"Half a watt goes in here . . ."
". . . must come out there!"
"I’m talking about power, Chucko."
"I know you are, or you wouldn’t be here."
"Look! It heard the word power and responded, just like wedo!"
Those of you who are old enough might remember this routine fromFiresign Theatre’s classic comedy album I Think We’re AllBozos on This Bus. Those merry jokesters were talking about amechanical model of the government, but I’m talking about sonicpower, Chucko–raw, un-adulterated audio wattage.
In any sound system, this undeniable force is supplied by a poweramplifier. The technology that underlies this indispensable device hasremained essentially unchanged over the past 40 years, ever sincetransistors replaced vacuum tubes. But now, at the dawn of a newmillennium, we also face a new era in power-amp technology. Theindustry is making quantum leaps as it pursues its goal of designingpower amps that are smaller, lighter, less expensive, and more powerfulthan previous generations.
EM author Rudy Trubitt examined the design philosophies behind thepower amps of several major manufacturers in his article "The Power andthe Glory" (see the August 1993 issue). One might think that little haschanged over the past half-dozen years; after all, power amps seem likethe type of product that evolves at a glacial pace. Indeed, somedesigns have changed very little, but several companies have developednew or significantly revised designs since Trubitt visited the subject.Clearly, it’s time to take a fresh look.
To this end, I interviewed representatives from five power-ampmanufacturers: Mackie Designs, Hafler, QSC, Crown, and Velodyne SiliconSystems. Some of these companies have begun to manufacture power ampsrelatively recently, while others are well-established names in thefield. Of course, their spokespeople agree on some points, butit’s what they disagree about that can be especiallyenlightening.
Learn Your ABCs
Before we delve into the fine points of power-amp design, let’sreview a few basics. As its name implies, a power amplifier boosts thepower of a signal. Typically, a power amp accepts a line-level signaland increases its voltage and/or current without changing the shape ofthe input waveform. The amplified signal is sent to a speaker, whichconverts the signal into acoustic sound waves. Power amps are used inthree primary applications: studio monitoring, live soundreinforcement, and instrument amplification.
Unlike most studio gear, a power amp draws some serious current fromthe AC outlet. This current is converted to DC by the amp’s powersupply. In a traditional power supply, a power transformer decreasesthe incoming AC voltage, which is then converted to a DC voltage by aset of diodes and several large capacitors. One common type of powertransformer is called toroidal because it looks like a doughnut (ashape known as a toroid in mathematical terms). This shape plays animportant part in how the transformer functions because thetoroid’s magnetic field is confined more to its core, reducingleakage into the audio circuitry. Toroidal transformers are made ofiron for its electromagnetic properties.
The DC voltage from the power supply is symmetrically arrangedaround the ground point (0V). For example, the output from the powersupply might be ±50 VDC. These positive and negative voltages,which are called the power-supply rails, operate the amp’sinternal circuitry.
In particular, these voltages provide power to a set of outputtransistors, which perform the actual amplification. The outputtransistors amplify the input signal by drawing power from a set ofcapacitors in direct proportion to the input signal’s voltage asit varies over time. As the capacitors discharge in this process, theyare replenished by the power supply.
The power-supply rails determine the maximum amplitude that the ampcan produce. For example, if the rails are at +50V and -50V, the ampcan produce signals of nearly 100 volts peak to peak. If the ampproduces a signal that exceeds this limit, the tops and bottoms of thewaveform are cut off; this is called clipping.
One of the most important characteristics of any power amp is theefficiency with which it uses AC power to amplify the input signal.Unfortunately, most conventional designs are very inefficient, usingless than 50 percent of the AC power they draw from the wall. Theremainder of that power dissipates as heat within the amp. Most poweramps therefore require large heat sinks, and many use fans to cooltheir components. In addition, many include thermal-protectioncircuitry, which shuts down the amp if things get too hot.
That Amp’s Got Class
Power amps fall into several classes, the simplest being A, B, andAB. Class A power amps have output transistors that handle both thepositive and negative swings of the waveform. Class B designs have oneset of transistors that handles the positive swings, and another thathandles the negative swings. Class AB amps also have separate sets oftransistors to handle the positive and negative swings, but there issome overlap as the signal changes polarity; when the signal is near0V, both sets of transistors are conducting.
Most of Mackie’s current power amps, including the ones withinthe company’s powered speakers, use a conventional Class AB designwith conventional power supplies. According to Cal Perkins,Mackie’s director of new technology (and self-described corporatecynic), the company’s amps have the most efficient Class AB designin the marketplace because they use lightweight toroids, which reducethe transformer’s weight by about 50 percent.
Hafler uses a variation of Class AB called trans-nova, designed byJim Strickland in 1980. Jerry Cave, Hafler’s managing director,explains: "It’s a different way of using transistors in thecircuit, requiring fewer gain stages and a much simpler signal path.This lets us get voltage and current gain out of both transistorsinstead of one or the other. Trans-nova combines the linearity of ClassA with the efficiency of Class AB. Class A is typically only 25 percentefficient; trans-nova is typically 65 percent efficient."
Some amp designs use multiple power-supply rails. When the peaks ofthe output signal are small to moderate, the low-voltage rails areused; most musical material falls within this range most of the time.When the output peaks are large (as in momentary, loud transients), thehigh-voltage rails are used. If the same set of output transistors isused with both sets of rails, this is called a Class H design; ifdifferent transistors are used with the different rails, it is calledClass G.
The multirail approach is more efficient than single-rail designs.According to John Subbiondo, marketing manager for QSC, "Transistorsare most efficient when they are all the way on; if they’re onlyhalfway on, half the power is lost to heat. With a multirail design,the transistors are closer to being fully on more of the time."Consequently, multirail designs can lower the AC current draw andcooling requirements by as much as to 40 percent. QSC’s mostpowerful amps use four rails on each side of ground.
The New Switcheroo
Another way to improve the efficiency of a power amp is to use aswitching power supply (also known as a switch-mode, active, orelectronic power supply). Long used in computers and other devices, aswitching power supply converts the incoming 60 Hz AC power to a muchhigher frequency, often in the 200 to 500 kHz range. This improves theperformance of the transformer (the behavior of which is frequencydependent), allowing smaller, lighter transformers to be used.
QSC uses switching power supplies in its PowerLight series of amps.Subbiondo cites several additional advantages to amps that use thisapproach: "You can make them very quiet; you don’t have a big humfield from a large transformer, and the hum field you do have isoutside the audio range. In addition, you can have a purer path fromthe audio circuitry to the speakers. And because the transformer issmaller, it tends to have lower impedance and fewer losses, whichtranslates into less voltage sag and better performance under highdemand."
However, Perkins points out that switching power supplies have theirown set of problems. "True, a switching power supply eliminates most ofthe low-frequency magnetic fields, but it produces a lot ofhigh-frequency noise. This noise can be eliminated, but in many casesit isn’t."
Many people believe that using a switching power supply results in amore open sound with a clearer high end, but that it also inherentlycompromises bass response. Others say that this problem is related topoor design implementation: correct design provides better voltageregulation, which results in improved low-frequency performance overconventional power supplies.
Mick Whelan, vice president of new product development at Crown,sees this debate as similar to the one about analog versus digitalaudio. "Some people love it and others hate it," he says. "A lot ofpeople believe that you don’t get good bottom end without a goodchunk of iron."
Perkins believes that this debate is more smoke and mirrors thananything else. "A power supply is nothing more than anenergy-conversion and energy-storage mechanism," he says. "It sucks ACfrom the wall, converts it to DC, then reconverts it to a modulatedsignal, which is the audio. This process is a function of the totalenegy-storage capacitance and/or inductance of the system. If you havethe same energy-storage capability, it doesn’t matter if it’sin a capacitor or an inductor; the number of joules is the same."
Perkins gives an example from his past experience working at JBL: "Idesigned an amp for JBL with a switching power supply that had moreenergy storage than the biggest iron supplies they had at the time. Acompeting product on the market at the time had roughly the sameaverage power level but had one-twentieth the energy storage. When weconnected the two products to a pair of speakers and cranked up thesound using material with a lot of bass, it was like listening toentirely different speakers. There was no bass in the competingproduct."
According to Subbiondo, switching power supplies offer anotheropportunity, called power-factor correction. "This is a way of drawingcurrent from the wall more efficiently," he says. "Supplies thataren’t power-factor corrected do something called peak-voltagerectification, in which the filter capacitors charge up only when theincoming AC voltage equals or exceeds their reservoir voltage. It drawscurrent only at that time, so you get big current spikes. Power-factorcorrection lets you continuously draw current over the entire ACwaveform. This lowers peak AC requirements by 40 percent, which mightnot mean much in a 200-watt amplifier, but it makes a big difference ina 9,000-watt amp."
Switcheroo, Part Deux
One of the most interesting recent developments in power-amptechnology is the switching power amp, also known as Class D. In thisdesign, the input signal is converted into a pulse-width modulated(PWM) square wave by alternately turning two output transistors on andoff in the 100 kHz range. (Class D amps are therefore sometimes calledPWM amps.) This square wave is then processed through a lowpass filter,which yields an amplified version of the input waveform.
Theoretically, Class D topology offers some significant advantagesover more conventional designs, the most important of which is muchhigher efficiency. This means that you need less heatmanagement–which translates into lighter weight, smaller size, andlower cost. Class D amps have been used in powered subwoofers for sometime, partly because their distortion is less audible in the low range.One of the premier manufacturers of powered subs is Velodyne Acoustics,which recently started a new division, Velodyne Silicon Systems (VSS),to develop Class D amps for other manufacturers to use in theirproducts.
However, this type of amp presents a number of significant obstaclesto high-fidelity operation. According to Bill Ciullo, vice president ofengineering at VSS, "Class D is very difficult to do well. In the realworld, it tends to sound terrible unless you do some fancy designtricks. For example, if there is any overlap or gap between thetransistors turning on or off, you get major distortion." This can evencause the amp to literally explode!
In addition, Cave says, "Class D amps have output inductors that actas filters. When you change the impedance of the speaker load, thefrequency response of the amp changes. If you know the impedance of thedriver you’re using, it works fine."
Ciullo says that Tripath was the first company to make Class D ampssound good (although, in typical marketing fashion, Tripath calls itsdesign "Class T"). "They solve the overlap/gap problem at the front endby adding broadband noise to the input signal," