Few figures have engendered more controversy among audiophiles than Bob Carver. Virtually every product he developed at Carver Corporation, which he founded in 1978, contained an unusual feature with an impressive-sounding name. Supporters hailed these features as important innovations. Detractors dismissed them as gimmicks.
Carver preamplifiers had Sonic Holography, a fancy handle for interaural crosstalk cancellation. His amplifiers had Magnetic Field power supplies. His tuners and receivers had the Asymmetrical Charge-Coupled FM detector. And there were the infamous “t-mods,” affordable solid-state amplifiers modified to match the “transfer function” (and thus the sonic character) of ultra-expensive high-end amplifiers (a la Mark Levinson, Conrad Johnson and Carver’s own Silver Seven).
In 1994, Carver left the company following a long-standing dispute with his board of directors. “My idea was to stay our original course,” he recalls. “We would be allied with independent dealers with well-trained salespeople, and our products would be unique and aimed at people who historically bought my kind of products. The board’s idea was to expand distribution, then explode the product line. I felt this approach was doomed to failure, because success in the audio industry means being very focused. It was frustrating to watch, and so I left.”
Losing control of their companies is a common fate for entrepreneurs who take their companies public. “It’s an oft-told tale,” Carver agrees. “It’s painful, but it’s not unusual.”
At Sunfire Corporation, the company he founded after leaving Carver Corporation, Bob Carver continues to create innovative designs with high-fallutin’ names - and to create controversy. Sunfire’s first product was a power amplifier with a tracking power supply, dubbed “the tracking downconverter.” The next was a tiny 11-by-11-by-11-inch subwoofer said to deliver the same performance as much larger subs. The specified output of the True Subwoofer’s tiny amplifier - 2,700 watts - has raised more than a few eyebrows.
When asked to explain their products, many high-end designers break into a bullshit-baffles-brains monologue. Carver, by contrast, is lucid and down-to-earth, not what one would expect given the unabashed self-promotion in his companies’ advertising and literature. In person, what comes across is Carver’s gentlemanly demeanour, love of music and passion for audio - and his wonderfully creative approach to equipment design.
GB: One of the foundation technologies for Sunfire is the tracking downcoverter, which you use in your power amplifiers and subwoofers. What does it do?
BC: A regular amplifier has a power supply rail that’s parked up high in the sky. The audio signal varies under the power supply rail. The difference between the instantaneous value of the audio signal and the power supply rail is converted into heat - a great deal of heat. In the tracking downconverter amplifier, the power supply tracks the audio signal, so that the power supply rail is always six volts above the instantaneous value of the audio signal, rather than staying at 100 volts as in a regular amplifier.
The tracking downconverter was an idea I had about 18 years ago. It’s a generic form of a tracking power supply. I got a patent for it, but could not get it to work. It would always blow up. I gave up after a year of around-the-clock work, and decided to make the magnetic field amplifier instead. That is the product I used to start Carver Corporation, and it was very successful. A few years ago, I decided to have another go at the original circuit, and got it to work in spades. That was the basis of Carver Corporation’s Lightstar amplifier, and also the Sunfire amplifier.
GB: How did you get it to work?
BC: It’s so obvious now. I had to slow down the transition speed in the switching devices in the power supply. Engineers are always taught that the faster the switching, the more efficient the device. With the tracking downconverter, very rapid switching generates electromagnetic interference. That causes an electrostatic field inside the switching device, which causes a short circuit, and the part blows up.
It’s not the rep rate that causes the failure, it’s the transition rate. The rep rate is the frequency - how many times it switches per second. The transition speed is how long it takes to make the switch. Reducing the transition speed was the breakthrough. Once the transition speed was controlled, things stopped blowing up.
GB: In a conventional audio amplifier, you have AC coming in, then a power transformer to convert the line voltage to the voltages needed by the amplifier, followed by a bridge rectifier to convert to DC, storage capacitors to act as a reservoir and probably some power supply regulation to keep rail voltages steady. The audio circuitry functions as a tap, modulating the flow of energy from that big reservoir. How does your design differ from this model?
BC: Instead of feeding the amplifier directly, the reservoir after the bridge rectifier and power transformer feeds the input to the tracking downconverter. The output of the tracking downconverter feeds the amplifier. As far as I know, it’s the only tracking power supply in existence for home audio amplifiers. It may be the only tracking power supply in existence for anything.
