This week, instead of talking about what is inside the loudspeakers, let’s talk about what I listen for when sound is coming out of them. Specifically, let’s talk about one spatial aspect of the mix – where instruments and voices are located in two-dimensional space. (This will be a short posting this week, because it includes homework…)
Step 1: Go out and buy a copy of Jennifer Warnes’s album called “Famous Blue Raincoat: The Songs of Leonard Cohen” and play track 2 – “Bird on a Wire”.
Step 2: Close your eyes and really concentrate on where the various voices and instruments are located in space relative to your loudspeakers. If you hear what I hear, you’ll hear something like what I’ve tried to represent on the map shown in the figure below.
A map of the locations of many of the instruments in Jennifer Warnes’s recording of “Bird on a Wire”.
I’ve used some colour coding, just to help keep things straight:
Voices are in Red
Drums are in blue
Metallic instruments (including cymbals) are in green
Bass is gray
Synth and Saxophone are in purple
Note that Jennifer sings her own backup vocals, so the “voice”, and the two “bk” (for backup – not Burger King) positions are all her. It also sounds like she’s singing in the “choir” on the left – but it’s hard for me to hear exactly where she is.
Whenever I’m listening to a pair of loudspeakers (or a car audio system, or the behaviour of an upmix algorithm) to determine the spatial properties, I use this map (which I normally keep in my head – not on paper…) to determine how things are behaving. The two big questions I’m trying to answer when considering a map like this revolve around the loudspeakers’ ability to deliver the (1) accuracy and (2) the precision I’m looking for. (Although many marketing claims will use these words interchangeably, they do not mean the same thing.)
The question of accuracy is one of whether the instruments are located in the correct places, both in terms of left and right, but also in terms of distance. For example, the tune starts with a hit on the centre tom-tom, followed immediately by the bigger tom-tom on the left of the mix. If I have to point at that second, deeper-pitched tom-tom – which direction am I pointing in? Is it far enough left-of-centre, but not hard over in the left loudspeaker? (This will be determined by how well the loudspeakers’ signals are matched at the listening position, as well as the location of the listening position.) Secondly, how far away does it sound, relative to other sound sources in the mix? (This will be influenced primarily by the mix itself.) Finally, how far away does it sound from the listening position in the room? (This will be influenced not only by the mix, but by the directivity of the loudspeakers and the strength of sidewall reflections in the listening room. I talked about that in another blog posting once-upon-a-time.)
The question of precision can be thought of as a question of the size of the image. Is it a pin-point in space (both left/right and in distance)? Or is a cloud – a fuzzy location with indistinct edges? Typically, this characteristic is determined by the mix (for example, whether the panning was done using amplitude or delay differences between the two audio channels), but also by the loudspeaker matching across the frequency range and their directivity. For example, one of the experiments that we did here at B&O some years ago showed that a difference as small as 3 degrees in the phase response matching of a pair of loudspeakers could cause a centrally-located phantom image to lose precision and start to become fuzzy.
Some things I’ve left out of this map:
The locations of the individual voices in the “choir”
Extra cowbells at around 2:20
L/R panned cabasa (or shaker?) at about 2:59
Reveberation
Some additional notes:
The triangles on the right side happen around 2:12 in the tune. The ones on the left come in much earlier in the track.
The “synth-y fx around 2:20” might be a guitar with a weird modulation on it. I don’t want to get into an argument about exactly what instrument this is.
I’ve only identified the location of the bass in the choir. There are other singers, of course…
You might note that I used the term “two-dimensional space” in the beginning of this posting. In my head, the two dimensions are (1) angle to the source and (2) distance to the source. I don’t think in X-Y cartesian terms, but Polar terms.
An important thing to mention before I wrap up is that this aspect of a loudspeaker’s performance (accuracy and precision of phantom imaging) is only one quality of many. Of course, if you’re not sitting in the sweet spot, none of this can be heard, so it doesn’t matter. Also, if your loudspeakers are not positioned “correctly” (±30 degrees of centre and equidistant from the listening position) then none of this can be heard, so it doesn’t matter. And so on and so on. The point I’m trying to make here is that phantom image representation is only one of the many things to listen for, not only in a recording but also when evaluating loudspeakers.
Take a balloon and blow it up. It will look something like the drawing in the centre of Figure 1.
Fig 1. A balloon in its natural state (the middle), the same balloon being squished (on the left), and the same balloon being stretched (on the right).
If you put your hands on the top and bottom of the ballon and compress them, you’ll make the balloon shorter, but you’ll also make it wider (as is shown on the left side of Figure 1). This is because the air inside the balloon is under a higher pressure when you squeeze your hands together, and that pressure pushes harder on the parts of the balloon that aren’t being held in by your hands.
If, instead, you grab the top and bottom of the balloon and stretch them further apart, you’ll make the balloon narrower (as is shown on the right side of Figure 1). This is because you’ve created more space inside the balloon, thus lowering the internal pressure pushing out on its walls. The lower the pressure, the less the air pushes outwards, so the balloon collapses.
What we’ve described here is basically the same as what happens to a sealed loudspeaker cabinet (if you’re not careful when you do its mechanical design). However, instead of your hands squeezing a balloon (a sealed “cabinet” of air), we use the signal from a power amplifier to pull the woofer into the cabinet. This increases the pressure inside the cabinet, since it’s sealed and there’s nowhere for the air to go. As a result, the trapped air tries to push outwards on the sides of the cabinet. If the walls of the cabinet are thin, then the cabinet itself will act like the balloon and expand.
Similarly, if you put a positive signal on the power amplifier, you’ll move the woofer out of the cabinet, reducing the air pressure inside it and “sucking” the sidewalls of the loudspeaker in (if they are so thin that they can move).
Generally speaking, you want a loudspeaker to behave as a piston in a baffle (like this). This means (in this case) that you want the woofer to move in and out of a loudspeaker cabinet and you don’t want the cabinet itself to move at all. (We’ll talk later about why this is a bad thing.)
There are a number of different ways to do this. One way is to make your loudspeaker cabinet out of something very, very, very stiff (and probably, as a result, very, very, very heavy). For example, we make our prototypes out of 22 mm or 25 mm thick MDF (and sometimes we use two sheets of it, glued back to back, to make sure that things are stiff enough). Most of our loudspeakers like the BeoLab 9, for this week’s example, have enclosures that are made of plastic, so we use some different methods to achieve the stiffness and rigidity we need to ensure that our enclosures are not singing along with the drivers.
Fig 2. A “naked” BeoLab 9 with the Acoustic Lens assembly removed.
The first step in ensuring that the walls of the loudspeaker cabinet are stiff enough is to make them thick enough. How thick is “enough” depends on the loudspeaker itself. A subwoofer with a small enclosure volume will have to withstand more internal air pressure than a midrange in a larger enclosure. Of course, we can’t simply make the walls of the cabinet out of overly thick plastic, since that would not only be an unnecessary waste of materials, it would increase the weight of the loudspeaker (a consideration for shipping) and the cooling time of the plastic when it’s in production (a consideration for the production line timing and costs).
A second way to make the sidewalls stiffer is to use ribs – usually on the inside of the cabinet. These are moulded as part of the sidewall itself – they aren’t just glued on to the inside surface of the cabinet. If you take a look at Figure 3, you can see the ribs on the inside of the sidewalls of the BeoLab 9. These run diagonally and vertically in the case of this loudspeaker – but they’re not just randomly placed inside the loudspeaker. They have been strategically placed using simulations in the early development stages, and measurements of the early prototypes. These measurements are done using very small accelerometers glued to the sides of the loudspeakers and monitoring their outputs while playing signals through the loudspeaker drivers. (In case you’re wondering, the depth of the ribs is about 22 – 25 mm, depending on where you measure.)
Fig 3. A closeup of the ribs on the inside of the woofer enclosure of the BeoLab 9.
A third tactic is to use a plastic that is stiffened by adding things to it. The usual method is to use fibre-reinforced plastic. The fibres in the plastic help give it a structural strength that you can’t get from just plastic alone.
A fourth possibility is to create a laminate material where you build up the enclosure using layers of different materials (or layers of the same material with different structural composition). This increases stiffness in the same way that a sheet of plywood is stiffer than a sheet of wood.
So, in the case of the BeoLab 9 (as with many of our other loudspeakers), three of these tactics were used. The plastic is thick enough, it has strengthening ribs on the inside, and it is a laminate (if you slice it open, you’ll see that it is a layer of foamed plastic, sandwiched between “skin” layers of solid plastic).
However, when they (I wasn’t part of the BeoLab 9 development team – I was still working in the Automotive Department at the time) got to the last stages of the development, it became obvious that there was a problem. There was an audible resonance caused by the woofer. Some digging around resulted in the finding out that the sides of the cabinet were moving too much. In essence, the problem was almost exactly as I described with the balloon in Figure 1.
Fig 4. A not-to-scale cross section of a BeoLab 9 prototype showing a simplified explanation of its “ballooning” effect before the problem was fixed.
As I tried to show in Figure 4, when the woofer moved inwards, it pushed the sidewalls of the loudspeaker outwards (shown with the blue lines). When the woofer moved outwards, it sucked the sidewalls inwards (the red lines). (I’m over-simplifying here, but not enough to start a fight.)
However, as I said, this was discovered rather late in the development process. The problem had to be fixed, but the question was how to do it without starting from scratch and creating new moulding tools for making the plastic enclosure. The solution was to use the sidewalls to reinforce each other. Since the movement of the opposite sides was in opposite directions (i.e. the left and right sides of the loudspeaker either wanted to move apart or together at any given moment) they could be braced by connecting them with a bridge. Take a look again at Figure 3. You’ll see a metal rod that goes straight across the middle of the woofer enclosure. You can see it just above the back of the woofer in Figure 5 as well.
Fig 5. A view into the woofer cabinet of a late-stage prototype BeoLab 9.
That piece became known in the acoustics department as the “dog bone” because its final version had the basic shape of a cartoon dog bone. The end result is that the rod is included in the BeoLab 9 construction to prevent the sidewalls of the woofer cabinet from moving when you’re playing loudly.
Here’s another photo showing the internal PCB with the amplifier and filters – not because it’s relevant to this discussion, but just because it might be interesting…
Fig 6. The electronics (analogue filters and amplifiers) of a BeoLab 9. As you can see here, this circuit board normally lives inside the woofer enclosure. The power supply board is not shown in this photo.
How is a loudspeaker cabinet like a nude ballet? (Or: so what?)
So, we’ve seen how we get rid of parts of the loudspeaker moving when they shouldn’t. The question is “why is it a big deal?” Well, it’s a little like Sir Robert Helpmann’s comment about the difficulties in choreographing a nude ballet – the problem is that some parts of the body keep moving when the other parts have stopped.
In theory, a loudspeaker should behave the same at all frequencies (that statement can mean a lot of different things – and I am happy to argue that it is both true and false, depending. So don’t try to start a fight with me on that one, and please don’t mis-quote me and say that I’m contradicting Siegfried Linkwitz or anyone else by taking that statement out of context on a hi-fi forum somewhere else…) As I mentioned in a previous posting, we like to pretend that a loudspeaker is just a moving piston in an infinite baffle, since that behaves pretty well. Of course, no one actually believes that this model is true – but it’s a comforting ideal.
Take a look at the shape of the blue and red versions of the loudspeaker cross-section in Figure 4, above. You might notice that, when the woofer goes outwards, the cabinet goes inwards. When the woofer goes inwards, the cabinet goes outwards. (this is an oversimplification of the truth, but let’s go with it for now). This means that, from the point of view of the air pressure radiating from the entire loudspeaker, the woofer goes positive and the cabinet goes negative at the same time. So, the sound pressure radiating off the cabinet cancels the sound pressure radiating off the woofer as can be seen in an example of two opposite-polarity sources causing destructive interference as in the animation below. (Note the intersection of the red and green curves where you will hear no sound at all…)
Not only that, but the cabinet as a sound source has a very different directivity than the woofer by itself. So, the result is… complicated. The truth is even more complicated, since the cabinet will not behave as nicely as this – it will resonate better at some frequencies than others, making some notes “bloom” – they’ll appear to be louder (or quieter, depending on where you are and how the room interacts with the loudspeaker) and they might even appear to come from a different direction.
So, the moral of the story here is that you want all parts of the loudspeaker to not be moving when the parts that are supposed to be moving are doing so.
And, before you go and listen to your loudspeakers and look for “blooming” and blaming it on panel resonances or vibrating cabinets, don’t forget that your room modes are more likely a primary source of mis-behaviour when it comes to some bass notes sounding different from the others. However, if you have problems with your room acoustics, we can’t fix that with ribs and dogbones – unless you also bring in a large dog. (I know, I know – the linked paper is about humans – but it does make a mention of animals in the abstract…)
Before we start talking about curves and corners, let’s have a quick review on the concept of interference. At its most fundamental level, sound is just a relatively small, relatively fast change in barometric pressure. If the instantaneous pressure is higher than average (which happens to be the same as the pressure inside your head), then your eardrum is pushed into your head. If the instantaneous pressure is lower than average, then your eardrum is pulled out of your head. When your eardrum moves in and out, you hear sound.
One way to create a high pressure is to take a loudspeaker driver and push it outwards. In order to create a low pressure, we pull it inwards, as is shown in the animation below. The red thing is a piston which is basically the way we like to pretend a loudspeaker driver (like a tweeter) behaves. The grey thing is a very, very wide loudspeaker cabinet. The red semicircles show the high pressure zones that expand outwards from the front of the loudspeaker. The green semicircles show the low pressure zones.
If you have two sound sources, their pressure differences (relative to the average pressure) add. So, if you have two high pressures arriving at your eardrum, it will be pushed farther into your head than if only one of them arrived at your ear. Similarly, if you get two low pressures arriving at your eardrum, it will be pulled further out of your head than if only one of them was present. HOWEVER, if you have a high pressure and a low pressure arriving at the same time in the same place (for example, at your eardrum) then they cancel each other and, if they have the same magnitude, your eardrum won’t move at all and you won’t hear anything. (This is how noise-cancelling headphones work. The sound from the headphones is in theory identical to the sound coming to your ears from outside the headphones, except that it’s opposite in polarity, so the sum of the two sounds is nothing.)
Fig 1. The bottom plot is the result of adding the signals shown in the top and middle plots, moment by moment in time. For example, the red stars show the values at one moment in time. One plus one (the top two plots) equals two (the bottom plot).
Fig 2. The bottom plot is the result of adding the signals shown in the top and middle plots, moment by moment in time. For example, the red stars show the values at one moment in time. One plus negative one (the top two plots) equals nothing (the bottom plot).
Keep all that in mind as you read on…
Diffraction
It should not come as a surprise that a sound wave will bounce off a hard surface like a flat concrete wall. The question is “why?” The answer to this question can be complicated – but the simple version is that the molecules in a concrete wall are harder to move than the molecules in air – so we have a change in acoustic impedance. This is essentially a measure of how easily the molecules in the substance are moved by a sound wave… sort of… (Let’s leave it at that, since we really don’t need this article to be a thorough discussion of acoustic impedance.)
The interesting thing is that an acoustic wave will be reflected off any change in impedance. So, you don’t have to be going from a low impedance to a high impedance (as in the case of a sound wave trying to move from air into concrete). It will also reflect on a boundary where you change from a high to a low impedance (for example, a sound wave in concrete trying to get out into air – in this case, the sound wave will bounce off the surface of the concrete and move back into it rather than “leak” out into the surrounding air.).
Imagine yourself standing in a long tunnel. You clap your hands, and the sound wave travels down the length of the tunnel until it reaches the end – what happens then? Well, the answer is “two things”. Some of the sound leaks out of the tunnel. However, since the air inside the tunnel has a higher acoustic impedance than the air outside the tunnel (because the sound is freer to go where it wants on the outside), the end of the tunnel is a boundary where there is a change in acoustic impedance. And, as we saw in the last paragraph, this means that we will get a reflection. So, even though the end of the tunnel is open, it reflects your hand clap back into the tunnel. So, some sound leaks out and some reflects. (One way to really experience this is to notice your ears pop when you enter a long tunnel on a fast-moving train. When you first enter the tunnel, your ears pop because of the sudden change in pressure. Some time later, you might notice that your ears pop again. This is because the high-pressure wave front that the train made when it entered the tunnel travelled to the opposite end of the tunnel, bounced back and hit you again.)
If you didn’t know that the second sound was a reflection off the end of the tunnel (for example, you didn’t hear the first hand clap because you were wearing earplugs) you might think that it was a direct sound from someone down at the far end of the tunnel. So, if you’re not the person doing the clapping, but you’re in the tunnel with that person, you get two sounds – the direct sound and the reflection.
There is another, less obvious case where you have a change in acoustic impedance. This is when you have a sound wave travelling along next to the surface of something, and the surface ends. For example, if, in the animation at the top of this page, the surface of the loudspeaker cabinet was not as wide, there would be a corner where the face of the loudspeaker meets its side. At that corner, the acoustic wave front “sees” a change in acoustic impedance. Consequently, there is something like a reflection that starts at the corner. in essence, the corner of the loudspeaker is a boundary that radiates like a second sound source (just like the end of the tunnel in the example above).
So, if we modify the animation at the top of the page to include a narrower cabinet, the result would be something like the animation below.
As you can see there, the corner of the loudspeaker becomes a second source that radiates its own sound waves after the original, direct sound hits it.
This effect is called acoustic diffraction and it has some significant implications on the sound of a loudspeaker. This is directly because of the interference (see above…) between the direct sound from the loudspeaker driver and the secondary sound source caused by the corner.
Remember we saw above in Fig.1 that, when you have two high pressure zones that meet each other, you get more pressure than either one of them alone. Now take a look at the animation above and look for the places where the black curves from the two “sources” intersect. This is where you’ll get an increase in pressure, and therefore more energy than just the direct sound by itself. As you can see in Fig. 3, below, this means that you have an angle off-axis to the front of the loudspeaker where the signal is louder than it is directly on-axis. Of course, this also means that there will be some angles where you hear less (because the secondary wavefront from the corner cancels the direct sound) – these are where the red and green curves (the high and low pressure zones) intersect.
Fig 3a. The straight lines show the places where the high pressure zones overlap each other (the red lines cross the red lines) and the low pressure zones overlap each other (the green lines cross the green lines), creating constructive interference and therefore higher sound pressure level (in other words, it’s louder).
As you can see in Fig 3a, there are different angles where the high pressures add to give you an even higher pressure (notice that the low pressures also add to give you an even lower pressure. The result is that, along those lines, you get constructive interference and therefore the sound is louder than it is elsewhere. I’ve only shown three such angles in this diagram – there are more. You might note as well that the origin of the high pressure lobes is not the centre of the loudspeaker driver (the piston shown in red). It’s somewhere between the primary and secondary sources (in this case, the loudspeaker driver and the corner).
Fig 3b. The straight lines show the places where the high pressure zones overlap with the low pressure zones (the red lines cross the green lines), creating destructive interference and therefore lower sound pressure level (in other words, it’s quieter).
Fig 3b shows two angles where the high and the low pressures overlap, causing destructive interference and therefore cancellation. Therefore, the sound is quieter along those lines than it is elsewhere.
The real world
What does all of this mean in the real world?
Well, as you’ve probably already guessed, the first conclusion is that building a loudspeaker that has sharp corners is probably a bad idea. For example, if you wanted to build a loudspeaker, and you just put a tweeter on the front and made sharp right angles where the sides meet the front, you will have a problem with diffraction off those corners. As you can see in Figure 3, you will get a boost in the signal at some angles off-axis to the front of the loudspeaker, and some cancellation at other angles. The amount by which the signal will be boosted, the angles where you’ll have the effects, and the frequencies where you’ll have the problems are all dependent on the specific dimensions of the device. For example, the further away the loudspeaker’s corner from the driver, the lower the frequency that will be affected.
Let’s take a real-world example. The very first version prototype of the BeoLab 5 was really just a “normal” three-way loudspeaker that was used to test the ABC algorithm, So, there was a woofer in a cabinet with a microphone for the ABC development, but on top was just a small cabinet with a midrange driver and a tweeter, as you can see in Figures 4 and 5.
Fig 4. BeoLab 5 and its early prototype. This version was a late prototype of the lens geometry and the ABC demonstration / test device.
Fig 5. The original “conventional” tweeter and midrange used for comparison to the lens.
Fig 6. The original “conventional” tweeter enclosure used for comparison to the lens.
Figure 6 shows the “conventional” tweeter cabinet version of one of the BeoLab 5 prototypes which was placed on top of the white woofer cabinet shown in Figure 4 when the Acoustic Lens assembly was removed. As you can see, this is an example of how-not-to-make-a-loudspeaker (if you’re worried about diffraction). We have a tweeter in a flat surface and (some sharp-ish corners at the sides of the loudspeaker face). The result of this is that we have exactly the same problem shown in Figure 3a and 3b, above. We can see this in the measurement of the horizontal directivity of the loudspeaker, shown in Figure 7, below.
Fig 7. A directivity plot of the tweeter in the conventional cabinet shown in Figure 6. The X-axis is the frequency in Hz (ranging from 1.8 kHz to 20 kHz). The Y-axis is the horizontal angle of radiation where 0° is directly on-axis for the tweeter. The contour lines are in 0.5 dB increments and have been normalised to the on-axis response. The red ovals are peaks in the response caused by diffraction off the cabinet edges
It may be a little difficult to read this plot, so I’ll explain a little. The entire plot has been normalised to the on-axis magnitude response of the tweeter. In other words, the measurement doesn’t show the response of the tweeter – it shows how the response changes as you move around the loudspeaker in the horizontal plane. The X-axis is the frequency of the signal in Hz, ranging from 1.8 kHz to 20 kHz. The Y-axis is the horizontal angle of radiation of the loudspeaker where 0° is directly on-axis, in front the tweeter. The lines in the plot can be thought of as a kind of topographical map with a difference of 0.5 dB per contour. So, if you think of a straight “ridge” in the plot along the 0° line in the middle, the plot generally falls off (in other words, the signal is quieter) as you move around to the side and back of the loudspeaker. You can see that, at the high frequencies, the lines are closer together which means that you lose more level at high frequencies than at low frequencies as you come around to the side of the loudspeaker. This is traditionally called loudspeaker “beaming”. The interesting thing to look at are the four red oval areas. The larger ones are centred around 3.2 kHz and ±40°. The smaller ones are up at about 7.5 kHz and about ±15°. Because they’re in red, this means that they are louder than the on-axis response, so they are peaks in the topographical map. These peaks are the direct result of diffraction off the edge of the loudspeaker cabinet. I count 4 red contour lines at the lower frequency peak, which means that we have a beam that is about 2 dB (remember, 0.5 dB per line * 4 lines) louder around 3 kHz at 40° off-axis to the loudspeaker.
