections, I applied sound absorbing panels that are effective down to below 200Hz. The absorption material on Wall #1 is effective down to below 150Hz. Here’s a diagram:
For the ceiling I applied a different solution in the form of a reflector that reflects most sound down to about 300Hz and absorbs some sound in the 100Hz to 300Hz range while absorbing lots of sound below 100Hz. I did this to prevent the room from becoming too dead, which results in an unnatural boominess. Here’s a diagram of that treatment:
Notice that no sound reflected off the ceiling treatment will reach the listener’s ears until it has bounced off the rear wall.
To improve the response of the floor and ceiling reflections, I designed my speakers using a MTM arrangement with the midrange units spaced according to D’Appolito’s technique which greatly narrows the vertical dispersion, thus reducing the output levels that reach the floor and the ceiling in the range from about 280Hz up to the crossover to the tweeter. To get a similar effect from the tweeter, I added a felt ring to the baffle to absorb off axis sound off the tweeter. The shape of the felt was carefully designed to create a tweeter off axis response which is nearly identical to the mid/woofers in both the vertical and horizontal planes.
Finally, to demonstrate visually what the effects of the treatments have done, I created another response chart that shows the response characteristics of the reflections that are arriving from all six primary reflection points. The responses of each reflected sound has been tailored to reflect the treatment used AND the controlled off axis characteristics of the speaker. I also removed the notches in the bass response since the bass speaker in my speaker setup is close to the floor and not likely to have the out of phase effects I discuss above and calculated based on my tweeter location being the source of all sound originating at the speaker.
Here is the example chart:
Notice that the ripple effect from the carpet comb filter issue has been removed. This is due to the controlled dispersion of the D’Appolito arrangement and the tweeter felt ring.
So far I have only discussed the primary, first order reflections. Maybe another day I will create a part three of this series discussing reverb and higher order reflections.
One of the things I completely failed to mention is that below about 150Hz, the boundary effect will reinforce the bass from the speaker just because it is near a reflective surface. When frequency’s 1/4 wavelength is more than the distance between the speaker and a reflective surface, the combined output of the reflection and the source is 6dB louder than the source alone. So, if a speaker is within 3 feet of a wall, the output below about 94Hz will be 6dB louder than if the speaker is not by the wall. Since the speaker is already sitting on the ground, there is a 6dB gain for the floor reflection as well. So, placing a speaker in a corner will create an 18dB boost in the SPL over the same speaker being in a true anechoic chamber.
This detail means that my charts above are only useful above about 150Hz since I did not include the boundary reinforcement in my simulations.
As a fun side note, the effect I just described also applies to listeners. If you are sitting in a corner, the bass SPL could be increased by 18dB.
FAQ:
What methods are you using to determine your reflections?
I use a mirror and a lamp. I place the lamp bulb right where my head would be aimed at the wall about where I think the reflection should be. Then I place a simple handheld makeup mirror flat against the wall and move it around until the reflected light hits the tweeter of the speaker. I then mark that spot on the wall. After that I move mirror until I have the reflection point for each driver in the speaker and mark the wall appropriately.
I’m using a variation of your design for the one inch OC panels for absorption do you also have plans for diffusion/diffusers?
If you check out my HT Construction Thread (LINK), you''ll notice I have more diffusors than I do absorbers. I plan on adding some more to the ceiling over the head of the listeners. While I was writing this thread and creating all the charts and diagrams, I was planning on creating a part three where are I go into late reflections (as opposed to early reflections), RT60 (room decay) and reverb. That thread would discuss the purpose and need for diffusors. The lack of activity on this thread is not encouraging, so I may never work on a part three.
What methods do you use to determine the placements of diffusers?
I spend lots and lots of time finding all the dual and triple reflection points on the walls and ceiling and place the diffusors at those points to convert what would be late echoes hitting my head into smooth reverb. It works well.
I was personally going to just use the mirror method for reflections; I don't have much faith in that system.
The mirror system is the finest available because you can quickly get the perfect location for each speaker reflection on every wall and ceiling.