With a tracking-power-supply amplifier, we can get the same performance as a regular amplifier, with reduced cost. Or, for the same cost, we can improve power output and the ability to drive low impedances by a factor of three or four. What we did was improve performance. The tracking downconverter is what gives the Sunfire amplifier its almost-magical ability to double its power continuously as impedance falls. You get 300 watts at eight ohms, 600 at four ohms, 1,200 at two ohms. And of course the amplifier doesn’t get hot.
A tracking power supply is a sort of holy grail of amplifier designers. Everyone worth his salt thinks, “If only I had a tracking power supply, my amplifier would have immense power and immense headroom. It wouldn’t get hot. It would be able to drive virtually any real-world load with ease without a huge heat sink and 100 output transistors.”
GB: Why have only you done it, and not all these others who dream of it?
BC: Once, when I was giving a talk at the Boston Audio Society, I was asked that very question. The founder of Analog Devices raised his hand and said, “I have an answer for that. Because it’s hard to do.”
GB: With a signal that suddenly becomes louder, does slowing down the transition cause compression, or does your six-volt headroom prevent that?
BC: The power supply is slightly faster than the amplifier. The speed of the amplifier and power supply are related to bandwidth. The power supply has ever-so-slightly higher bandwidth than the amplifier itself. The amplifier is no slouch; it’s just under 100 kHz. The power supply is a little bit faster - actually it would be fine if it were just the same speed - and then it gets a six-volt head start. Between those two things, the amplifier tracks any transient you can blast into it, as long as it’s band-limited to 100 kHz.
GB: How do you plan to apply tracking downconverter technology in the future: to push performance higher, or to push price and amplifier size downward? It looks like you’ve been doing more of the former. Do you plan to do more of the latter?
BC: No. Not at all.
GB: Do you plan to license the technology to other companies?
BC: We’re doing that right now, but not in the high-end audio field. It’s being done for professional and subwoofer amplifier applications.
GB: Your power amplifiers have two sets of output terminals: one for current source and one for voltage source. What’s the point?
BC: The voltage source has very low output impedance like a solid-state amplifier, so it has the sound of a solid-state amp. The current source has output impedance like a vacuum-tube amp, so it sounds more like a vacuum-tube amplifier. With the current-source output, the output impedance is exactly one ohm. We did a survey of tube amps and they average about an ohm.
We recommend driving an electrostatic or ribbon loudspeaker with a current source, and driving the low-frequency driver with a voltage source. Most people who have panel loudspeakers bi-wire them using that suggestion, and they like it better. There’s a sweet glow to the midrange and a soft high end that you associate with tube amps; along with the tightly controlled bass you expect from a good transistor amplifier. When you hook the panel up to the voltage source, it sounds like a transistor amp. The midrange is a little laid back and doesn’t have that glow; and the highs are sometimes a little sibilant. GB: Does the difference between tube and solid-state sound come down just to output impedance?
BC: Not entirely, but almost entirely. Ninety per cent of the sound quality that we typically attribute to vacuum tubes comes from the output impedance.
GB: Many people think the output transformer, and associated factors like soft clipping and core saturation, create the tube sound. Does the transformer raise output impedance, and thus create tube sound?
BC: The forward impedance of the output tubes and the transformer together usually comes out at about 10 ohms. When you use 20 dB of feedback, it reduces it to around one ohm. That’s basically what a vacuum-tube amplifier is all about.
GB: If output impedance is what creates tube sound, why did you build the Silver Seven tube amplifiers at Carver? Why not just add a resistor to a conventional amplifier, as you’ve done with the Sunfire amps to generate the current-source output?
BC: Building a big tube amplifier was a dream I had carried with me since my childhood, when I saw a big McIntosh in a window. I went to college, went into the service, got married, had children. When I finally got around to building my dream amplifier, it got rave reviews by guys like Bascom King and Harry Pearson. Those guys love tubes in the first place.
GB: After you’ve designed an amplifier, does listening to it change the design?
BC: Not appreciably. But it changes some of the finishing touches. After I built the original Sunfire amplifier, I listened to it alongside a Silver Seven. The Silver Seven sounded like a tube amp and the Sunfire sounded like a solid-state amp. I modified the circuits slightly, so that the Sunfire sounded as close as I could possibly make it to the vacuum-tube amp. Part of that was putting in the one-ohm resistor; but there were other things we had to do to get the two amps to sound as much alike as possible. Essentially, this involved giving it the distortion profile of a tube amplifier - in the current-source setting.
GB: How audible are the differences between solid-state amplifiers?
BC: What I’m going to say will fly in the face of what most people believe. I believe that you can take two solid-state amplifiers, and provided neither one is overloaded in any fashion, they’ll sound identical. That’s a big if. Amplifiers are overloaded in three basic ways. They’re overloaded in amplitude; they’ve overloaded in current; they’re overloaded in speed. It’s very easy to do this if you don’t have a big juicy amplifier. Obviously a little Radio Shack amplifier is not going to be able to touch a big Jeff Rowland or a Mark Levinson or a Sunfire amplifier. Provided the amplifier has flat frequency response and sufficiently low distortion, both of which are trivial these days, and provided there are no interface problems, the differences will always be the subtle differences associated with overload, either momentarily, like slew-rate limiting or clipping, or just running out of drive current.
GB: So the main problem is overload, which you can overcome with brute force, i.e. a really big amplifier, or elegance, something like a tracking downconverter. Was the tracking downconverter what enabled you to build a very small subwoofer?
BC: Having left Carver I wanted to build a product that people would really notice. A very small subwoofer would certainly attract attention…provided it had performance. The smallest box I could fit a driver with a 10-inch basket into was 11 by 11 by 11inches.
GB: The power specification for the True Subwoofer - 2,700 watts - has been controversial.
BC: No one believes it, because the amp’s the size of a candy bar, albeit a big $2.25 Hershey bar. When Don Keele, a world-respected subwoofer scientist and reviewer, measured it, he measured 2,500 watts continuous for at least two minutes. And it’s capable of twice as much peak output - maybe 5,000 watts.
GB: Is it just raw power that enables you to get so much bass from so small a box?
BC: The True Subwoofer is a small box with a small driver that produces the same output and bass extension as a large box with a large driver. It has not been possible until now to build a small subwoofer with this kind of output. Conventional drivers are not very efficient at converting input power into motion. They operate close to “stall mode” where a lot of power is turned to heat. To get lots of output from a small box, we have to increase input power by a factor of 10 or more. Two hundred watts becomes 2,000 watts, and that burns out the voice coil. What was necessary was to make a driver that would remain highly efficient even in a small box. This requires a big magnet and lots of motion, which pulls the driver out of stall. When it’s out of stall, it’s efficient and there’s far less heat.
This flies in the face of conventional driver design, because that type of driver generates vast amounts of back emf [back electromotive force, which results when the motion of the voice coil in the magnet gap generates a voltage]. The back emf prevents current from flowing through the voice coil, so there isn’t any output. To overcome the back emf, you need a special amplifier that can swing 100 volts, rather than 20 or 30 volts. When you compute the input power - 100 volts squared divided by four ohms, the resistance of the speaker - that’s over 2,000 watts. That’s what the amplifier would deliver if the load were purely resistive. But it’s not. It’s a loudspeaker load with high back emf. The real power delivered to the voice coil is only a few hundred watts.
GB: Is that the explanation of the controversial power specification of the True Subwoofer? That its amplifier delivers a few hundred watts into the resistive component of the load, but that it has to be capable of delivering far more power in order to deal with the reactive component of the load.
BC: You can think of it in those terms. That’s called “imaginary power.”
GB: I lose the ability to think when people start talking complex arithmetic and imaginary numbers.
BC: It’s hard. I’d forget about it at this point.
GB: What was the basic insight that led to the True Subwoofer?
BC: Necessity is the mother of invention. This woofer began two-and-a-half years ago with an eviction notice from my wife Diana. “Get this stuff out of the room,” she said. “If you want to have a big woofer in the room, it has to be invisible.” I began thinking of ways to do this. I’d think about when I was driving around, when I’d go to an opera, when I was going to sleep at night. Gradually, it occurred to me that one of the fundamental defining equations for efficiency, (Bl)2/re [where B is magnetic strength, l is voice-coil excursion, and re is voice-coil resistance], is true over all frequencies, not just for the midband as the equation is classically interpreted. It turns out I was right. That meant I could increase output and bass extension by using a big magnet and long-excursion driver.
GB: It sounds like a very unusual driver design.
BC: Oh yes. It requires careful attention to the suspension, surround and the magnetic geometry to keep it linear. They can’t collapse under the box pressures, which are around 150 pounds. So the rubber-rolled surround has to be sufficiently rigid so that it doesn’t cave in and act like a passive radiator in reverse.
GB: How do you obtain acceptable transient response with materials sufficiently massive to withstand these pressures and produce these excursions?
BC: Transient response is the mirror of frequency response. If you have good amplitude response, you have good transient response. If you have good transient response, you have good frequency response. The trick is to get the amplitude response correct, which is challenging enough. It requires good motor structure, good surround design, a good amplifier, the proper equalisation and so on.
GB: What about sensitivity?
BC: Sensitivity is so low it’s preposterous. It’s not so much the mass, although that does drive down sensitivity. The main thing is the back emf associated with the large magnet. The sensitivity is somewhere in the 60s or 70s [dB @ 1 watt @ 1 metre]. But it doesn’t matter because I’ve got this super amplifier to drive it.
GB: Isn’t distortion high with such a small box and driver?
BC: At full output, distortion is less than 10 per cent from 18 Hz upward. Most subwoofer manufacturers don’t specify distortion, because it’s very high: 10, 20 per cent or more. Even the Velodyne, which has a high-gain servo system, has one per cent distortion at low levels; but at full output, distortion is close to 10 per cent.
GB: How tolerant is human hearing of distortion at very low frequencies?
BC: Incredibly tolerant. It’s all but impossible to hear distortion at very low frequencies.
GB: Some speaker manufacturers maintain that distortion is important at those frequencies.
BC: Believing in low distortion is like believing in motherhood and apple pie. Low distortion is a good thing, and every manufacturer strives for ever-lower distortion. But it’s well-known how much distortion we can hear, and at what levels and under what circumstances we can hear it. If we apply those rules artfully, we can design products with vastly improved performance.
Take DVD as an example. DVD contains immense amounts of distortion as a result of the compression algorithms used for audio and video. A lot of information is thrown away. The components in the signal that are tossed out represent tremendously high distortion. But we don’t notice it because the designers know what’s important and what’s not.
Philosophically and conceptually, any audio product can be approached that way. It’s impossible to hear 20 per cent distortion if the thing’s operating at 9 Hz, because the first overtone is below the threshold of audibility unless it’s so loud that it’s hurting your ears � 130 dB or more.
GB: What about when you move higher in frequency?
BC: The thresholds change, but the rule still applies.
GB: In addition to the active driver, the True Subwoofer has a mass-driven unit that moves out when the active woofer moves out, and moves in when the active unit moves in. You’d expect the outward movement of the active unit to pull the mass-driven cone in, and vice-versa. How is the mass-driven woofer made to operate in phase with the active woofer?
BC: The force that moves the driven woofer is the electricity going through the voice coil. The force that moves the other woofer is the reaction force. To understand how this works, imagine standing on a chair and holding a Slinky in your hands. Pretend there’s a little weight on the other end of the Slinky. Your hand represents the voice-coil-driven woofer. The springs in the Slinky represent the spring of the air in the box, and the mass on the bottom represents the mass glued onto the other woofer. If you start moving that Slinky up and down with your hand, when your hand goes down, the bottom end of the Slinky comes up, and when your hand comes up, the bottom end of the Slinky goes down. It’s just the opposite of what you’d expect.
GB: That’s when the rate of movement is above the resonant frequency of the system.
GB: So you get this spring effect in the subwoofer whenever the input frequency is above the resonant frequency of the subwoofer, which, given the driver design, must be very low.
BC: It’s 18 Hz.
GB: Is the new 13-inch version of the True Subwoofer just a bigger cousin, with a little more output and lower extension?
BC: The Signature Subwoofer is the same system, but slightly larger. It has 6 dB more output, 116 dB as opposed to 110 dB, and it’s tuned to 16 Hz rather than 18 Hz. It uses the same amplifier, but larger drivers � equivalent to a 10 1/2-inch driver rather than eight inches.
GB: Your Theater Grand processor has some nifty ease-of-use features, such as the ability to set itself appropriately when you turn on a source. But it has a bunch of Carver-sque features as well. What does the video auto-correlator do?
BC: It’s mostly for video tapes. It’s a dynamic circuit that artificially extrapolates additional bandwidth that has been truncated by the VCR. It sort of reconstructs the signal. That is summed into the video circuits and presented to the CRT. If it’s done artfully, it gives all the appearance of a much more sharply delineated picture without ringing - that little bright line around objects that you get when you turn the sharpness control up. There’s a lot of what I call “psycho-optics” involved.
GB: What do you mean by psycho-optics?
BC: I mean fooling the eye into believing that there’s more definition on the screen than is actually present in the program material.
GB: What about the Holograph Generator and the Wall-to-Wall sound options?
BC: The Theater Grand has two kinds of music modes. Wall-to-Wall is a digital mode, which is similar to the jazz and arena and all that DSP-generated stuff we’re familiar with. Then we have an analog music mode, which is my Sonic Holography. I originally designed Sonic Holography for two-channel systems. On the Theater Grand it’s optimised for two-channel sources, except when it’s in the Theater mode. Then it’s optimised for home theatre. It takes care of the front hemisphere of the soundstage and adds some of the cues that allow front-to-back dimensionality, which usually disappears when you use three front speakers. Normally, three front speakers give a flat curtain of sound across the front stage. To get a sense of depth, you have to add time delay to the centre channel. With holography, that delay occurs in our ear-brain system because of the holographic image. The side channels fill in the rest of the room.
GB: For home theater where it’s not practical to have three identical to have identical speakers across the front, do you favour the use of centre-channel speakers?
BC: I don’t know. I’m such a two-channel fan. The illusion that can be created from a two-channel source by our ear-brain systems decoding the timing cues the spatial cues can be so incredibly beautiful and so deliciously three-dimensional, that it’s difficult to warm up to five speakers. Five speakers provide a nice sense of envelopment and immersion, but it’s different from the three-dimensionality that a good two-channel system can have. But for theatre, I believe a centre channel is better to lock the dialog in.
GB: Is sonic holography essentially crosstalk cancellation?
BC: Yes. The crosstalk cancellation has been adjusted to produce a realistic presentation on regular stereo recordings. Years ago, Harry Pearson wrote a wonderful series of articles on generating a soundstage with a stereo system. What you want is a sense of layered depth on a soundstage that extends behind the speakers and is wider than the speakers. I’ve adjusted Sonic Holography to really give a soundstage in spades using Harry Pearson’s treatise as a model.
In real life, for every sonic event we have two sound arrivals: one for the each ear. Every time I snap my fingers, you hear two sounds: one in your left ear, one in your right. In three-channel playback, every time I snap my fingers, you might hear six � one in each ear for each of the three speakers. Those unwanted sounds need to be cancelled if we’re to generate a replica of real life.
GB: But this reduces the sweet spot.
BC: Yes it does. It was one person wide when I started, and I managed to get it to be three people wide.
GB: What does the five- and seven-channel music mode do?
BC: It splits the soundfield into two hemispheres. The front hemisphere uses our latest thinking on Sonic Holography. The rear hemisphere gets a mix of L+R and L-R information. With a seven-speaker setup, more of the L+R information is apportioned to the forward position; and more of the L-R information to the rearward speakers. The Sonic Holography gives very precise imaging in the front hemisphere of the instruments within the soundfield. The rear hemisphere immerses one in the soundfield because it has a great deal of hall ambience, but very little precise imaging.
GB: You said you love two-channel and what it can do. Do you like this better than regular stereo?
BC: I don’t know yet.
GB: The Theater Grand does DTS as well as Dolby Digital decoding. What’s your view of DTS? Is it going to confuse the market?
BC: I hope not. DTS seems to be settling in to a peaceful coexistence with Dolby Digital in the marketplace, with DTS appealing mainly to a dedicated set of people. That’s our market really.
GB: Does it sound better?
BC: I can’t say.
GB: If it doesn’t sound better, what’s the point? It’s another system that does the same thing as Dolby Digital.
BC: We know off the bat that it has higher bit rate. Is that audible? I haven’t been exposed enough to it to know. But even if it isn’t today, it always has to hold the promise of being better because there’s more stuff there.
GB: What do you plan to do next at Sunfire?
BC: Right now, I’m working on my Cinema Ribbons. They’ll be sold as a set of five. Each speaker is a just a little bigger than your fist. But it goes down to 80 Hz and it puts out the same SPL as a big box with two eight-inch drivers and at least one centrally located tweeter.
GB: How do you get that kind of output from a small speaker?
BC: By applying the same back emf and high box pressure concept as my subwoofer. It uses a special ribbon driver with very long excursion and large magnet. That produces very high back emf and box pressures several times that of an ordinary speaker, which lets the speaker go down to 80 Hz and produce high SPLs.
I call it the world’s first fly-by-wire loudspeaker. The system includes a controller that goes between the processor/preamp and power amplifiers. It’s absolutely fundamental to the system. It has comprehensive speaker protection; a low-frequency excursion limiter, which I call the low-frequency auto-throttle; and a fixed excursion limiter for frequencies beginning around 200 Hz. It uses feedforward to reduce low-frequency distortion and make the frequency response nice and flat. It also performs active equalisation, with three positions for speaker location - in the corners, against walls or out in the room.
GB: When DVD-Audio appears, with multiple channels and higher-resolution coding, what do you think it will do for music in the home? Will it be a repeat of the quadraphonic disaster, or are we smarter this time around?
BC: It should be able to work if it’s done properly. I would like it to be encoded in a stereo two-channel format for the front hemisphere, with some kind of crosstalk cancellation to give a more realistic presentation, and then use the additional channels as ambience fill for the back hemisphere, so that you can be immersed in sound.
GB: Do we need more than 16 bits and 44.1 kHz sampling? Will 96/24 sound better?
BC: I’m a great believer in being artful with existing technology. If you do a great job with what’s available, you can get something that’s absolutely superb. Ordinarily, when you make a change, you stumble for a while until the new technology is brought up to the level of the old one, even if the new technology has the capability of ultimately surpassing the old.
GB: We saw that with the transition from analog to digital.
BC: Absolutely. In spades.
GB: Can home audio reproduction get much better than it is now?
BC: It can get a lot better. Some day, your audio system is really going to fool you, so that you can close your eyes and believe you’re in the presence of a real live orchestra. Today even the best systems can’t do that.
GB: That sounds like getting the room to disappear. Some companies have spent a ton of money trying to do that, and haven’t succeeded.
BC: Either the room has to disappear, which is very difficult � maybe insurmountably difficult, or we have to make the room part of the acoustic venue, and work with the room. That would be my approach.
GB: But you listen to an orchestra in a hall hundreds of feet long and wide. The largest home listening rooms are tens of feet long and wide. How do you bridge that gap?
BC: Here’s one way to do it. Get two small speakers, like Cinema Ribbons, and put them at arms’ length in front of you in the middle of the room. Then use crosstalk cancellation, and make a special recording so that what we would hear is so much in the near field that the bounces from the room walls are insignificant. Or we could use a very large room with lots of sound absorbing material - lots of traps everywhere. Then use lots of speakers, or two speakers with Sonic Holography, or a combination of both.
Whatever technique we use, some day in the future there will be audio systems that can fool us. It probably won’t be in my lifetime, but we’ll have it.
Gordon Brockhouse Analyzes How The Sunfire Designs Work
The number-one enemy of virtually any electronic or electro-mechanical product is heat. Excess heat shortens the life of electronic parts - sometimes with dramatic abruptness. Designers often have to add weight and cost to the products to deal with heat issues; often they have to compromise performance as well.
Rather than getting rid of excess heat, Sunfire’s power amplifiers and subwoofers prevent heat from becoming a problem at all. Here, paraphrasing Bob Carver’s white papers on the True Subwoofer and Sunfire amplifier, is how they work.
The Tracking Downconverter
In a conventional amplifier, the voltage on the power supply rails is constant � typically 90 volts or so. Let’s imagine such an amplifier driving a four-ohm loudspeaker, which at a given moment in time (during a peak) is drawing 10 amperes. The output voltage will be 40 volts. Ten amperes are flowing through the output transistors; and there are 50 volts across the transistors (the difference between the 40 volts flowing to the loudspeaker and the 90 volts on the power supply rail). That means that 500 watts are not going to the output, but being turned into heat.
This example corresponds to 400 watts output power (assuming a purely resistive load). With the same amplifier putting 400 watts into an 8-ohm load, the output transistors produce 238 watts of heat. At 100 watts into a 4-ohm load, heat output is 350 watts. With an 8-ohm load, heat output is 218 watts. At 25 watts, heat output is 225 watts with a 4-ohm load and 134 watts with an 8-ohm load.
This has two consequences. First, large, heavy heat sinks are needed to conduct all this heat away from the output transistors. Secondly, the power supply must be large enough to drive the loudspeaker load and generate all that heat.
The 20-ampere output transistors commonly used in high-end and mainstream amplifiers are rated for 200 watts. That’s the amount of heat they can dissipate. That effectively limits the amount of current the transistors can produce. Going back to our first example, when there are 50 volts across the transistor, it can safely deliver 4 amperes into a 4-ohm load. When there are 70 volts across the transistor (as there would be at 100 watts output into a 4-ohm load), the device can deliver 3 amperes. To deal with low-impedance loads, several of these must be used in parallel. This drives up heat sinking and power supply requirements.
Sunfire amplifiers use a tracking power supply (the “tracking downconverter”). The voltage on the power supply rail is kept at a steady six volts above the output signal. That means that the voltage across the output transistors is always 6 volts. At 400 watts output into a 4-ohm resistive load, the output transistors generate 60 watts in heat. At 100 watts, they generate 30 watts. With heat output so low, there’s no need for heat sinking; and power supply requirements are dramatically reduced.
And because there are only 6 volts across the output transistors at all times, they can always deliver their rated 20 amperes. The Sunfire I uses 12 of these devices per channel, giving it peak-to-peak current capability of 240 amperes.
The True Subwoofer
To produce sound, you have to move air. To produce lots of deep bass, you have to move a lot of air. You can use a very large radiating surface with limited back-and-forth movement. Or you can use a smaller radiating surface that moves a lot.
Most subwoofers take the former approach. Sunfire’s takes the latter. Its two drivers (one active, one passive) have cone excursion of 2 1/2 inches. The diameter of the radiating surface is 8 inches. The combined air displacement is 251 cubic inches. According to Sunfire, this is comparable to three or four conventional 15-inch drivers installed in a large cabinet.
There is a linear relationship between cabinet volume and power requirements. If you want to reduce cabinet volume by half, you have to double input power to produce the same output. Moving from a two-foot cube to a one-foot cube, you have to increase amplifier power by a factor of eight to maintain the same output level.
The Sunfire True Subwoofer’s cabinet is very small - 11 by 11 by 11 inches. To get lots of deep bass from a cabinet this small requires a very powerful amplifier. Sunfire rates its subwoofer amplifier at 2,700 watts!
The driver design that displaces so much air also prevents it from burning up. Conventional moving coil speakers are inefficient at turning input power into cone motion. To move the cone, they require a lot of current passing through the voice coil. This produces heat. More input power means more heat.
You can increase driver efficiency by using a more powerful magnet and longer voice-coil excursion. Normally, this is completely impractical. This combination results in high amounts of back electromotive force (emf). The motion of the voice coil in the magnet gap produces a voltage. If the amplifier cannot overcome the voltage produced by the motion of the speaker, no current can flow into the voice coil, and there is no sound.
For such a driver to work, you need an amplifier capable of very wide voltage swings. The tracking downconverter amplifier used in the Sunfire subwoofer can swing over 100 volts. Because of heating problems and space limitations, it would be impossible to use a conventional amplifier that could swing 100 volts into a four-ohm load. But because a tracking downcoverter amplifier runs cool, does not require heat sinking or a large power supply, it can be made small enough to work in this application.
With this kind of amplifier, back emf becomes an asset. The Sunfire sub’s long driver excursion and huge magnet (225 ounces) produce large amounts of back emf, which prevents damaging amounts of current from flowing into the voice coil. Consequently, the voice coil does not burn out. The huge magnet enables the long excursion even though the amount of current flowing in the voice coil is small.