This cabinet was built compare the directivity of a normal box-shaped loudspeaker to one with an Acoustic Lens. A close-up of the lens used for this comparison is shown below in Figure 8.
Fig 8. A first-generation Acoustic Lens on an early BeoLab 5 prototype.
You’ll note in Figure 8 that the Acoustic Lens is slightly different from the final version (hence the “first-generation” qualifier) . This version also suffered from diffraction artefacts caused by the sharp edges where the face of the lens structure meets its side. This was corrected in the second generation version shown below in Figure 9.
Fig 9. A second-generation Acoustic Lens on an early BeoLab 5 prototype.
Notice that the second-generation lens has curved transitions from face to side to reduce the diffraction problem. This curvature was eventually extended to wrap around the entire structure as can be seen in the photo of the final BeoLab 5 tweeter lens in Figure 10, below.
Fig 10. An Acoustic Lens on an BeoLab 5 tweeter. Notice that the entire structure is curved from the face through the transition to the sides and to the back.
You may notice that the difference in these two designs was that the original one had sharp corners on the sides. The diffraction effects of these corners were easily visible in the first directivity measurements of the Lens, so the second prototype with the curved transition from front to side was made to eliminate this problem. The directivity measurement of the prototype shown in Figure 9 is seen below in Figure 11.
Fig 11. A directivity plot of the tweeter in a second-generation Acoustic Lens prototype shown in Figure 6. The X-axis is the frequency in Hz (ranging from 1.8 kHz to 20 kHz). The Y-axis is the horizontal angle of radiation where 0° is directly on-axis for the tweeter. The contour lines are in 0.5 dB increments and have been normalised to the on-axis response.
You’ll see in Figure 11 that there are two significant differences between the directivity of a tweeter in the prototype Acoustic Lens and a conventional cabinet (shown in Figure 7). The first difference is that the beaming effect (seen as a convergence of the contour lines at the high frequencies in Figure 7) does not happen with the lens. The contour lines are much more parallel resulting in a behaviour known as “constant directivity”. This is a way of saying that the loudspeaker has a directivity that is the roughly the same throughout its entire frequency range (rather than beaming in the high end).
The second difference is that the peaks in the 3 kHz and 8 kHz areas, seen in Figure 7, are gone. This is because there are no corners at the edge of the loudspeaker cabinet to cause diffraction. You may note a peak in the magnitude responses off-axis above 15 kHz. We actually don’t know what causes this, however, since it is so high in frequency and only +1 to +1.5 dB, and since this is still only a prototype, it wasn’t really considered to be a significant issue.
Wrapping up
So, I’ve killed two birds with one stone in this article (or “two flies with one smack” as they say in Denmark). On the one hand, we’ve seen that, if you’re worried about the directivity and/or the off-axis response of your loudspeaker (I know, the latter is a sub-set of the former…) sticking a tweeter (or a midrange, or a woofer, depending on dimensions and frequency ranges) on the front of a rectangular box is probably a really bad idea. (On the other hand, it’s a pretty easy, and therefore cheap, way to build a speaker, which is why such a design is so popular I guess…) And, on the other hand, we’ve seen one of the characteristics of Acoustic Lenses – being a more constant directivity than a tweeter-on-a-box. The fact that the tweeter mounted in an Acoustic Lens had less diffraction is not because of the Lens geometry in particular, but because of the shaping of its surroundings as part of the development process of BeoLab 5.
There are more stories like this one. For example, if you look carefully at the “plates” of the BeoLab 5 and the prototype in Figure 4 (the part the tweeter and midrange drivers are mounted in), you might notice that the prototype plates are flat, whereas the BeoLab 5 plates curve downwards. This is not because someone thought the curve would look pretty. This was because the circular edge of the prototype plates also caused diffraction, resulting in a higher-level lobe in the vertical plane. Sloping the plates downwards puts their sharp edges in the “shadow” of the plates themselves, reducing the diffraction effects. So, you can see that diffraction and its effects on directivity is one of the other issues that we worry about when we’re building a loudspeaker.
Rather than talk about technologies inside B&O equipment, this week I’ll try to go through a couple of strategies on how to properly calibrate the main channels in a surround system – and how to do it improperly, but make it sound better for your friends. I’ll use the example of a BeoPlay V1, a BeoVision 11 or a BeoSystem 4 as the heart of the system – but the basic concepts are the same for any other surround processor.
Location, location, location
The first step in setting up any surround sound system is the correct placement of your loudspeakers. There are two standard configuration recommendations. The first is from the International Telecommunications Union, in a document called Recommendation ITU-R BS.775-2 – Multichannel stereophonic sound system with and without accompanying picture (available as a PDF file from the ITU here). The second is called Recommendations for Surround Sound Production from the Producers and Engineers Wing of the Recording Academy of the National Academy of Recording Arts and Sciences (or NARAS – better known as the people that bring you the Grammys). Tha recommendation can be downloaded as a PDF file from here).
The short versions of these two recommendations are as follows:
ITU-775
The ITU standard configuration is the one people who do research into multichannel audio use for their experiments. It’s also the one we use at Bang & Olufsen when we’re testing our loudspeakers in the Acoustics Department or tuning the parameters in the TrueImage upmixing algorithm. The nice thing about this configuration is that it matches a surround sound system for someone who sits on a sofa placed against a wall, and has their surround loudspeakers adjacent to the same wall.
In a perfect loudspeaker configuration, all of your loudspeakers are the same distance from the listening position. They have all been calibrated to have the same loudness at the listening position. Also, they are all large, full-range loudspeakers (and therefore, you do not need a subwoofer).
The Centre Front loudspeaker should be in the centre, at the front (we’ll call that 0°). The Left Front and Right Front loudspeakers should be at ±30° relative to that angle. The Left and Right Surround loudspeakers should be located symmetrically at an angle of between ±100° and ±120°.
Fig 1. ITU 775 recommendation for 5-channel loudspeaker configuration
The ITU-775 document doesn’t specifically state the standard configuration for a 7-channel system, but it does provide a recommendation for a 5-channel system that uses 7 loudspeakers (in cases where you have a larger system and you use two loudspeakers per surround channel). However, the recommendation is still a pretty good recommendation for a 7-channel setup. (This also makes sense, since, if you have 7 loudspeakers, you may occasionally like to use them as a 5-channel system without having to place extra loudspeakers in your room.) If you dig around, you’ll see that this also fits the typical setups used in re-recording studios for doing 7-channel mixes and mastering for Blu-ray releases of films. A good example of this is Tron Legacy, which was produced using a system very much like the one shown below – with matching loudspeakers at 0°, ±30°, ±90° and ±150°. (this also makes sense from a radially symmetry perspective, since, ignoring the centre channel, you have equal loudspeaker spacings of 60°.
Fig 2. ITU 775 recommendation for 7-loudspeaker configuration
NARAS
The NARAS recommendation is a little different, although the people that wrote it were aware of the ITU recommendation (which came first…), so they made sure that their version didn’t contradict the existing standard. Their version uses the same layout for the front three loudspeakers, but suggests that the surround loudspeakers be a little further back – within the ±110° to ±150° angle, with an “optimal range” of ±135° to ±150°.
Like the ITU standard, the NARAS document also recommends that all loudspeakers be the same type of full-range loudspeakers, all the same distance from the listening position, all level-adjusted to be the same at the listening position.
Fig 3. NARAS recommendation for 5-channel loudspeaker configuration
The Real World
Both the ITU and the NARAS standards are really designed by and for people who work and live in recording studios or run perceptual experiments involving multichannel audio. This means that they have one chair and no friends – at least when they watch movies and listen to music… However, if you have a sofa and friends, then you will start having some questions – or at least some doubts.
For example, if you are “normal” (whatever that might mean) but a little careful about your surround sound setup, you probably have something that looks like the drawing below.
Fig 4. A reasonably good multichannel setup in the real world with a “correct” calibration scheme.
What happens if we were to calibrate this system “perfectly” using the centre of the sofa as our reference “sweet spot” as shown in Figure 4? We’d apply a delay to the Centre Front loudspeaker to make the time of arrival of its signals match the Left Front and Right Front loudspeakers (usually done by setting the Speaker Distance). We’d also apply a delay to the surround loudspeakers to do the same. We’d also probably drop the levels of the centre and surround loudspeakers to match the Left Front and Right Front signals (because they’re closer, and therefore louder).
However, let’s think about what happens if you sit on the left side of that sofa? Now, the Left Surround loudspeaker is very close to your left ear – and that has some serious implications on your experience. Firstly, since sound pressure doubles with every halving of distance, (assuming that this diagram is to scale) then sitting on the left side of the sofa means that you’ll get roughly a 6 dB boost (possibly more, if you’re leaning…) in the signal from that one loudspeaker. In addition, since that loudspeaker is so close and arriving at your listening position early, your brain will be able to figure out that the loudspeaker is close because you’re pretty good at localising sources when they’re near your head. The same problem, albeit on a much smaller scale, happens with the centre loudspeaker. If its time-alignment delay is calibrated using the centre position, then, if you’re sitting on the left side of the sofa, then the Left Front loudspeaker’s signal will arrive before the Centre Front. The end result of this is that, if you’re sitting on the side of the sofa, you’ll have too much from one of the surround loudspeakers and the intelligibility of the dialogue will be reduced a little.
So, how should we calibrate the system to make things a little better for your friends? Take a look at Figure 5, below.
Fig 5. A reasonably good multichannel setup in the real world with an alternative calibration scheme.
What I’m trying to show with this diagram is that both the distance and the level for each loudspeaker should be measured to the closest person in your listening area. So, in this case, the Left Front and Left Surround loudspeakers are calibrated to the left position on the sofa. However, the Centre Front loudspeaker is calibrated at the centre of the sofa. The result of this is that the centre speaker will be delayed – but less than it would have been if you had calibrated it as in Figure 4, because the Left Front loudspeaker is closer to the person on the left side of the sofa than to the person in the centre of the sofa. Also, the Surround loudspeakers will be delayed much more than they would have been using the scheme in Figure 4. However, they’ll still be symmetrical (so the person in the “sweet spot” won’t feel like the surround channels are lopsided, and the friends on the sides of the sofa won’t notice that they’re sitting on top of a loudspeaker… Also, this will result in the centre channel being a bit louder and the surround channels being a little lower in level – both of which are technically incorrect for the person in the sweet spot, but at least it’s a mistake in the right direction – so you’re improving intelligibility of the dialogue
If you do calibrate the system this way, you’ll technically be incorrectly calibrated at the sweet spot, but your friends on the sides of the sofa will be much happier – and you won’t notice too much. Of course, if you have a BeoPlay V1, a BeoVision 11 or a BeoSystem 4, you can make this configuration just one of your nine available Speaker Groups – you can always use another one for a “perfect” calibration for the sweet spot when you’re home alone.
If, after aligning your system using this method, you still find that the dialogue is a little hard to understand, and the surrounds are a little hot (this is often the case when your sofa and the surround loudspeakers are all situated against the same wall, you should not be afraid to do the following:
make the centre channel one or two milliseconds early. You can do this by telling your surround processor that it’s about 30 to 60 cm farther away than it really is.
raise the level of the centre channel 1 or 2 dB
drop the level of the surrounds as much as necessary – in my experience, it’s not unusual to have to drop them by as much as 6 dB if you’re against the same wall with them. (Note that, if you have a BeoPlay V1, a BeoVision 11 or a BeoSystem 4, you can do this using the “Fader” adjustment in the Sound menus. This will merely control the relative levels of the Front and Surround / Back loudspeakers – so it’s a one-fader solution to doing it manually for each loudspeaker output.)
If your listening area is larger, the technique is the same – you calibrate any given loudspeaker in the system to the closest listening position, and then tweak to taste.
I guess that the big message here is “just because your system is configured ‘correctly’ doesn’t mean that it can’t sound better”. Don’t be more afraid to tweak the adjustments on your calibration than you would be to add cream and sugar to your coffee, or salt and pepper to your meal in a restaurant. As Duke Ellington once said: “If it sounds good, it is good.”
The following question recently arrived in my inbox via our customer service department.
“I am an admirer of B&O Hifi products as of over 20 years, but a great mystery for me is how you achieve great sound reproduction in the typically minimalist Scandinavian interior design environmment with polished floors, bare walls and bare glass windows. Effectively such environments are acoustical disasters !?!”
I thought that this was a great question – so it’s the topic of this week’s article.
Of course, you are correct. A room comprised of large flat reflective surfaces with little acoustical absorption has a very different acoustical behaviour from a recording or mastering studio where the final decisions about various aspects of a recording are made. And, consequently, this must have an effect on a listener’s perception of a recording played through a pair (assuming stereo reproduction) of loudspeakers in that room. The initial question to be asked is “what, exactly, are the expected effects of the room’s acoustical behaviour in such a case?” The second is “if the room has too much of an effect, how can I improve the situation (i.e. by adding absorption or changing the physical configuration of the system in the room)?” The third, and possibly final question is “how can we, as a loudspeaker manufacturer compensate (or at least account) for these effects?”
The effect a room’s acoustical behaviour has on a loudspeaker’s sound can, at a simple level, be considered under three general headings:
early reflections
room modes
reverberation
Early Reflections
Early reflections, from sidewalls and the floor and ceiling, have an influence on both the timbre (tone colour) and the spatial characteristics of a stereo reproduction system. Let’s only think about the timbral effects for this article.
Fig 1. The sound arriving at a listener from a loudspeaker in a room with only one wall. Note that the sound arrives from two directions – the first is directly from the loudspeaker. The second is a “first reflection” off the wall.
Let’s start by assuming that you have a loudspeaker that has a magnitude response that is perfectly flat – at least from 20 Hz to 20 kHz. We will also assume that it has that response regardless of which direction you measure it in – in other words, it’s a perfectly omnidirectional loudspeaker. The question is, “what effect does the wall reflection have on the measured response of the loudspeaker?”
Very generally speaking, the answer is that you will get a higher level at some frequencies (because the direct sound and the reflection add constructively and reinforce each other) and you will get a lower level at other frequencies (because the direct sound and the reflection work against each other and “cancel each other out”). What is potentially interesting is that the frequencies that add and the frequencies that cancel alternate as you go up the frequency range. So the total result looks like a comb (as in a comb that you use to comb your hair, if, unlike me, you have hair to comb). For example, take a look at Figure 2.
Fig 2. Distance to loudspeaker = 2 m. Distance to wall = 1 m. Wall is perfectly reflective and the loudspeaker is perfectly omnidirectional.
So, you can see in Figure 2 that, at the very low end, the reflection boosts the level of the loudspeaker by a little less than 6 dB (that’s two times the level!) at the listening position. However, as you go up in frequency, the total level drops to about 15 dB less before it starts rising again. As you go up in frequency, the level goes up and down. This alternation actually happens at a regular frequency spacing (i.e. a notch at every 200 Hz) but it doesn’t look regular because the X-axis of my plot is logarithmic (which better represents how we hear differences in frequency).
What happens if we move the wall further away? Well, two things will happen. The first is that the reflection will be quieter, so the peaks and notches won’t be as pronounced. The second is that the spacing of the peaks and notches in frequency will get closer together. For example, take a look at Figure 3.
Fig 3. Distance to loudspeaker = 2 m. Distance to wall = 3 m. Wall is perfectly reflective and the loudspeaker is perfectly omnidirectional.
Similarly, if we move the wall closer, we do the opposite, as in Figure 4.
Fig 4. Distance to loudspeaker = 2 m. Distance to wall = 0.25 m. Wall is perfectly reflective and the loudspeaker is perfectly omnidirectional.
So, if you have a room with only one wall which is perfectly reflective, and you have a perfectly omnidirectional loudspeaker, then you can see that your best option is to either put the loudspeaker (and yourself) very far or very close to the wall. That way the artefacts caused by the reflection are either too quiet to do any damage, or have an effect that starts at too high a frequency for you to care. Then again, most room have more than one wall, the walls are not perfectly reflective, and the loudspeaker is not perfectly omnidirectional.
So, what happens in the case where the loudspeaker is more directional or you have some fuzzy stuff on your walls? Well, either of these cases will have basically the same effect in most cases since loudspeakers are typically more directional at high frequencies – so you get less high end directed towards the wall. Alternatively, fuzzy stuff tends to soak up high frequencies. So, in either of these two cases, you’ll get less high end in the reflection. Let’s simulate this by putting a low pass filter on the reflection, as shown in Figure 5, 6 and 7 which have identical distances as the simulations in Figures 2, 3, and 4 – for comparison.
Fig 5. Distance to loudspeaker = 2 m. Distance to wall = 1 m. Wall is absorptive and/or the loudspeaker is directional at high frequencies .
Fig 6. Distance to loudspeaker = 2 m. Distance to wall = 3 m. Wall is absorptive and/or the loudspeakers is directional at high frequencies.
Fig 7. Distance to loudspeaker = 2 m. Distance to wall = 0.25 m. Wall is absorptive and/or the loudspeaker is directional at high frequencies.
What you can see in all three of the previous plots is that, as the high frequency content of the reflection disappears, there is less and less effect on the total. The bottom plot is basically a proof of the age-old rule of thumb that says that, if you put a loudspeaker next to a wall, you’ll get more bass than if it’s farther from the wall. Since there is not much high frequency energy radiated from the rear of most loudspeakers, Figure 7 is a pretty good general representation of what happens when a loudspeaker is placed close to a wall. Of course, the exact behaviour of the directivity of the loudspeaker will be different – but the general shape of the total curve will be pretty similar to what you see there.
So, the end conclusion of all of this is that, in order to reduce undesirable artefacts caused by a wall reflection, you can do any combination of the following:
move the loudspeaker very close to the wall
move the loudspeaker farther front the wall
sit very close to the wall
sit farther away from the wall
put absorption on the wall
However, there is one interesting effect that sits on top of all of this – that is the fact that what you’ll see in a measurement with a microphone is not necessarily representative of what you’ll hear. This is because a microphone does not have two ears. Also, the direction the reflection comes from will change how you perceive it. A sidewall reflection sounds different from a floor reflection. This is because you have two ears – one on each side of your head. Your brain uses the sidewall reflections (or, more precisely, how they relate to the direct sound) to determine, in part, how far away a sound source is. Also, since, in the case of sidewall reflections, your two ears get two different delay times on the reflection (usually), you get two different comb-filter patterns, where the peaks in one ear can be used to fill in the notches in the other ear and vice versa. When the reflection comes from the floor or ceiling, your two ears get the same artefacts (since your two ears are the same distance to the floor, probably). Consequently, it’s easily noticeable (and it’s been proven using science!) that a floor or ceiling reflection has a bigger timbral effect on a loudspeaker than a lateral (or sideways) reflection.
Room modes
Room modes are a completely different beast – although they exist because of reflections. If you pluck a guitar string, you make a deflection in the string that moves outwards until it hits the ends of the string. It then bounces back down the string, bounces again, etc. etc. See the diagrams and animations on this page – they might help. As the wave bounces back and forth, it settles in to a total result where it looks like the string is just bouncing up and down like a skipping rope. The longer the string, the lower the note, because it takes longer for the wave to bounce back and forth on the string. You can also lower the note by lowering the tension of the string, since this will slow down the speed of the wave moving back and forth on it. The last way to lower the note is to make the string heavier (i.e. by making it thicker) – since a heavier string is harder to move, the wave moves slower on it.
The air in a pipe behaves exactly the same way. If you “pluck” the air in the middle of a pipe (say, by clapping our hands, or coughing, or making any noise at all) then the sound wave travels along the pipe until it hits the end. Whether the end of the pipe is capped or not, the wave will bounce back and travel back through the pipe in the opposite direction from whence it came. (Whether the pipe is closed (capped) or open only determines the characteristic of the reflection – there will be a reflection either way.) It might help to look at the animations linked on this page to get an idea of how the air molecules behave in a pipe. As the wave bounces back and forth off he two ends of the pipe, it also settles down (just like the guitar string) into something called a “standing wave”. This is the pipe’s equivalent of the skipping rope behaviour in the guitar string. The result is that the pipe will resonate or ring at a note. The longer the pipe, the lower the note because the speed of the sound wave moving in air in the pipe stays the same, but the longer the pipe, the longer it takes for the wave to bounce back and forth. This is basically how all woodwind instruments work.
What’s interesting is that, when it comes to resonating, a room is basically a pipe. If you “pluck” the air in the room (say, by putting sound out of a loudspeaker) the sound wave will move down the room, bounce off the wall, go back through the room, bounce of the opposite wall, etc. etc. etc. (other things are happening, but we’ll ignore those). This effect is most obvious on a graph by putting some sound in a room and stopping suddenly. Instead of actually stopping, you can see the room “ringing” at a frequency that gradually decays as time goes by. However, it’s important to remember that this ringing is always happening – even while the sound is playing. So, for example, a kick drum “thump” comes out of the speaker which “plucks” the room mode and it rings, while the music continues on. You can see this in Figure 8, below.
Fig 8. The concept of the effect of a room mode. See the text below for an explanation.
Figure 8 shows the concept of the effect of a room mode. The sound coming out of the loudspeaker is shown on the top plot, in black. The response in the room is shown in the middle plot in blue. You can see there that the room keeps “ringing” at a frequency after the sound from the loudspeaker stops. The red plot on the bottom is the difference between the two plots – in other words, the “sound” of the room mode in isolation (note that it’s at a different scale than the top two plots to make things easier to see).
There are two audible effects of this. The first is that, if your music contains the frequency that the room wants to resonate at, then that note will sound louder. When you hear people talk of “uneven bass” or a “one-note-bass” effect, one of the first suspects to blame is a room mode.
The second is that, since the mode is ringing along with the music, the overall effect will be muddiness. This is particularly true when one bass note causes the room mode to start ringing, and it keeps ringing when the next bass note is playing. For example, if your room rings on a C#, and the bass plays a C# followed by a D – then the room will be ringing at C#, conflicting with the D and resulting in mud. This is also true if the kick drum triggers the room mode, so you have a kick drum “plucking” the room ringing on a C# all through the track. If the tune is in the key of F, then this will not be pretty.
If you would like to calculate a prediction of where you’ll have a problem with a room mode, you can just do the following math:
metric version: room mode frequency in Hz = 172 / (room length in metres)
imperial version: room mode frequency in Hz = 558 / (room length in feet)
Your worst modes will be the frequencies calculated using either of the equations above, and multiples of them (i.e. 2 times the result, 3 times the result, and so on).
So, for example, if your room is 5 m wide, your worst-case modes will be at 172 / 5 = 34.4 Hz, as well as 68.8 Hz, 103.2 Hz and so on. Remember that these are just predictions – but they’ll come pretty close. You should also remember that this assumes that you have completely immovable walls and no absorption – if this is not true, then the mode might not be a problem at all. (If you would like to do a more thorough modal analysis of your listening room, check out this page as a good start.)
Sadly, there is not much you can do about room modes. There are ways to manage them, including, but not exclusive to the following strategies:
make sure that the three dimensions of your listening room are not related to each other with simple ratios
place your loudspeaker in a node – a location in a room where it does not couple to a problematic mode (however, note that one mode’s node is another mode’s antinode)
sit in a node – a location in a room where you do not couple to a problematic mode (see warning above…)
Reverberation is what you hear when you clap your hands in a big cathedral. It’s the collection of a lot of reflections bouncing from everywhere as you go through time. When you first clap your hands, you get a couple of reflections that come in separated enough in time that they get their own label – “early reflections”. After that, there are so many reflections coming from so many directions, and so densely packed together in time, that we can’t separate them, so we just call them “reverberation” or “reverb” (although you’ll often hear people call it “echo” which is the wrong word to use for this.
Reverb is what you get when you have a lot of reflective surfaces in your room – but since it’s so irregular in time and space, it just makes a wash of sound rather than a weird comb-filter effect like we saw with a single reflection. So, although it makes things “cloudy” – it’s more like having a fog on your glasses instead of a scratch. Think of it like the soft focus effect that was applied to all attractive alien women on the original Star Trek – you lose the details, but it’s not necessarily a bad thing.
So what are you gonna do about it?
Fine, this is a short-form version of what a room’s acoustics does to the sound of a loudspeaker, but how do we, as a manufacturer of loudspeakers, ensure that our products can withstand the abuse that your listening room will apply to the sound? Well, there are a number of strategies that we use to do what we can…
1. ABC. The BeoLab 5 has a proprietary system built-in called Adaptive Bass Control or ABC. Pressing a button at the top of the loudspeaker starts a measurement procedure that is performed using a built-in microphone that measures the loudspeaker’s behaviour in two locations. Actually, what it’s doing is looking at the difference in the loudspeaker’s response in those two positions of the microphone to determine the radiation resistance that the loudspeaker “sees” as a result of reflective surfaces in the room. The ABC algorithm then creates a filter that is used to “undo” the effects of some of the low-frequency effects of the room’s acoustics. For example, if the radiation resistance indicates that the loudspeaker is close to a wall (which, as we saw above, will boost the bass) then the filter will reduce the bass symmetrically. That way, the loss in the filter and the gain due to the wall will cancel each other.
2. Position switches. ABC in the BeoLab 5 is a very customised filter that, in part, will adjust the loudspeaker’s response for placement near a wall or in a corner. Almost all of the other BeoLab loudspeakers (and other sound systems such as the BeoPlay A8 and A9, for example), include a manual-adjusted “position switch”. This allows you to use one of three filters that we have customised in the development of the loudspeaker to account for its behaviour according to whether you have placed it away from a reflective surface (“Free”), near one surface (“Wall”) or in a Corner. This is not just a filter that adjusts the bass level. The three filters for “free”, “wall”, and “corner” have been calculated using three dimensional measurements of the acoustical behaviour of the loudspeaker. So, the filters for the BeoLab 3 are completely different from those for the A9, for example, because they have very different directivity characteristics.
3. Sound design in multiple rooms. As I talked about in a previous posting, when we do the sound design of all of our loudspeakers, we tune each of them in at least 4 or 5 rooms with very different acoustical behaviours ranging from a very “dead” living room with lots of absorptive and diffusive surfaces to a larger and very “live” space with a minimalistic decorating, and large flat surfaces (just like the description in the original question). Once we have a single sound design that is based on the common elements those rooms, we test the loudspeakers in more rooms to ensure that they’ll behave well under all conditions.
Wrap-up
Of course, I haven’t covered everything there is to know about room acoustics here. And, of course, you can’t expect a loudspeaker to sound exactly the same in every room. If that were true, there would be no such thing as a “good”concert hall. A room’s acoustical behaviour affects the sound of all sound sources in the room. On the other hand, humans also have an amazing ability to adapt – in other words you “get used to” the characteristics of your listening room. Back when I was working as a part-time recording engineer in Montreal, I did a lot of recordings in churches. Typically, we (the producer and I) would set up a control room with loudspeakers in a back room, and the musicians would sit out in the church. When we arrived to set up the gear, the first thing was to set up the monitor loudspeakers and a CD player, and we would play CD’s that we knew well while we set up everything else. That way, we would “learn” the characteristics of the control room (since we already knew what was on the discs and the characteristics of the monitor loudspeakers). So, if all of our CD’s sounded like that had too much bass, then we should do a recording with too much bass – it was the fault of the control room.
However, there is no debate that, due to lots of issues (the first two that come to mind are frequency range and directivity) two different loudspeakers will behave differently from each other in two different rooms. In other words, if you listen to loudspeaker “A” and loudspeaker “B” in a showroom of a shop, you might prefer loudspeaker “A” – but if you took them home, you might prefer loudspeaker “B”. This would not be surprising, since what you hear is not only the loudspeaker but the loudspeaker “filtered” by the listening room. This is exactly why, when you are buying a loudspeaker, you should audition it in your home in order to ensure that you will be happy with your purchase. And THIS is why you can arrange a home demonstration of Bang & Olufsen loudspeakers through your dealer.
As I mentioned in an early posting about two months ago, if I didn’t have time to write enough, I’d cop out and post some photos of development models of some of our loudspeakers. It’s been a busy week. So, here are some photos of the BeoLab 18 and BeoLab 14.
Early BeoLab 18 prototype. This photo was taken on the floor of the Cube.
Early BeoLab 18 prototype – acoustic lens closeup. The lens itself is an SLA prototype and is attached to the top of the milled plastic cabinet using modelling putty.
Early BeoLab 18 prototype showing a closeup of the port. More modelling putty was added to the bottom of the port opening to smooth the curve as part of an investigation about reduction of turbulence noise.
The first BeoLab 14 prototype – subwoofer and satellite. Note that the DSP and amplifiers would be external for this prototype. The white modelling putty at the top of the pipe used to port the cabinet was put there to reduce turbulence noise. It is not a prototype of the final geometry of the port flare – it was just a quick-and-dirty solution for the initial demos.
“… the power output is phenomenal. Moreover, the detail is incredible and the tonal balance is spot on. The vocals in Antony & The Johnson’ Twilight are intense, the throb of Jeff Beck’s guitar in So Real resonates sublimely, whilst classical works yield profound levels of clarity. The mi-range is highly detailed, the treble is smooth and accurate, and the bass is rich and velvety. You can push the volume without risk: even at high levels the speakers have plenty in reserve.”
I realised this week that I’ve been throwing around words like ” voice coil”, “suspension”, “surround” and “spider” when I talk about loudspeakers, but many people don’t know what these things are – or how a loudspeaker driver works in general… So, this week, I thought I might back up a couple of steps and talk about some basics. There’s nothing here about Bang & Olufsen loudspeakers specifically – it’s just an introduction to how loudspeaker drivers (like woofers, for example) work.
Back in 1820, a Danish guy named Hans Christian Ørsted was in the middle of giving a lecture when he noticed that, when he switched a circuit on and off, a compass sitting nearby on the desk moved a little. Since he had been poking around with experiments in electricity and magnetism for years, it didn’t take him long to put two and two together and come up with the idea that, when you run electrical current through a wire, you get a magnetic field around it. (Interestingly, not only did Ørsted figure this out (unless you believe that Romagnosi did it), but he also wrote papers on aesthetics – and he was the first person to isolate the element aluminium – and he founded Den Polytekniske Læreanstalt which, today we call the Danish Technical University. So, he had a big influence on B&O in many respects.)
Nowadays, we know that, by putting current through a wire, you produce a magnetic field that has magnetic lines of force that encircle the wire. The direction of the lines of force are dependent on the direction of current, and the extent by which the magnetic lines of force extend away from the wire is dependent on the amount of current.
Depending on whether or not you believe Benjamin Franklin you can use your right or left hand to determine the direction of the lines of force. Figure 1, below, shows a right hand (which indicates that we believe Benjamin Franklin and we say that current runs from the positive terminal of a battery to the negative terminal, which is, in fact, incorrect.). The thumb points in the direction of the current and your other fingers wrap around the wire in the same direction as the magnetic lines of force (which go from North to South on a magnet).
Fig 1. The right hand rule shows the direction of the magnetic lines of force around a wire as a result of putting current through it.
If you take that same wire, keep running current through it, and coil it up like a spring, you can make a slightly more useful magnet that actually has a North pole and a South pole. Again, you can use your right hand to figure out which end is which – you wrap your fingers around the spring in the same direction of the current going through the wire, and your thumb will be pointing towards the North pole of the magnet, as is shown in Figure 2.
Fig 2. If you have a coil of wire and you put current in it, you get a magnetic field. Note that, if you run the current the other way (by reversing the battery), the magnetic poles will reverse, so North will be on the left of this diagram.
Now, if you take two magnets and you put them end-to-end with the North of one facing the North of the other (or South to South), they’ll push each other apart. If you put them North-to-South, they’ll pull each other together.
So, let’s do something weird. We’ll make a coil of wire, and we’ll put it in a strange-looking permanent magnet that is a bit like a horseshoe magnet that has been wrapped around itself to make a circular plug in the middle which is one pole (say, North) and a ring around it which is the other pole (say, South) as is shown in Figure 3.
Fig 3. A coil of wire about to be put in the gap of a strange-looking permanent magnet.
Now, if I put current through the wire, I’ll make a magnetic field around it that will either push against or pull towards the magnetic field of the permanent magnet around it. In other words, if it’s free to move, it will.
Now, since a loudspeaker, generally, is a thing that is used to convert electrical energy into acoustical energy; and, since in order to create acoustical energy (i.e. make noises) we need to move air molecules; we can use this strange device we’ve made in Figure 3 to our advantage. However, let’s be a little more methodical about this…
So, let’s build a dynamic moving coil loudspeaker driver, bit by bit. We’ll start by talking about its name. The “dynamic” part means that the basic principle that does the work is electromagnetism (as opposed to electrostatics or some esoteric methods like using plasma, tesla coils, or cats). The “moving coil” part is because, uh, the part of the device that moves in the magnetic field of the permanent magnet is a coil of wire.
What we want to do is to put the wires of the coil inside a magnetic field that is as strong as we can make it (within reason, of course). The easiest way to do this is to make the “gap” the coil sits in as small as possible (and, of course, to use as strong a magnet as we can fit or lift or afford to buy). So, let’s make a small gap for the coil to sit in.
We start by making a “bottom plate” and connect a “pole piece” – this results in a shape that looks like a disc with a cylinder. It’s made out of soft iron because soft iron is a really good magnetic conductor. (In other words, if you stick a magnet on a piece of soft iron, the soft iron basically becomes an extension of the magnet without losing very much magnetic force.) That bottom plate and pole piece assembly is shown in Figure 4, below. I’ve made it red just to keep things clear later. It’s usually not red in real life.
Fig 4. The bottom plate and the pole piece, both typically made of soft iron.
As you can see already, the “plug” in the middle of the magnet in Figure 3 is already visible as part of the pole piece in Figure 4. However, in order to make the strength of the magnetic field greater (in other words, in order to concentrate the magnetic lines of force) we want to make the gap (where the coil is going to sit) narrower. This can be done by making the cylinder on the pole piece a little bigger in diameter – but only where the coil of wire will be. That’s done by putting a ring around it, as is shown by the blue part in Figure 5, below.
Fig 5. A ring has been added around the pole piece to reduce the gap width. (Note that the gap doesn’t exist yet – we’ll need to put in a couple of more pieces first.)
Now we add the magnet as you can see in Figure 6.. This looks like a ring that sits on the disc part of the pole piece. The top of the ring is one pole (say, South) and the bottom is the other pole (say, North) of the magnet. However, this means that the North pole of the magnet is extended to the top of the cylinder on the pole piece because (as I said earlier) the soft iron is a good magnetic conductor.
Fig 6. The blue ring is the permanent magnet, typically made of ferrite or neodymium.
You can see in Figure 6 that the gap between the top of the pole piece and the magnet is pretty big, so let’s make it smaller by putting a “top plate” on the top of the magnet. This is another disc of soft iron, where the hole is just a wee bit bigger than the diameter of the ring around the top of the pole piece as shown in Figure 7. This means that the South pole of the magnet is now the inside edge of the hole in that disc, so we’ve made a circular gap (between the top plate and the ring on the pole piece) that is very small, and therefore has a very concentrated magnetic field.
Fig 7. The top plate, also made of soft iron.
Unfortunately, we can’t just make a coil of wire and stick it in the gap and hope that it’s going to behave. Instead, we take a roll of cardboard (or something else) – a bit like the cardboard tube in the middle of a roll of toilet tissue – and wrap the coil of wire on that. That cardboard roll that supports the coil is called the “former” – it’s shown in Figure 8.
Fig 8. The light blue tube is the former, around which the voice coil is wound. You can’t see the voice coil because it’s hidden by the top plate. (Actually, you can’t see it because I didn’t draw it – but if I had, you’d just see some wires sticking out from the gap – depending on the type of coil we had.)
One little extra piece of information here. Since the voice coil, sitting in a magnetic field is the system that essentially converts electrical energy into movement, we call it the loudspeaker driver’s “motor”. Of course, it isn’t a motor that causes something to spin – but it does cause something to move.
Great. Now we have the coil of wire (the “voice coil”) wrapped around the former, sitting in the magnetic field. So far so good. Now we can put current through the wire and it will want to move in or out of the magnetic (depending on which direction we’re sending the current in). Now, our first problem is that, even if the voice coil and former moved out and in, there is nothing there to push and pull the air molecules in front of it – so it won’t make a lot of noise. So, let’s start putting up a surface that can move some air. We’ll start by putting on a “dust cap” which seals off the end of the former. This is the bump that you see on the front of a woofer in the middle of the cone – so we’re starting to get out to the visible “pretty face” of the loudspeaker driver. The dust cap is shown in Figure 9. Note that the dust cap is not always the same diameter as the former. Note as well that it is usually, but not always convex. Note as well that some drivers don’t have a discrete dust cap (like the BeoLab 3 woofer, for example).
Fig 9. The dust cap has been added to the front of the former.
Now we have a problem. We can put current into the coil and get it to move, but there is nothing there to stabilise it. What we need is something to make sure that it doesn’t fall down when you put the loudspeaker on its edge (as most are…). So, we’ll put in a stabiliser. It has to keep the former centred in the magnetic gap, but it also has to be flexible to allow the former to move in and out of the magnet. This part of the loudspeaker is called the “spider” – it looks like a disc that has wiggles in it that can stretch as the former moves in and out. This spider is shown attached to the former in Figure 10. Note that its outside will attached to something else, later.
Fig 10. The spider has been added. It is glued to the former, but is not attached to the coil or the top plate.
Welcome to later. Now we need a frame to attach the outside edge of the spider and some other parts of the loudspeaker to – as well as to allow us to attach the whole loudspeaker to a cabinet. This part is called the “basket” – it doesn’t do much other than act as a structural support for everything – a bit like the steel beams in a building. The basket is shown in Figure 11. It may be interesting to note that the basket for an automotive loudspeaker driver is a little different from one for a home loudspeaker because it has to be able to deal with the possibility of a nasty accident. For example, a friend who knows such things once told me that it’s a bad idea to put a woofer intended for a home loudspeaker in a car door because if you’re ever in a side impact collision, it’s not inconceivable that the magnet will rip away from the basket, shoot across the car and cut your legs off. So now I’ve warned you…
Fig 11. The basket is glued and/or riveted to the top plate. In addition, the outside edge of the spider is glued to the basket.
Now we can put the rest of the loudspeaker parts on. We attach a “diaphragm” or “cone” which makes the moving surface bigger. That’s the medium-dark green part in Figure 12. If we left it at that, when we moved the loudspeaker in and out of the magnet, it would sag, because the spider isn’t strong enough to keep the whole thing vertical. So, we add a “surround” which is usually made of foam or rubber (or fabric, in the old days). The surround is a flexible ring that is glued to the basket and the edge of the diaphragm. It’s the lightest green thing in Figure 12.
Fig 12. An entire moving coil loudspeaker. The light green ring is the surround and the darker green ring inside it is the diaphragm or speaker cone, glued to the top of the former and the dust cap.
So, now when you put current through the voice coil, it pushes out of (or pulls into) the magnet and moves the former, dust cap and diaphragm with it. This causes the spider and the surround (usually grouped into what we call the “suspension”) to stretch.
Fig 13. A cross section of a (not very) simplified model of a moving coil dynamic loudspeaker driver.
If we take the device in Figure 12 and cut it in half, we get a cross section like the one shown in Figure 13. And, just to prove that I’m not lying, I cut apart a real woofer – it’s shown in Figure 14. And then, not satisfied that I had done enough damage, I did it again to a BeoLab 3 woofer – those photos are in Figures 15 to 19. Another good example is this picture.
Fig 14. An actual moving coil dynamic loudspeaker, after I was very mean to it.
Fig 15. A BeoLab 3 woofer – after I was finished with it… You can see here that this particular loudspeaker driver does not have a separate dust cap and diaphragm. Also, you’ll notice that there is a flared cone that is used to connect the former to the outside edge of the diaphragm.
Fig 16. A BeoLab 3 woofer, showing some of the components.
Fig 17. A BeoLab 3 woofer, showing some more of the components. The magnet assembly is hidden inside the silver can at the bottom of the photo.
Fig 18. A BeoLab 3 woofer, showing some more of the components.
Fig 19. A BeoLab 3 woofer, showing some one more component.
That’s about it for this week. If you want to do a little more digging for yourself, you can look into things like the difference between overhung and underhung voice coils, neodymium vs ferrite, or just watch some relaxing, cool, tangentially-related videos like this one or this one or this one or this one. Or maybe just this.
Addendum
For the purposes of this explanation, I said that the top of the pole piece is the North pole of the permanent magnet, and the top plate’s inner edge is the South pole. However, there is no fixed convention for this. Manufacturers will almost always ensure that, when you put a positive voltage on the positive terminal of the loudspeaker, the diaphragm will move outwards. However, the north/south-ness of the magnet and the direction the voice coil is wound, and which end of the wire goes to which terminal vary not only from manufacturer to manufacturer, but model to model within one manufacturer’s portfolio.
One question that I am occasionally asked is why the BeoVision 11 is not able to decode audio signals encoded in Dolby TrueHD and DTS HD-Master Audio. I am not able to answer this question. However I can discuss what the implications are with respect to audio and, more specifically, audio quality in a playback system.
Before we start this discussion, let’s get a couple of terms defined:
LPCM: Linear Pulse Code Modulation. This is what most people call “uncompressed digital audio”. It’s the way digital music is stored on a CD, for example. It’s also the way digital audio is encoded to be able to send it around a computer or digital signal processor and do things like filter it, mix it, or just change the volume. So, at some point, in any system that has any digital audio signals anywhere in it (even if it’s just to change the volume, for example), it will be sending the signals around as LPCM-encoded audio.
CODEC: COmpression-DECompression. This is a way of encoding the LPCM audio signal so that it takes up less space (or less bandwidth). This is the audio equivalent of converting something to a “ZIP” file – sort of. Some CODEC’s are “lossless” – meaning that, if you take a signal, compress it and decompress it, you get back everything you put in – a bit-for-bit match of the original. (If you didn’t, it wouldn’t be “lossless”). Other CODEC’s are “lossy” – meaning that, when you compress the signal, some stuff is thrown away (this is why audio professionals call lossy CODEC’s like MP3 “data reduction” instead of “audio compression”). Hopefully, the stuff that’s thrown away is stuff you can’t hear – but that debate is best left out of this discussion.
Bitstream: Some Blu-ray players allow you to choose whether you send the audio data that’s on the disc directly out its digital output, unchanged OR to decode the audio data to LPCM before sending it out. Typically, this shows up in the menus on the player as a choice between sending “bitstream” or “LPCM”. So, for the purposes of this article, I’ll use these two terms as meaning those two things. However, if we’re going to be accurate, this is an incorrect definition, since an LPCM signal is a stream of bits, and therefore is also a bitstream. But now I’m being purposely punctilious…
Let’s start this discussion by building a fairly standard home theatre system. We have:
A Blu-ray player connected to…
A BeoVision or BeoPlay television or a BeoSystem X OR some other brand’s AVR (Audio Video Receiver) or Surround Processor via an HDMI cable
which is connected to…
A surround sound setup with 5 or 7 main loudspeakers (some of those might be inside the television) and maybe a subwoofer, all connected to the television using power amp, or line level (or PowerLink, for B&O customers) connections
Let’s also assume the following:
The Blu-ray player is connected to the television’s input with an HDMI cable.
Both the Blu-ray player and the television (or AVR or Surround Processor) have been certified by Dolby and DTS to decode all the encoding formats in which we’re interested for this discussion
(note that this is not true of all devices in the real world – but we’ll use the assumption for now).
The player does what you tell it to do. In other words, if you set its output to bitstream, it sends the stream of bits that are on the disc it’s playing – and nothing else.
The question is: What format should I use to send the audio signals from the Blu-ray player to the television? Should I set my Blu-ray player to send a Bitstream to the television and decode the audio signals there? Or maybe it’s better to tell the Blu-ray player to decode the signals and send LPCM to the television.
The answer, unfortunately, is potentially complicated… but let’s look at what the implications are for the two options.
Audio Quality
If you do a search on the web, you’ll find all sorts of answers to this question. Some of the answers are correct, some are partly correct, and some are just plain wrong – actually, some aren’t even wrong.
The basic reason for this is that Bang & Olufsen, like every other company that manufacturers audio/video components that support these formats must adhere to a very strict set of regulations that are set by Dolby and DTS in order to receive certification for our products. Also, before we can deliver a new product to our dealers, it must be thoroughly tested by both Dolby and DTS to ensure that we meet their exacting standards for their formats. In other words, a Dolby TrueHD decoder is a Dolby TrueHD decoder – regardless of what the box it’s in looks like.
This means that every device that is certified to decode Dolby TrueHD or DTS-HD Master Audio and convert that format to an uncompressed PCM audio signal (footnote: the “standard” format that is used by the internal Digital Signal Processing (DSP) of the device) does that decoding in the same way. This is true whether the device is a Blu-ray player, a DVD player, an AVR, or surround sound processor. Note that this does NOT, however, mean that a loudspeaker connected to any of these devices will give you the same sound quality – there are many, many other links in the audio chain that have an effect on sound. It merely means that the conversion from a compressed (either lossy or lossless) signal to an uncompressed signal will be identical, regardless of where it’s done.
So, whether:
you decode in the player and send PCM to the television, OR
you send the bitstream to the television and decode the signal there,
there is no difference in audio quality
Now, you might tell me “But I went to a user forum and saw a posting from a person who did a test with his Acme Blu-ray player and his Flybynight AVR and he reports that he can EASILY hear the difference in quality in the audio when he changes from LPCM to Bitstream on his player.” This may, in fact, be true. However, the reason that there is an audible difference is not due to one of the devices having a better decoder or some very minor issue like jitter on the HDMI signal. There are at least two very basic and very simple explanations as to why this might happen.
The first is a simple level difference. If you do an ABX test of a system where A or B is 1 dB louder than the other, but are identical in every other aspect, your listening test subjects will have no difficulty identifying what X is. (An ABX test is one where you can listen to 3 things – an “A”, a “B” and an “X”. “X” is guaranteed to be identical to either “A” or “B” (which are guaranteed to be different). Your task is to identify whether “X” matches “A” or “B”. So, if the Bitstream is just 1 dB (or less!) louder than than the LCPM version, then you will hear a difference – although you might not notice that the difference is a simple loudness. It’s likely that you will perceive the louder one as just being “better”.
The second is that some AVR’s can be set to apply different post processing to LPCM signals than they do to signals decoded internally. So, it’s possible that there could be something as simple as a bass or treble difference between the two signals – but it could be something as complicated as a lot of spatialisation and reverberation with a big EQ curve – or anything in-between. So, in this case, the difference between Bitstream and LPCM coming in from the player is in the post-processing differences in the AVR.
However, as I said, this is not the full story. Let’s look at more pieces of the puzzle to see what the differences are.
Audio during fast-forward / fast-rewind
Different players behave differently when you are fast-forwarding or fast-rewinding through a disc. If you are sending an encoded bitstream from the player, many devices will not deliver any audio when you are moving through your disc at a higher speed. This behaviour is different from brand to brand and model to model. However, in many cases, these same players WILL deliver audio if the output is set to PCM.
Listening Levels
In a theory, there should be no difference in audio levels (i.e. how loud the signals are) caused by moving from a decoder inside your player to one inside your television. However, as I mentioned above, this world isn’t perfect – and one of the results of that imperfection is that there may, indeed, be a difference in level caused by switching from a bitstream to a PCM output from the Blu-ray player. If this does happen, then it’s likely because of some extra (or different) post-processing that is happening to the signal(s).
Mixing of Extra Audio Channels
One of the cool features of DVD and Blu-ray (that I, personally, never use) is that you can watch a movie whilst listening to the director (or someone else) talk about the movie – a feature usually called “Commentary” or something like that. This is fun, because there’s nothing that can make IronMan 3 more interesting than hearing about how someone accidentally spilled a cup of coffee on their computer keyboard while they were rendering a 3D CGI version of an Audi falling into the ocean.
In order for this feature to work, the system has to take the audio signals for the movie and mix in an additional audio channel. However, some players are not able to mix these two sounds together if they’re outputting a bitstream. They can only do it if they’re decoding the movie AND the commentary separately, and then mixing the two LPCM streams. Of course, to then re-encode the result, just to send out a bitstream would be silly.
Another example of these extra sounds that may not make it through a bitstream output are the “clicks” or cute noises that are assigned to menu items.
So, with some players, if you want to hear these extra sounds, you’ll have to output a decoded LPCM stream.
Channel Allocation and Routing
This is where things get a little debatable. Usually, when people say “5.1” they mean the following channel allocations (in no particular order)
Left Front
Centre Front
Right Front
Left Surround
Right Surround
LFE
And, when they say “7.1” they mean
Left Front
Centre Front
Right Front
Left Surround
Right Surround
Left Back
Right Back
LFE
We’ll call these the “standard audio channel allocations” for this article. However, as I talked about it a previous article, “7.1” actually has seven “legal” variants that include non-standard audio channel allocations like front-wide and height channels.
Let’s assume, temporarily, that you have a disc that has some audio channels on it that should be routed to, say, a ceiling loudspeaker. The question is: “IF your surround processor has a “ceiling speaker” output, and IF there is a “ceiling” audio channel on the disc, can the signal get to the correct loudspeaker?” The answer is complicated. . .
Firstly, the question is “how can the disc ‘know’ that the audio channel is a ‘ceiling’ channel?” Well, both Dolby TrueHD and DTS HD-Master Audio (for example) have the option to include “metadata” (this is a fancy word meaning “data about the data” – or, in other words “information about the audio signals”) that can tell the decoder something like “audio channel #6 should be sent to a ceiling loudspeaker”. Other CODEC’s (like Dolby Digital, for example) do not have the possibility to have non-standard audio
Secondly, the question is “if the player decodes the signal to LPCM, and its decoder knows that the channel should be routed to the ceiling, can it ‘tell’ the surround processor via the HDMI metadata where to route the signal?” The answer to this question is dependent on the version numbers of the HDMI transmitter in the player and the HDMI receiver in the surround processor. If you are using HDMI 1.3 or earlier, then you cannot have non-standard channel allocations with an LPCM signal. If you have HDMI 1.4 or higher (for both the transmitter and the receiver) you might be able to get the correct metadata across from device to device. (if you look here, you can see that, if your HDMI transmitter or receiver is version 1.2 or earlier, then you cannot send the Dolby or DTS lossless codec’s – so this will also not work for non-standard channel allocations.)
So, the only way to guarantee that your complete system can support non-standard audio channel allocations (assuming that your surround processor has the ability to output them) is to send a bitstream from the player and decode at the end of the chain.
However, the question to ask after you’ve answered all of that is “how many commercially-available recordings include non-standard audio channel allocations?” The answer to this, as far as I’ve been able to figure out is “none”. (If you know of any examples of this that prove me wrong, please leave a note in the “replies” – I’m looking for materials! But read the rest of this paragraph first…) Of course, there are SACD’s where the LFE channel should be directed to a height channel – but SACD’s don’t include metadata to tell the player about the routing. Dolby ProLogic IIz and dts Neo:X have height channel outputs, but that’s a different system than a CODEC with discrete output channels. There are some other formats like Imax, Auro3D, and Dolby Atmos that use height channels, but they’re not available in consumer media.
So, this, at least for now, is an solution without a problem.
Bass Management
Some Blu-ray players (and even some good ol’ DVD players) include a bass management system that will filter the bass out of the main channels and add it to the LFE output.
IF your player can do this, then:
it can only do it to a decoded signal – so the bass management in the player will not work with a bitstream output
you should be sure that it is doing it (if you want it to do so) or that it is not doing it (if you don’t)
Latency
Some forum discussion groups have highlighted an issue with some specific players that exhibit lip-synch problems when they decode the signal to LPCM internally. The few comments about this that I have read in these fora indicate that switching the player’s output to bitstream appears to fix this problem. However, this should be considered a “work-around” for a bug in the player’s software. There should be no difference in synchronisation of sound and picture whether you’re decoding in the player or the surround processor. If there is a lip synch problem, then the people that made the player haven’t done their jobs properly.
Conclusion
So, to wrap up: the big things to remember here are that,
in any audio playback system, audio that is stored (and/or transmitted) in a CODEC has to be converted to LPCM somewhere
there is no difference in quality of decoder – in other words a Dolby (or DTS) decoder in one device won’t be better than a Dolby (or DTS) decoder in another device. If it were, then Dolby (or DTS) would not have approved one of them
there are some issues not related to audio quality that are affected by the location of the decoder in an audio chain, but these are typically very small (or even non-existent) issues for almost all consumers
Finally, I didn’t talk about jitter – which is term that a lot of people throw around as being one of those evils in every audio system where you can place all of your blame for everything that is bad about the system. This puts it in a league with things like “society“, “television“, “Canada“, and “The Boogie“. I’ll talk about that sometime in the future.
One last thing
Everything I’ve said above is only true for an HDMI connection. If you have an S/P-DIF or TOSLINK connection, then the discussion will be quite different, since you cannot have more than 2 channels of LPCM-encoded audio on those systems – so the only way to get multichannel audio through it is to use a lossy CODEC. So, if you have a multichannel source and you decode to LPCM before sending it out on S/P-DIF or TOSLINK, you will wind up with a 2.0 channel downmix.
Makes: One individual with reduced faith in loudspeaker reviews
Total time: approximately 2 hours
Directions
Take a woofer and put it in a cabinet
Connect an amplifier to it
Put it in a sauna
Set the room temperature to 20° C and wait until everything in the room is the same temperature
Measure the woofer’s on-axis response with a microphone
Look at the pretty plot of its magnitude response
Turn up the thermostat to 100° C and wait until the woofer warms up
Measure the response again
Look at the new pretty plot of its magnitude response
Scratch your head while you ask yourself why the two measurements look so different.
The setup
When you read a magazine review of a loudspeaker, it will include a measurement of its “frequency response” (more accurately called its “magnitude response”) which shows (ignoring a bunch of things) how loud different frequencies are when they come out of the loudspeaker assuming that they all came in at the same level.
However, as we saw in a previous article, for a Bang & Olufsen loudspeaker, this magnitude response is dependent not only on the loudspeaker, but how loudly you’re playing the signal.
Unfortunately, it gets much worse than this… For example, if we take a woofer (say, the one from the recipe above, for example) we can explain its electromechanical characteristics by breaking it down into different components (both actual and analogical). For example, the suspension (which is comprised of the surround and the spider) can be thought of as a spring. The electrical analogy for this is a capacitor. If you take all of the moving parts in the loudspeaker driver, they all add up to a mass that has to be moved – the electrical analogy for that mass is an inductor (since an inductor has some electrical “inertia” just like the mass of a bunch of loudspeaker bits). Some of the components are not an electrical analogy – they are real electrical components. For example, the voice coil, since it’s a coil, acts as an inductor. And, since it is a thin bit of wire, it also has some resistance to the flow of electrical current through it, so it’s also a resistor.
Fig 1. A simplified version of the actual electrical and electrical analogies of mechanical components in a loudspeaker driver.
If you look at the diagram above, you’ll see a very simplified “circuit” that shows the components of a moving coil dynamic loudspeaker. If these components don’t look familiar to you, don’t worry, it’s not important. Some components in the circuit are actual electrical things (like the resistance of the voice coil, shown in red) and others are analogies – electrical representations for a mechanical component in the system (such as a capacitor representing the “spring” of the surround and the spider).
If you know how each of these components behaves, and you know the correct values to put in for a given loudspeaker, and you know how to do the right math, then you can come pretty close to getting a decent prediction of the response of the loudspeaker that you’re modelling with the circuit. However, if you just put in one value for each component, then you’re assuming that they never change – in other words that you’re dealing with a “linear” system.
The problem is that this assumption is incorrect. For example, the Voice Coil Resistance – the amount that the wire in the voice coil resists the flow of current through it when the loudspeaker driver is not moving – changes with temperature. The hotter the wire gets, the higher the resistance goes. (This is a normal behaviour for most resistors.) If the voice coil resistance changes, then the whole system behaves differently, since it isn’t the only component in the circuit. So, as we change the temperature of the voice coil, the total response of the loudspeaker changes.
Sadly, the temperature of the voice coil isn’t only dependent on the room temperature as it seemed to be in our recipe for a Befuddled Speaker Enthusiast. As soon as you start playing sound with a loudspeaker, it starts heating up. The louder the signal (either because you turned up the volume or because your Metallica album just came on) the hotter it gets. So as you play music, it heats and cools. The speed with which it heats up and cools down is dependent on its “thermal time constant” – a big woofer with a giant magnet will take longer to heat up and cool down (and therefore have a longer thermal time constant) than a little tiny tweeter.
So, now you should have at least three questions that deserve answers:
How much does the temperature vary when I play music?
How does the response of the loudspeaker change with temperature?
How much does the response of the loudspeaker change with temperature?
What are you going to do about it?
1. Voice coil temperature
As I’ve talked about in a previous article, a loudspeaker driver is, give or take, about 1% efficient. That means that 99% of the power that you push into it (from the amplifier) is not converted into sound. Unfortunately, all of that power is lost as heat – almost all of it at the voice coil of the loudspeaker. So, the louder your music, the hotter your voice coil gets. Of course, if you have a way of cooling it (by using other parts of the loudspeaker as a radiator to your listening room) then it won’t get as hot, and it will cool down faster.
Let’s take a BeoLab 5 as an example (since that’s where we’re headed anyway…). Let’s take some relatively new-ish pop music (which has been mastered to be fairly loud due to a war that has been going on for years) and play it on a B&O player through Power Link (B&O’s version of a line level signal) at maximum volume on a BeoLab 5 whilst monitoring the temperature of the voice coils. What you’ll see if you do this is something like the plot below.
Fig 2. The temperatures (in °C) of the voice coils of the four drivers in a BeoLab 5 as a result of playing pop music at full volume on a BeoSound 5. The X-axis is the time in minutes. (green = tweeter, light blue = midrange, dark blue = mid woofer, red = woofer)
As you can see in the screenshot in Figure 2, the lower woofer (a 15″ driver connected to a 1000 W Ice Power amplifier) varied with a maximum of about 205° C. While it was playing this music at this level, it rarely dropped below 120°C.
This means that the difference in temperature of the woofer was 185°C at a maximum (205°C – 20°C) and rarely below 100°C.
In case you are wondering, this temperature cannot be measured directly, since it would destroy the voice coil if we tried to do so. Instead, what we do is to measure the temperature of the loudspeaker driver magnets, and use that real-time data input in addition to the signal that we’re sending to the drivers to calculate the temperatures of the voice coils based on thermal models of each of them. As you can see in Figure 3, below, the magnet temperatures are very different, and react much more slowly. These measurements were taken at exactly the same time as the ones shown in Figure 2. (Note that, although the mid woofer and woofer voice coils are roughly the same temperature, the mid woofer magnet is hotter than the woofer magnet by about 20°C or so. This just proves that their thermal models are different.)
Fig 3. The temperatures (in °C) of the magnets of the four drivers in a BeoLab 5 as a result of playing pop music at full volume on a BeoSound 5. The X-axis is the time in minutes. (green = tweeter, light blue = midrange, dark blue = mid woofer, red = woofer)
2 & 3. Loudspeaker response changes
So, now the question is “what does this change in temperature do to the response of the driver?” We’ll only deal with one driver – the woofer.
As I mentioned above, the thing that changes most in the model shown in Figure 1 is the loudspeaker driver’s voice coil resistance. For those of you with a background in reading electrical circuits, you may notice that the one shown in Figure 1 has some reactive components in it which will result in a resonance at some frequency. For those of you without a background in reading electrical circuits, what this means is that a loudspeaker driver (like a woofer) has some frequency that it “wants” to ring at – if you thump it with your thumb, that’s the frequency that you will hear ringing – a bit like a bell with a low pitch.
As the voice coil resistance goes up, its resistance increases, and we generally lose sensitivity (i.e. level or loudness) from the woofer. In other words, the hotter it gets, the quieter it gets. However, this only happens at the frequencies where the resistor is not “overridden” by another component – say the mechanical resonance of the woofer or the inductance of the voice coil.
The total result is shown for various temperature differences in Figure 4. Notice that these plots show the change in magnitude response of the driver with CHANGES in temperature. So, the blue curve at the top is the change in magnitude response (which is 0 dB at all frequencies – in other words no change) when the loudspeaker is playing at the same temperature it was measured at (let’s say, 20°C or room temperature). As the temperature of the voice coil increases above that temperature, you can see that you lose output in two frequency bands on either side of a “bump” in the response – this is at the resonant frequency of the loudspeaker driver.
So, the louder you play, the more low end you lose, apart from a peak in the response (which also rings in time) at the resonant frequency of the driver.
Fig 4. Sensitivity of an example woofer vs. the change in temperature of its voice coil in degrees Celsius
In case you’re wondering, the plot shown in Figure 4 is pretty close to the actual response of the 15″ woofer in the BeoLab 5 at different temperatures above room temperature.
The solution
Interestingly, all of the stuff I said above is true for every loudspeaker. So, if you’re the kind of person who believes that the only proper loudspeaker is one where you have nothing but a loudspeaker driver (in a cabinet of any kind, or not) and an amplifier – and no weird filtering or mucking-about going on, then you’ll have to live with the kind of unpredictable behaviour that you see above. This happens all the time to every dynamic loudspeaker. Since, in a passive loudspeaker, there’s nothing you can do about this (except for trying to keep the drivers cool somehow) you don’t often hear passive loudspeaker manufacturers talking about this little skeleton in their closet…
However, since a BeoLab 5 “knows” the temperature of the voice coil of the woofer, and since it has been programmed with the curves shown above in Figure 4, we can do something about it.
In essence all we need to do is to take Figure 4, flip it upside down and make a filter that “undoes” the effect of temperature on the loudspeaker’s response. In other words, if (because the woofer gets 160°C above “normal”) it drops 3 dB at 20 Hz, the BeoLab 5 knows this and adds 3 dB at 20 Hz. So, built into the BeoLab 5 is a set of filters that are used, depending on the temperature of the woofer’s voice coil. These filters are shown in Figure 5.
Fig 5. Magnitude responses of the compensating filter for the woofer from the previous plot vs. the temperature of its voice coil in degrees Celsius
It’s important to note three things here.
This can be done because we know the behaviour of the woofer at different temperatures (this was measured as part of the development process)
This can be done because the loudspeaker “brain” (the DSP) knows the temperature of the voice coil in real time as you’re playing music
This filter should only be applied to the woofer. The mid woofer and the other drivers have different behaviours and should not be affected by this correction curve. Therefore, this filtering can only be done because the BeoLab 5 is an active loudspeaker with independent filtering for each loudspeaker driver.
Some extra information
You should be left with at least one question. I said above that, as the music gets loud, the woofer heats up, so you lose output, so we add a filter that compensates by putting more signal into the driver.
“Waitaminute!” I hear you cry… “The problem is caused by the signal being too loud, so you make it louder!?” Well… yes.
However, there is one more trick up our sleeve. In a previous posting, I mentioned in passing that we also have Thermal Protection in almost all of the loudspeakers in the B&O portfolio. This means that the DSP brain knows the temperature of the drivers and, in a worst-case situation, turns the levels down to protect things from burning up. So, if we go back to the example of a BeoLab 5 playing at full volume, let’s see what’s happening to the signal levels.
Fig 6. The gains (in dB) applied to the signals sent to the four drivers in a BeoLab 5 as a result of playing pop music at full volume on a BeoSound 5. The X-axis is the time in minutes. (green = tweeter, light blue = midrange, dark blue = mid woofer, red = woofer)
These curves in Figure 6 show the gains applied to the four loudspeaker drivers in a BeoLab 5 at the same time as the measurements shown in Figures 2 and 3 were being made. In fact, if you look carefully at Figure 2 around the 23 minute mark, you’ll see that the temperature dropped – which is why the gain in Figure 6 increases (because it can!) in response.
Now, don’t panic. The BeoLab 5 isn’t screwing around with the gains of the drivers all the time. Remember that this test was done at FULL VOLUME – which, for a BeoLAb 5 is VERY LOUD. The gains shown in Figure 6 are a last-ditch effort of the loudspeaker to protect itself from a very mean customer (or the very mean children of a customer who is away for the weekend). This is the equivalent of the airbags deploying in your car. You know that if the airbags are outside the steering wheel (or if the gains in the BeoLab 5 dropped by 15 dB or so) something significant occurred…
Thanks to Gert Munch for his help in cleaning up the mistakes I made in the drafts up to and including the penultimate version of this article.