SECONDARY REFLECTIONS
Secondary reflections are the reflections that could bounce off two or more walls and still arrive at the listener's ear pretty much sounding the same as the direct sound, just delayed and lower in SPL. Since all of this happens in a three dimensional environment, I can only diagram the most basic forms of secondary reflections in my two dimensional models.
Here is a diagram of the four most important simple secondary reflections using only the walls as reflectors:
Here is a diagram of the only simple secondary reflection utilizing the ceiling and rear wall:
I could go through the math to calculate the delay in mS and the reduction in SPL based on distance traveled, but I won't at this time.
The point is, these sounds will reach the listener's ears as distinct echoes, not as smooth decay or such.
I have two sample audio files demonstrating the difference between a clap in my HT with lots of echo and the same room after being partially acoustically treated presenting reverb instead of echo.
Part of the treatment process is to diffuse the secondary reflections (as well as all non-absorbed reflections) to prevent the sound from fully and completely reflecting directly back at the listener. If the sound is diffused, what would originally reflect right to the listener will be split up and reflect off all the room's surfaces and eventually hit the listener's ears at every possible delay interval imaginable. This is reverb.
Here are a couple of sample diagrams of how I diffused the basic secondary reflections in my HT:
Note that the sample audio files were recorded before I applied all of the diffusion to the walls, so the final sound is actually more diffuse that the so called "treated" clap file suggests.
Now that I have explained the nature of the reflections - their properties in SPL, phase shift, delay, and such - let's look at how we measure them in a way that can be translated in what we hear.
The standard method for measuring room reflection is the RT60 chart. This chart measures the echo decay of the room in time and SPL.
All of the data I have used above suggests a RT60 type chart such as this:
I limited the chart to the real reflections we have already diagramed and a 12dB window. In a proper chart, the SPL range would extend down about 30dB and the time axis would last for about 60mS.
Notice that the initial sound is followed by 5 reflections that are relatively loud within 10mS. Our brains will interpret those first 5 reflections as part of the original signal and give the impression that a tick on a ride cymbal in a recording is actually nearly 9 mS long. All of the adjectives I used above apply to the sound represented in this chart.
Now, let's look at a chart of the sound we could expect in the treated room:
Notice that aside from the floor reflection, the entire reflection pool has been eliminated and instead of echoes happening after 12mS, there is a smooth reverb arriving at the listener's ear from the reflections being diffused and reflected off other surfaces and diffused again.
The treated room will wound more natural and detailed to the listener and present less distortion to the mind.
THIS VERY THING IS WHY I CLAIM THAT ACOUSTICS PLAY AS BIG A ROLE IN SOUND QUALITY AS THE SPEAKERS.
The best speakers in the world placed in an untreated room will be perceived as being fat and blurred when compared to average speakers in a well-designed listening environment.
As a side note, in my own room, the floor reflection is actually much less loud and limited to lower frequencies than we perceive as detail due to my use of two midrange drivers in a D'Appolito arrangement and the felt rings around the tweeters. So, the reality of that floor echo is much different for me.
Here is a chart that more accurately represents how my RT60 histogram might look:
FAQ:
1) A question about carpet vs. hardwood floors. I always thought carpet was better too. But won''t hardwood be more of a reflective surface? How would you treat the primary reflections on the floor? Wouldn''t a rug have the same effect as carpet?
I choose not to treat the floor because it is impractical. Instead, to get the most natural sound, a fully reflective floor is better than a semi-reflective floor, especially with the comb filtering that carpet brings to the table. Different carpets and rugs have different effects, so results will vary. In some of the very high-end mastering/mising studios I have contracted on, we did treat the floor reflections the same way we treated the wall reflections with large acoustic panels - either absorbers or diffusors. This is possible, and it can look nice, but it makes that part of the floor useless for any other purpose.
2) With my speaker's waveguide and my mains using dual woofers, does that help reduce the comb effect the carpet has?
If both of those woofers are operating in the midrange, and assuming they spaced them to take advantage of the D'Appolito principle of controlled dispersion, then you make have less of a problem with the midrange. However, the wave guide for the tweeter is too shallow to truly eliminate vertical transmission of sound. The design of the wave guide on your speakers is merely a less effective way to cut down on edge diffraction. Sound from the tweeter will still reflect off the floor.
