|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
216.80.143.40
The baffle step, killer of DIY speakers by beginners, is pretty well known, but something very important is often ignored / underestimated / overlooked: destructive interference, often in the critical 100-160 Hz region, between the direct wave from the woofer and the reflected wave from the floor.
It is a difficult problem to solve:
- Lower the woofer, higher in frequency the interference will be;
- The medium must be close to the tweeter, and both at a reasonable height.
- Larger the distance between the woofer and the medium, lower the Fx should be.
- If crossed over low, the medium can also be 'victim' of floor reflection.
2.5 and 3.5 ways speakers using two identical woofer don't have too much trouble, 4 identical drivers is even better, but 'traditional' 3 ways speakers are suffering.
Some manufacturers tried to resolve this, like PSB with its discontinued Stratus Gold.
Do you know other examples of good solutions?
Follow Ups:
Don't really understand this?
I have noticed that two-way floorstanders that I have seem to have an ideal distance from the floor that they tolerate. In my case, about 2-3 inches is ideal. More and the bass disappears. Less and the floor seems too prominent of a factor in the sound.
I would really like to understand this issue and know why stand mounted monitors don't seem to be so sensitive to height. Is it because they have no real bass output??
Above:
Power response curve of Grand Utopia BE.
You are seeing (and experiencing) the Allison Effect in action.The speaker's direct sound and the reflection of that sound from nearby boundaries will alternately cancel each other or reinforce each other at specific frequencies determined by the distance between the drivers and those boundaries.
The frequency of power response nulls (cancellations) due to the Allison Effect can be determined using the following formula:
Frequency of Null = 1130 (velocity of sound)/Distance in feet x .3 (quarter wave coefficient).
Power response nulls become much more severe when the woofers are the same distance from multiple boundaries.
Below the frequency at which direct and reflected sound cancel each other (creating nulls), they reinforce each other, creating shelving bass boost (easilly heard as you move speakers closer and closer to boundaries - bass response shelves up as the bass drivers get closer to the boundaries and the speakers can sound boomy and distorted).
That is what you are seeing in the example that I attached above (power response curve of the Grand Utopia BE)... a severe cancellation around 100 Hz (you can work out the distance between the Grand Utopia BE's woofers to nearby boundaries in this example using the formula I provided) followed by shelving bass boost at frequencies below the cancellation (where the direct and reflected sound reinforce).
The null and the bass boost issues could be reduced by placing the speakers so that the woofers are a different distance from each room boundary and none of the boundary distances are multiples of the others. With such placement the power response curve of the speakers would flatten out.
Edits: 07/18/12
Above:
Power Response curve of a modern speaker design (Grand Utopia BE).
This is one of the rare instances where I disagree with Duke."Floor Bounce" otherwise known as the Allison Effect does not just occur on (one) axis but actually everywhere in the room. The direct sound from the speaker and the reflection of that sound from the nearest boundary will always be out of phase (and cancel each other) at a certain frequency determined by the distance between the drive unit and the nearest boundary (or boundaries as is usually the case).
The Allison Effect is a power response problem by definition and is all too evident aurally when listening to speakers who's designers have ignored this important aspect of wave mechanics. The Allison Effect can usually be heard as a suckout in the "power" passband or around 150 Hz to 300 Hz (given common speaker design woofer placement). The suckout is particularly destructive with orchestral music resulting in greatly diminished scale and a thin, unbalanced sound that emphasises high frequencies at the expense of lower frequencies. This also affects vocals especially the tenor range of male vocals and the mezzo range of female vocals (making them sound unnatural and muffled). Since the power response suckout is usually accompanied by a boost (wave addition) at frequencies below the cancellation, you also get "Barry White voice" where most male voices sound like Barry White.
Rock and pop music usually emphasize mid-bass frequencies (50 Hz to 75 Hz) and therefore power response cancellations are not as noticeable with these genres of music (unless the woofers are way up off the floor in which case the frequency of the power response notch moves down).
I would say that 95% of current speaker designs are afflicted with Allison Effect type power response problems, which is very obvious if you look at power response graphs (measurements of reverberant field energy) rather than on axis frequency response graphs. Like the above and below examples.
Edits: 07/14/12 07/14/12
Hi,You are greatly mistaken on a range of issues here.
First, what Stereophile publishes is the room averaged response, NOT power response.
Second, what is called "room averaged" for the purpose is only a narrow window, not anything like the whole room.
