#2 EXPANSION OF DYNAMIC RANGE WHAT IS DYNAMIC RANGE? Dynamic Range, DR, for a stereo, is the difference between the loudest and smallest sound that the system can reproduce. For human hearing, DR is about 120dB, with OdB being the smallest audible sound and 120dB being very loud. The scale is logarithmic, each 10dB = ten times, so 120dB = 100,000,000,000 times; which presents immense engineering challenges. Early in High Fidelity, the Long Playing record achieved a DR of 75dB, on a good day. The signal source was electromechanical, with stylus deflection being the analog of amplitude, which set the upper limit. So a chain of components ending in a pair of speakers that could manage 75dB could (possibly) reproduce the DR on the LP. As an industry, we did a good job at this. When the CD came along, the upper limit for DR was bumped to 103dB, that limit being set by the 16 bit word standard. Wow! From 75 yo 103dB! That was huge! Nearly 30dB, so nearly 1000 times. That upset the applecart. I was there, building speakers. It was a stretch. Some guys did OK, many did not; their product was noticeably lifeless by comparison. The point then was: conventional tech barely made the cut, with some effort. Today, HD’s 24 bit enables DR that surpasses that of human hearing, meaning that it is finally possible for recorded music to have ideal DR. The recording guys can capture it, but can we playback guys transduce it? Can we possibly wring another 15dB out of conventional tech? Nope. Ain’t gonna happen. Can’t. The proof is out in the open. When you hear guys saying that they don’t hear much difference in the dynamics between CD and HD, they’re right! But not for the reason they think. The bottleneck is not on the recording side, it’s on the playback side. Their speakers are not dynamically capable AND their passive filters are obliterating the improvement in resolution. Let’s take a look inside DR reproduction in the speaker. A speaker’s job it is to transduce an electronic time/voltage signal into time/sound pressure in your room. Perfect Dynamic Linearity means that: for each doubling of the input, you’ll see a doubling of the output, across the entire audio band. Here’s a 3D graph showing what perfect Dynamic Linearity in a speaker would look like. But what actually happens is something less than ideal. Instead of a 1: 1 relationship between input and output, you get something less. And the shortcomings are typically not evenly distributed across the audio band. In most cases the woofer gives up first; it can’t keep up with the rising outputs of the midrange and tweeter. No, surprise, really, its job is much harder. This is the reason why so many systems “glare” when you crank them: the tonal balance is shifting upwards. I’m sure you’ve noticed that the frequency /amplitude curves presented to you in the specs and magazines are taken at modest levels. No indictment, here, it’s very difficult to measure, in room, at high pressures. But you should know what’s going on; you should not expect that a “pretty” freq/amplitude curve will still be so at high amplitudes. WHAT’S CAUSING IMPERFECT DR AT THE WOOFER? #1 - The motor itself is nonlinear. Motor force at the cone falls off on both sides of center position. This happens for two reasons: as the voice coil moves fore and aft, more and more of it moves outside the magnetic gap, so the force generated falls off. And a less-than- perfectly elastic suspension ‘tightens’ as the moving system excurds, mechanically limiting it. This problem is even worse than it looks because the two performance functions that we want most are governed by adverse square functions. Bear with me, please. First, because your perceptions are logarithmic, each time you want the system to sound twice as loud, you actually need it to play four times as loud, so four times the excursion. Plus, each time you ask for lower bass, let’s say you’d like to go from 40Hz to 20Hz, halving the frequency, you’ll need four times the excursion. So, as you ask your system to play some combination of bass heavy music and nice amplitude, you are actually demanding huge amounts of excursion. That’s why your woofers bottom out. The critical point, here, is that the greater the percentage of time the coil is in the declining force zones, the less faithful it is to the signal’s DR across the entire range of the woofer. That’s why you hear ‘congestion’ on difficult music, which is not as evident on simpler tunes. What you’d like to see, and happily pay for, is motor behavior that looks more like the second plot. A few of today’s best makers of cutting edge woofers are taking special care on this aspect, knowing that customers, like me, are discerning. (Has anyone ever told you this before? No? Because they don’t know it, themselves.) #2 - Thermal compression drags down output, an effect that, unfortunately, sums with force/position effect. As you demand more from a woofer, in addition to causing more excursion, you are also pushing more current thru the voice coil. So it gets hot. Law of nature: heat generated is a square function of of current, so the effect come on quickly. Problem? Yes. Resistance rises linearly with temperature. As the coil warms, the resistance rises. Remember the woofer in series with the resistor (in section one)? Lower output, higher Q, loss of amp damping. But it’s worse: like the coils in your toaster (which are supposed to get hot) the rise time and cool-off times are slow. So when you punch a bunch of current thru your coil, the hotter it gets, the more the output falls. It takes a few milliseconds to hurt you, then the effects linger. As distortions go, this one is an real mess. What can we do about it? Two things are obvious: keep the coils cool and to spread the heat out among multiple woofers, thus multiple coils. Examine these two Satori woofers, a 9.5” and a 7.5”. You can see the heroic means of keeping the voice coil temperatures down. First, oversized coils spread the heat out and increase the radiating area. All things being equal, a larger coil would also spread out the magnetic flux that motivates it. So, by increasing the coil size, we do need to ramp up the magnetic circuit, and exponentially. So it gets expensive. Then, generous venting thru the pole piece, thru the center of the coil with both ends flared for smoother and quieter flow, a symmetrical spider that fully exposes the forward end of the coil and a perforated coil former form a system that effectively pumps cool air across the coil. Implicit in this thermal design is an extremely coercive neodymium boron drive system that focuses an intense magnetic field across the voice coil, raising sensitivity, so requiring less current. As you will see, one of the advantages that Next Gen tech brings is the ability to put more drivers into less space. So we can go ahead and pack baffle with woofers, which both beneficial to the system’s output/bandwidth envelope because they can do more work, but also to DR. Completing the strategy for keeping both temps and excursion low is a DSP “trick” that results in less current at the coil and less demand for excursion. Every bass system has a natural low frequency roll-off; a region where it can make little useful output. For all of these years, in conventional systems, we been sending full range signal into that region (shown in red), both heating it and forcing useless excursion. We can’t do this with passive filters, but the idea is simple, in the processor we filter out the unwanted energy but just fitting a high pass filter to the acoustic output; the blue line shows the filter. This approach cleans up the working part of the band and improves DR greatly because it both helps keep the coil cool by reducing total current flow, plus eliminates unproductive excursion. This is a rarity in engineering: all upside, no downside. A free lunch! To recap: multiple advanced woofers designed for motor force linearity and coil cooling, plus a high pass filter. It sounds simple, now that we understand it, no?
#2 EXPANSION OF DYNAMIC RANGE WHAT IS DYNAMIC RANGE? Dynamic Range, DR, for a stereo, is the difference between the loudest and smallest sound that the system can reproduce. For human hearing, DR is about 120dB, with OdB being the smallest audible sound and 120dB being very loud. The scale is logarithmic, each 10dB = ten times, so 120dB = 100,000,000,000 times; which presents immense engineering challenges. Early in High Fidelity, the Long Playing record achieved a DR of 75dB, on a good day. The signal source was electromechanical, with stylus deflection being the analog of amplitude, which set the upper limit. So a chain of components ending in a pair of speakers that could manage 75dB could (possibly) reproduce the DR on the LP. As an industry, we did a good job at this. When the CD came along, the upper limit for DR was bumped to 103dB, that limit being set by the 16 bit word standard. Wow! From 75 yo 103dB! That was huge! Nearly 30dB, so nearly 1000 times. That upset the applecart. I was there, building speakers. It was a stretch. Some guys did OK, many did not; their product was noticeably lifeless by comparison. The point then was: conventional tech barely made the cut, with some effort. Today, HD’s 24 bit enables DR that surpasses that of human hearing, meaning that it is finally possible for recorded music to have ideal DR. The recording guys can capture it, but can we playback guys transduce it? Can we possibly wring another 15dB out of conventional tech? Nope. Ain’t gonna happen. Can’t. The proof is out in the open. When you hear guys saying that they don’t hear much difference in the dynamics between CD and HD, they’re right! But not for the reason they think. The bottleneck is not on the recording side, it’s on the playback side. Their speakers are not dynamically capable AND their passive filters are obliterating the improvement in resolution. Let’s take a look inside DR reproduction in the speaker. A speaker’s job it is to transduce an electronic time/voltage signal into time/sound pressure in your room. Perfect Dynamic Linearity means that: for each doubling of the input, you’ll see a doubling of the output, across the entire audio band. Here’s a 3D graph showing what perfect Dynamic Linearity in a speaker would look like. But what actually happens is something less than ideal. Instead of a 1: 1 relationship between input and output, you get something less. And the shortcomings are typically not evenly distributed across the audio band. In most cases the woofer gives up first; it can’t keep up with the rising outputs of the midrange and tweeter. No, surprise, really, its job is much harder. This is the reason why so many systems “glare” when you crank them: the tonal balance is shifting upwards. I’m sure you’ve noticed that the frequency/amplitude curves presented to you in the specs and magazines are taken at modest levels. No indictment, here, it’s very difficult to measure, in room, at high pressures. But you should know what’s going on; you should not expect that a “pretty” freq/amplitude curve will still be so at high amplitudes. WHAT’S CAUSING IMPERFECT DR AT THE WOOFER? #1 - The motor itself is nonlinear. Motor force at the cone falls off on both sides of center position. This happens for two reasons: as the voice coil moves fore and aft, more and more of it moves outside the magnetic gap, so the force generated falls off. And a less-than-perfectly elastic suspension ‘tightens’ as the moving system excurds, mechanically limiting it. This problem is even worse than it looks because the two performance functions that we want most are governed by adverse square functions. Bear with me, please. First, because your perceptions are logarithmic, each time you want the system to sound twice as loud, you actually need it to play four times as loud, so four times the excursion. Plus, each time you ask for lower bass, let’s say you’d like to go from 40Hz to 20Hz, halving the frequency, you’ll need four times the excursion. So, as you ask your system to play some combination of bass heavy music and nice amplitude, you are actually demanding huge amounts of excursion. That’s why your woofers bottom out. The critical point, here, is that the greater the percentage of time the coil is in the declining force zones, the less faithful it is to the signal’s DR across the entire range of the woofer. That’s why you hear ‘congestion’ on difficult music, which is not as evident on simpler tunes. What you’d like to see, and happily pay for, is motor behavior that looks more like the second plot. A few of today’s best makers of cutting edge woofers are taking special care on this aspect, knowing that customers, like me, are discerning. (Has anyone ever told you this before? No? Because they don’t know it, themselves.) #2 - Thermal compression drags down output,an effect that, unfortunately, sums with the force/position effect. As you demand more from a woofer, in addition to causing more excursion, you are also pushing more current thru the voice coil. So it gets hot. Law of nature: heat generated is a square function of of current, so the effect come on quickly. Problem? Yes. Resistance rises linearly with temperature. As the coil warms, the resistance rises. Remember the woofer in series with the resistor (in section one)? Lower output, higher Q, loss of amp damping. But it’s worse: like the coils in your toaster (which are supposed to get hot) the rise time and cool-off times are slow. So when you punch a bunch of current thru your coil, the hotter it gets, the more the output falls. It takes a few milliseconds to hurt you, then the effects linger. As distortions go, this one is an real mess. What can we do about it? Two things are obvious: keep the coils cool and to spread the heat out among multiple woofers, thus multiple coils. Examine these two Satori woofers, a 9.5” and a 7.5”. You can see the heroic means of keeping the voice coil temperatures down. First, oversized coils spread the heat out and increase the radiating area. All things being equal, a larger coil would also spread out the magnetic flux that motivates it. So, by increasing the coil size, we do need to ramp up the magnetic circuit, and exponentially. So it gets expensive. Then, generous venting thru the pole piece, thru the center of the coil with both ends flared for smoother and quieter flow, a symmetrical spider that fully exposes the forward end of the coil and a perforated coil former form a system that effectively pumps cool air across the coil. Implicit in this thermal design is an extremely coercive neodymium boron drive system that focuses an intense magnetic field across the voice coil, raising sensitivity, so requiring less current. As you will see, one of the advantages that Next Gen tech brings is the ability to put more drivers into less space. So we can go ahead and pack baffle with woofers, which both beneficial to the system’s output/bandwidth envelope because they can do more work, but also to DR. Completing the strategy for keeping both temps and excursion low is a DSP “trick” that results in less current at the coil and less demand for excursion. Every bass system has a natural low frequency roll-off; a region where it can make little useful output. For all of these years, in conventional systems, we been sending full range signal into that region (shown in red), both heating it and forcing useless excursion. We can’t do this with passive filters, but the idea is simple, in the processor we filter out the unwanted energy but just fitting a high pass filter to the acoustic output; the blue line shows the filter. This approach cleans up the working part of the band and improves DR greatly because it both helps keep the coil cool by reducing total current flow, plus eliminates unproductive excursion. This is a rarity in engineering: all upside, no downside. A free lunch! To recap: multiple advanced woofers designed for motor force linearity and coil cooling, plus a high pass filter. It sounds simple, now that we understand it, no?