3) What about sound coming from the surround speakers. How should one treat a room for these? Even though most movies and surround music dosen't have much that comes from these speakers some SACD/DVDA''s do have A LOT of music and sound coming from them (Pink Floyd DSOTM, Dr. Chesky's 5.1 surround sound show, David Sanborn "Time Again", to name a few).
My article on acoustics out at http://www.soundenvironments.com thoroughly covers that subject. I discuss the principles of a "LEDE," or live end dead end, listening room and how it can be modified to work with a surround sound setup.
4) Since the panels I'm building are portable could a possible room arrangement be to have only 1 on the left and right walls at the primary reflection point (btw... the right wall has an walk way. It may not be in the primary reflection point but if it is what advantage do I have to treating it?) And use the other two panels for absorption in the front of the room? My speaker's baffle is 9.5" wide for the mains and rears. 7 7/8" for the center channel.
I would experiment. The width of the baffle suggests that most sound above 350Hz will not emanate behind the speaker, so unless you have significant edge diffraction, the reflections off the front wall will all be below about 350Hz. The only way to know is to stick a baffle there and see what happens. As for the side walls, you need to place a baffle on both sides of the room in a mirrored arrangement, even if there is a walkway. For two-channel listening it is essential that the sound from the left & right be identical. The second place I would experiment with the two extra panels is on the cross room reflection points for the opposite speaker. In other words, the left speaker’s reflection point off the right wall.
Here are the absorption coefficients for 2" thick 703:
Type in. (mm) 125 250 500 1000 2000 4000 NRC
703 Plain (51) 0.17 0.86 1.14 1.07 1.02 0.98 1.00
703, FRK (51) 0.63 0.56 0.95 0.79 0.60 0.35 0.75
How low does the effectiveness go for the 2" thick OC-703? It looks like 250 Hz as well with some absorption down to 125 Hz.
There is significant absorption down to about 225Hz with very little below that. If you can place the fiberglass up to 2" from the wall, the low frequency effectiveness can be extended down to about 180Hz. This is what I did with my panels and the results are pretty amazing.
The formula for determining the frequency at which a baffle will not direct sound forward anymore is:
C = Speed of sound = 1128ft/S
W = Width in Feet = 0.79166666, or 9.5 inches
C / W = F’
1128 / 0.7916666 = 1424.84Hz
That is the same frequency as the width of the baffle. Above that frequency all sound will radiate to the front of the baffle (except of diffracted and bend sound). At 1/4 wavelength all sound will bend around the baffle.
1424.84 / 4 = 356.21Hz
So, below 356Hz, all sound will disperse in all directions from the baffle. Between the root frequency (1.4 kHz) and the quarter wavelength (350Hz) the sound will start bending around the baffle in a curve shape pattern. So at 500Hz more sound will come around the baffle than at 1 kHz.
See how that works?
As for lower frequency absorption, the best tools are "bass" traps. Ethan Winer has plans for simple lower midrange traps that are more effective tha most. If you want a faster solution, using FRK compressed fiberglass with the foil on the outside, facing into the room, can be pretty effective. But then you get reflection at the higher frequencies, so ther are trade-offs.
I have mentioned the issue of our brains not being able to tell a direct sound from an echo if the echo arrives less than 10mS after the direct sound. In other places I have suggested that difference needs to be greater, like 20mS (or more), in order to assure that you can inherently hear the difference between a direct sound and an echo.
The reality, like nearly everything, is more variable.
When the echo comes from the side or rear in relation to the direct sound being in front of you, the typical line in the sand between our perception of direct and echo it about 10mS - and it varies from person to person.
However, if the echo is arriving from the same area as the direct sound, it takes a longer delay before our brains can tell any difference between it and the direct sound. In the situation that the echo is coming from the same area as the direct sound, the typical delay requirement to tell a difference is generally accepted as being 20mS.
So, the cool looking decay chart I posted above with the various arrival times of echoes from the various walls is not truly sufficient when tuning a room properly - especially small rooms. The direction of the reflections in relation to the direct sound needs to be known in order to know the impact they will have on the perceived accuracy of the direct sound.
1) What about diaphragmatic absorbers? I have a question, why fill the air gap with fiberglass? (I know there should be a .25 or .5'' gap between the plywood and fiberglass but why have fiberglass inside the panel? Simply designing a diaphragm panel of .25'' plywood (.74 lb/sq ft) and placing it on 2 by 4s would create an are gap of 3.75'' and a frequency of resonance of 102Hz. I realize the fiberglass helps absorb more of the sound wave at lower freqs but normally glass fiber loses its effectiveness when the wavelength of the sound is much larger than the thickness of the fiberglass (i.e. At 102Hz the wavelength is around 11 feet.) How does the fiberglass help if it in itself can't do a lot of absorbing at such low freqs.
- The fiberglass slows the speed of sound and thus lowers the tuning frequency of the trap. It also dampens the sound by absorbing a portion of the pressure wave. Finally, it widens the Q of the total panel. If you leave the fiberglass out, the panel isn't really absorbing much other that the specific frequency that makes the diaphragm vibrate.
2) Polycylindrical Absorbers. Are these practical for home use?
- Not in most cases.
3) Membrane absorbers. The master handbook of acoustics gives a simple example of a membrane being buliding insulation with kraft paper toward the sound. They also say that building insulation has not caught on as an inexpensive absorbing material due to cosmetic and protective covers required. Hmm.... I might have to think about using that as a bass absorber one day. Or is it even worth considering? I also have a feeling you aren't referring to this by Membrane absorbers.
- That is one solution. When I said membrane, I was refering to what the Handbook refers to as "Diagphramic" absorbers.
4) Helmholtz Resonators. I could use a lot of 10oz coca cola bottles to absorb 5.9 Sabins at 185 Hz each!! Unfortunately they have a bandwidth of .67 Hz so only 1 tangential mode (185.5 Hz) and 1 oblique mode (184.6Hz) would be affected by adding a coca cola bottle. I realize bottles and pots are not good Helmholtz resonators but I found the above amusing. However reading about these makes it seem hard to build properly (need a high Q factor (and this seems to have advantages, being higher Q''s have less losses, and disadvantages, being higher Q''s have reverb problems) and need precise placement to work.
- The two solutions below are variations of a simple Helmholtz resonator.
5) Perforated Panel Absorbers. These look very promising. Could I build a free standing model?
Yes you could. In fact there are hundreds of samples of these in the commercial industry and you see them all over the place. My ceiling reflectors are a variation on this design with an open back, which increases the tuning frequency.
6) Slat Absorbers hmm... the book doesn't mention much about these
- These are the most common version of Helmholtz resonators used in recording studios, especially drum room.
- You didn't mention hanging panels or huge masses of absorbing material in corners.
Which are generally the best routes to go with in a home to absorb low-mid freqs? Are any of these practical to make as a freestanding panel/resonator?
- I like my variation on that uses a monocylindrical front diffusor and an open back filled with fiberglass. In my less than perfect tests, my panels absorb in the 200Hz to the 600Hz range quite well, considering their size.
More on topic, is there any way to diffuse bass? (under 100hz?)
- Not in a typically sized room
Your diffusion panels are monocylindrical correct? These work very well to diffuse sound if it hits at 0 degrees. Don't they create a comb filter effect if sound hits them at a 45 degree angle? Does this cause any audible effect?
- Not really - at least not audibly. You are looking at this in terms of a monocylindrical unit on a very large wall. My walls are not large, nor is my room. The size and shape of my diffusors vary (to reduce both similarities in low frequency cutoff, vary the dispersion pattern of the high frequencies, and change the tuning of the cavity so the absorption frequencies vary). Now, in the truest sense, there are frequencies where there will be some comb-filtering, but these are limited and not consistent between units. My primary goal was to diffuse the midrange frequencies - from about 800Hz to 7 kHz.
The Quad-Diffusors are better than what I did, but I cannot afford to buy proper quad-diffusors and building my own would be too heavy to mount easily and removably.
The 2D quad-diffusors (Skylines, Art Diffusors) are better than the 1D versions in smaller rooms, but they have a narrower "Q'' and thus limited response.
My diffusors are more simple. They create a non-parallel, non-linear surface which will diffuse sound:
For frequencies above the wavelength of the diameter of the monocylindrical units I built, the sound will be diffused. So, if the diameter of the unit is 36" the lowest frequency which is diffuses will be 376Hz, however the perfect diffusion starts one octave above that at 752 Hz.
Now, these same units are also basically Helmholtz dampened resonators in that air pressure that goes around the front panel will be resonated in the cavity behind the front. If the width of the unit s 24", that means frequencies below about 280Hz will enter behind the unit and be absorbed by the mass of fiberglass inside the unit. Also, the shape will provide some basic absorption at the diameter (564Hz). Then there is the sound that can travel through the front material which will be absorbed by the mass of fiberglass.
Here are my assumed and poorly calculated properties for the larger diffusors I made for the side walls (above the absorption panels):
I was able to measure some of this data based on ground plane measurements and a test speaker, but as much is assumed performance via mathematical models as it is measured.
While I have described two easy methods for quickly finding the reflection acoustic points on the walls (or ceiling & floor) between the listener and the speakers, I have received numerous emails and PMs asking me to explain it again. It seems the more I explain these methods, the harder they sound to the people asking for my help.
So, today I made a few charts showing how to use a mirror to find the first reflections of sound between a speaker and the listener.
First, here is a diagram showing a room with a couch and two speakers. I added rays to the diagram to represent the acoustic energy which needs to be absorbed (or diffused) in order to reduce or eliminate the first order reflections which this entire series of articles attempted to expose.
This next drawing shows a simple way a single person with a tradition lamp can fairly easily find the points on the wall where the first order reflections are location so he can address them with absorption materials:
By placing a table lamp or adjustable height floor lamp where the speaker should be with the bulb at the very spot where the tweeter on the speaker should be located, one can simply move a hand mirror around on the wall to find the spots on the couch where the light is perfectly reflected. By marking the wall where the mirror perfectly reflects the light one will know the center of the area which requires acoustic treatment.
The lamp method can be inverted where the lamp is placed where the listener's head would be located and moving the mirror around until the light is perfectly reflected on the tweeter.
Here is another diagram using a mirror and two people:
This method is by far the fastest and doesn't call for any extra tools nor moving the speaker out of the way. One person sits in the various listening positions and another person moves the mirror around on the wall until the seated person can perfectly see the tweeter.
In both methods is it imperative that the mirror be held perfectly flat against the wall so that the glass is perfectly parallel to the wall. Otherwise using a mirror is pointless.
It is also important to treat the reflection from the speaker on the opposite side from each wall:
It is also important to consider the vertical aspects of treatment.
While the most critical reflections to treat are those from the tweeter, it is very advantageous to treat the woofer reflections as well.
Here is a diagram showing what might be required if the speakers are tower-sized with several woofers:
Here's some historical research to explain the "blurring" phenomena of early reflections:
Haas Effect
Also called the precedence effect - describes the human psychoacoustic phenomena of correctly identifying the direction of a sound source heard in both ears but arriving at different times. Due to the head's geometry (two ears spaced apart, separated by a barrier) the direct sound from any source first enters the ear closest to the source, then the ear farthest away. The Haas Effect tells us that humans localize a sound source based upon the first arriving sound, if the subsequent arrivals are within 25-35 milliseconds. If the later arrivals are longer than this, then two distinct sounds are heard. The Haas Effect is true even when the second arrival is louder than the first (even by as much as 10 dB). In essence we do not "hear" the delayed sound. This is the hearing example of human sensory inhibition that applies to all our senses. Sensory inhibition describes the phenomena where the response to a first stimulus causes the response to a second stimulus to be inhibited, i.e., sound first entering one ear cause us to "not hear" the delayed sound entering into the other ear (within the 35 milliseconds time window). Sound arriving at both ears simultaneously is heard as coming from straight ahead, or behind, or within the head. The Haas Effect describes how full stereophonic reproduction from only two loudspeakers is possible.