Third, the deep notch would be where it is (maybe 130 Hz) if the room in which the measurements where taken had a 2.6 m high ceiling and the measurement hight would be somewhere around 1m to maybe 1.3m...
In fact, given that this speaker has one bass driver very high up at the top of the speaker and another near the floor one would expect to see a shallow double notch, one a very low frequency from the driver high up in the air and another at fairly high frequency from the driver at the floor.
Moreover, disconnecting the bottom bass unit should drastically change the frequency and amplitude of the notch if it resulted from the Allison Effect but it in fact fails to do much at all.
So in fact it would seem that the Allison Effect fails to make it's appearance in these measurements, while the floor ceiling mode would appear to make a (predictable given the speakers geometry) appearance.
Ciao T
Sometimes I'd like to be the water
sometimes shallow, sometimes wild.
Born high in the mountains,
even the seas would be mine.
(Translated from the song "Aus der ferne" by City)
Edits: 07/14/12
The power response is NEVER, EVER eared by a listener!
A human being, despite quantum mechanics theory, is in practical, actual life at one point in space at a given time.
I don't even agree with the averaging of frequency response at different places. What you ear is what a measurement microphone records at one location. you can move, and then the frequency response will change, and can be measured by the same microphone at the same location. But you will never ear an 'average'.
As for the 'power response', it is theoretical (energy spread in all directions, that will NEVER reach your ears in the same amplitude versus frequency balance, 'thanks' to differential ansorption and phase changes).
Now, you could challenge this using quantum mechanics. THAT would be interesting, but I guess that we both need a higher-level physicist to help us doing that.
Hi,
> The power response is NEVER, EVER eared by a listener!
This goes against a fairly large body of work in resarching the subject, including but hardly limited to Floyd E'Toole's work so idolised by villastrangiato.
Now if you have made research that suggests the contrary I would be keenly interested to read more about it. Otherwise you may wish to peruse some of the links I suggested to villastrangiato for some remedial reading.
> A human being, despite quantum mechanics theory, is in practical,
> actual life at one point in space at a given time.
> I don't even agree with the averaging of frequency response at
> different places.
The problem is not where the human being is, but how the human hearing works (e.g. Haas et al) and the location where the human being listens to music on a HiFi.
The only place where the off axis response of the speaker has no bearing on the direct on-axis sound would be dozens of meters up in the air with both speakers and human suspended in free space.
Ignoring how uncomfortable and impractical such a "listening - non - room"
, it would deliver a pure free field.
In reality of course we tend to be quite surrounded by sound reflective surfaces, known to those skilled in the art as "Walls" and "Floor" and "Ceiling".
> What you ear is what a measurement microphone records at one location.
You seem confused. The Human hearing is remarkable for how it operates precisely in a way that remarkably nothing like a microphone.
> you can move, and then the frequency response will change, and can
> be measured by the same microphone at the same location. But you
> will never ear an 'average'.
You are missing a crucial point. For a certain time span (the so-called Haas Window) will integrate both first arrival and reflections into "one stimulus".
The first arrival dominates the perception of direction, the integrated sound power dominates the perception of loudness. Any sound arriving outside the Haas window n will be percieved as "Echo".
In terms of distance the Hass window is around 7 - 14m...
> As for the 'power response', it is theoretical (energy spread in all
> directions, that will NEVER reach your ears in the same amplitude
> versus frequency balance, 'thanks' to differential ansorption and
> phase changes).
The energy not directly radiated towards the listener will reach the listeners ears inside rooms, however it will reach it delayed.
And you are right, if there is much absorption (as opposed to diffusion) in the room this will alter the tonal balance of the reflected sound. This in one of the reasons why excessive absorption is not a good idea in listing rooms. Absorbtion can only be used if offers equal absorbtion across the rull range of frequencies.
> Now, you could challenge this using quantum mechanics.
Why? It is not necessary to do so.
Simple conventional acoustics and human hearing analysis suffice to completely reject your position and to confirm that indeed in acoustically small rooms (like recording studios, living rooms etc.) the perception of loudness (and hence frequency response) is dominated the speakers off axis response (which is often rendered as "power response".
Interesting this "power response" gig is also one of the reasons why both "baffle step" and "Floorbounce Notch" do not exist in real systems and real rooms.
The "floor bounce notch" as such has a frequency that is caused by delay, however the delay of the many LF reflections that will be present and reach the listener where they will be integrated will have many different "notch" frequencies. As the direct on-axis sound only contributes a small fraction to the total sound power there is no deep notch.
Similarly, as the "Baffle Step" only exists on axis, but not in the integrated sound (incidentally a small amount of correction, but less than 3dB may be desirable BTW, once all interactions are considered), so if you compensate the baffle step what you actually get at the listening position is a 6dB Boost, something that is trivial to confirm using Pink Noise and a simple spectrum analyser.
Ciao T
Sometimes I'd like to be the water
sometimes shallow, sometimes wild.
Born high in the mountains,
even the seas would be mine.
(Translated from the song "Aus der ferne" by City)
Thorsten babbles:
"Third, the deep notch would be where it is (maybe 130 Hz) if the room in which the measurements where taken had a 2.6 m high ceiling and the measurement hight would be somewhere around 1m to maybe 1.3m...
In fact, given that this speaker has one bass driver very high up at the top of the speaker and another near the floor one would expect to see a shallow double notch, one a very low frequency from the driver high up in the air and another at fairly high frequency from the driver at the floor.
Moreover, disconnecting the bottom bass unit should drastically change the frequency and amplitude of the notch if it resulted from the Allison Effect but it in fact fails to do much at all."
Cancellation or superposition in midbass frequencies has been measured (I've measured it myself many times, thanks) as a result of different path lengths contributing to what the microphone or ear is hearing. Layman is quite correct in that at typical listening heights (40 inches off the floor for example), a woofer mounted at a similar height off the floor will contribute essentially two sounds to what is heard - that which bounces off the floor AND ceiling and that which is direct. Depending on the distance to the listening point, the net difference of path length each portion of the wavefront travels determines whether the frequencies are cancelled or re enforced. Placing a driver near the ceiling or floor - in essence, coupling the driver to these surfaces - minimizes this problem.In such cases, the path length that looked more like the two hypotenuses of two 45 degree triangles begins to look more like the complex interaction of two or three room modes (ceiling to floor vs front wall to back wall vs. side wall to side wall). In that instance, given the wavelengths and inherent energy involved to support further reflection, the amount of cancellation/re enforcement is minimized. This is why experienced designers recommend and have had success in coupling bass drivers close to the floor. The closer such drivers are raised to typical ear height, the more opportunity there will be for cancellation and reenforcement by out of phase reflected wave fronts.
So effectively, there would not be two cancellation notches as Thorsten erroneously suggested - only potentially one. And that would depend on the distance to the listener. Longer distances would reduce the amount of phase difference. As an example, if one calculates the path length difference for a sound reflected from the floor or ceiling from a driver mounted 4 feet high to a person listening at 4 ft height 10 feet away - the path length difference amounts to almost 3 feet! If you double the distance to the listener, it drops in half to about 1.5 feet.
Hi,
You just cannot stop it, can you?
How about you have a look at the speaker, at the room it was measured in and how it was measured, before you again illustrate that you know little and assume much.
The Speaker incidentally has a 15" Woofer near the floor and an 11" at the top of the speaker at around 1.7m or so high... The room incidentally is only 5.5m in the longest distance, so around 3m listening distance is probably the max. that can be achieved...
No with 0.5m and 1.7m high drivers and 3m listening distance, what would be the cancellation frequencies from the path length differences?
And with a 8.5' high ceiling where would be the vertical mode and how would it be driven by these drivers? and what would be the result, if measured averaged over a reasonable width/hight window around the nominal listening position?
Ciao T
Sometimes I'd like to be the water
sometimes shallow, sometimes wild.
Born high in the mountains,
even the seas would be mine.
(Translated from the song "Aus der ferne" by City)
The room gain should and does IME fill-in the floor reflection cancellations to some extent. Compensating for it with an extra driver adds complexity and reduces clarity, BDS compensation throws away dynamics as well as eating 3-6db of power, and detail and nuance as well.I've lived with a pair of speakers with no low-pass (no BDS compensation) for a long time. Their actual diffraction behaviour is as smooth as possible, just like Olson predicted for a sphere. The enclosures are very quiet to boot.
So that's two causes of distortion at any level effectively eliminated.
I bought the speakers instead of QUAD 57s which I was set on, before I heard these.
There is no complete answer to anything in audio, nor is perfection possible, it's all interdependent even orthogonal.
Note that a post in response is preferred.
Warmest
Timothy Bailey
The Skyptical Mensurer and Audio Scrounger
And gladly would he learn and gladly teach - Chaucer. ;-)!
'Still not saluting.'
Edits: 07/13/12
First reflections in bass might not exist in real rooms because the wavelength is larger than the first reflection up until a point. But if a person doesn't believe that they can stack up three or four 7-10" woofer for bass to streach the source woofers so the bounce has a wider distance.
Room nodes are next to impossible to fix in a square\rectangle room. That ceiling node is a killer with deep dips around 70-80hz and nasty peaks around 100hz. The only way to fix that is with a room without parallel walls and ceiling. Better yet buy large wooded lot and and set up your hi-fi in a small clearing and sit in a nylon screened room (to keep the bugs off of you). The constant and natural diffusion of plant life and irregular ground is the best acoustics. Makes $200 worth of used vintage gear sound awesome!!
Right oh...
6" or 7" based MTM's on stands and a big shaggy rug on the floor in front of the speakers!! ;)
nt
Hi,
> The baffle step, killer of DIY speakers by beginners, is pretty
> well known,
Actually, here is a fun fact. The "baffle Step" only exists in anechoic on axis response, in the speakers power response there is no baffle step.
Now why is that? And what do you tink you hear when you lkisten to a given speaker in your living room, the power response or the on axis response?
Secondly, it is true that many small speakers sound subjectively "better" with several decibel bass lift in the power/in room/actually heard response. However, it must be understand why that is (Hint - read up on Robinson-Dadson and/or ISO 226:2003).
> but something very important is often ignored / underestimated /
> overlooked: destructive interference, often in the critical 100-160
> Hz region, between the direct wave from the woofer and the reflected
> wave from the floor.
Again, this supposed "interference" only exists on axis. Not in the diffuse field, which is what you actually hear, it is not present.
In other words, this "interference" is a simple result of misinterpreting the measured results and of being unaware of basic acoustics.
> It is a difficult problem to solve:
Hardly. The imple solution is to switch from single point, pseudo-anechoic gated measurements to multiple point room averaged measurements appropriate to the subject being investigated and voila, the problem has miraculously vanished.
This in fact is covered by acoustics 101...
> Do you know other examples of good solutions?
Use appropriate measurements and/or be aware of the limits and peculiarities of the method you use to measure?
BTW, a real problem in most domestic rooms is the vertical mode, as seated at around 1m Ear Height you are peripously close to the null of the mode of a 2.4m high ceiling, which happens at around 140Hz for common 2.4m high ceilings.
Thankfully my living/listening room is blessed with a over 4m high cathedral ceiling, so I simply do not have this problem and of course, if I did have this problem I would not mistake it for ïnterference from the floor reflection".
Ciao T
Sometimes I'd like to be the water
sometimes shallow, sometimes wild.
Born high in the mountains,
even the seas would be mine.
(Translated from the song "Aus der ferne" by City)
- The 'Power Response' is irrelevant. You are not everywhere in the listening room, you are at a specific position, hopefully more or less in front of your speakers. Reflected waves are attenuated by the reflection surfaces, and walls in the US are semi-transparent to low frequencies. Moreover, a large part of the reflected sound is out-of-phase. The 'Power Response' matters only in a all-concrete small lab, and the response at the listener's hears would still be different.
- The floor is typically a very strong reflector of low frequencies (and sometimes of mids and highs also - wood, stone, tiles). When out-of-phase, the reflection attenuates the direct sound.
As you, I don't have the ceiling's reflection problem, but often it is less critical than the floor reflection, except when the speakers are tall, the floor above made of concrete and low.
Darn things mucking up our sound. The walls to are a issue as is the ceiling heck your body's causing issues with absorption lets get ride of that too.
If the room is big enough to have a reverberant field then you hear the power response of the speaker.
Hi,
> The 'Power Response' is irrelevant.
Interesting. Really? So room treatment is a big scam to make people pay ton's of money for stuff they do not need?
> You are not everywhere in the listening room, you are at a
> specific position, hopefully more or less in front of your speakers.
Yes the sound you hear includes all the reflections that the Haas effect integrates with the first arrival...
> Reflected waves are attenuated by the reflection surfaces, and
> walls in the US are semi-transparent to low frequencies.
Most people do not live in the US AND even in the US you find many building that are actually very solid, though typical individual houses are indeed build in ways that raise eyebrows elsewhere...
The only house I ever lived in that had non-solid surfaces for the living room was a victorian semi-detatched house in London, the floor was wood (but only a few feet above solid earth (so quite reflective at LF) and the ceiling was wood too and hence had significant leakage... In all other cases all surfaces have been and are solid.
Arguably one needs to take the room acoustics of the taret room into account when designing speakers, so for example a speaker intended for an individual house build to US code would need a very different LF behaviour to one designed for a Condo in New York or a Flat Berlin or Tokyo.
> Moreover, a large part of the reflected sound is out-of-phase.
Really, would care to substantiate that?
Surely "in phase" and "out of phase" for a given wavelet depends on the time delay. As this varies greatly so will phase, hence the result is hard to predict. Moreover, we do not as such listen to sine-waves. and the human auditory system does not as such operate like a microphone.
There where reasons that in the days when only the Crown TEF (which I had access to) could make near anechoic measurements pink noise was used measurements in room (like when setting up speaker system in a studio).
> The 'Power Response' matters only in a all-concrete small lab, and
> the response at the listener's hears would still be different.
Yet you will find that in most cases that power response dominates the actual frequency response and perceived tonal balance at the listening position.
> The floor is typically a very strong reflector of low frequencies
> (and sometimes of mids and highs also - wood, stone, tiles). When
> out-of-phase, the reflection attenuates the direct sound.
And when in-phase the reflections amplify direct sound. In a solid structure pretty much all LF is reflected back and some will be in- and some out of phase, so with sinewaves we will get a lot of comb filtering.
However as the human hearing integrates sound in the Haas Window regardless of time delay (phase) and simply registers this as increased loudness.
Ciao T
Sometimes I'd like to be the water
sometimes shallow, sometimes wild.
Born high in the mountains,
even the seas would be mine.
(Translated from the song "Aus der ferne" by City)
The on-axis response only predicts the spectral balance of the first-arrival sound that reaches your ears. The power response predicts the spectal balance of the reverberant energy also, and I agree with Thorsten's giving greater weight to the power response.
The floor-bounce notch is a naturally-occuring comb-filter phenomenon, and as such is subjectively rather benign. It occurs every time two people have a conversation in a non-anechoic space. I'm not saying its effect is inaudible, just that it's not a big deal. The ears do not process sound the same way measuring equipment does; the microphone sums the first-arrival sound and the floor-bounce and gets a notch, but the ears can tell the difference and so you don't really hear that deep floor-bounce notch that looks so bad on a graph as long as it is naturally induced by the room's acoustics (and neither electronically generated nor on the recording, as in either of these cases it is readily perceived as coloration of the first-arrival sound).
One of my designs happens to minimize the floor-bounce notch through driver geometry, but that's more a side-effect than a major design goal.
Duke
Me being a dealer makes you leery?? It gets worse... I'm a manufacturer too.
Greater weight should not blanketly be given to on axis response or power response without first assessing the other half of the equation - room response. As noted by Toole, the ultimate response heard in rooms that are more highly absorptive will be more affected by direct, on axis loudspeaker response. Those rooms that are more highly reflective will have the opposite situation in which power response plays a vital role. Blanket statements are seldom accurate in this arena. When it comes to acoustics, the details determine what the outcome will be and generalizations are practically useless - all generalizations except for this actual statement about generalizations, of course.
: )
Well yes, if the room is absorptive enough that there is little energy delivered to the ears via reflections, then obviously the direct sound matters the most. But unless you have a highly damped room or a nearfield setup, that is unlikely to be the case. In most home listening rooms, once you are beyond about 5 feet back from the speakers (this distance of course varies with speaker characteristics and room acoustics), more energy arrives at the ears via reflections than via the direct sound.Toole notes that the following factors all correlate well with subjective preference:
1.The first-arrival sound’s frequency response at the listening position;
2.The spatially-averaged response across a window plus or minus 30 degrees horizontal and plus or minus 10 degrees vertical;
3.The spatially-averaged response of the early reflections from the four walls, ceiling, and floor;
4.The sound power, or the sum total of the speaker’s acoustic energy radiation; and
5.The directivity index, which looks at the difference between the first-arrival sound’s frequency response and the sound power (summed omnidirectional response).
Note that four of the five have more to do with what's happening off-axis than with what's happening on-axis.
I have no problem with taking the room and setup and listener positions into account when choosing speakers, but without such specific information it makes sense to look at what is likely to matter in most rooms. The direct sound matters and so does the reverberant sound, and in general the role of the latter is greatly under-appreciated in part because it is generally under-reported.
Duke
Me being a dealer makes you leery?? It gets worse... I'm a manufacturer too.
Edits: 07/14/12
Duke said:
" The direct sound matters and so does the reverberant sound, and in general the role of the latter is greatly under-appreciated in part because it is generally under-reported."
On that, I'm sure everyone here will agree - even Thorsten.
nt
nt
nt
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: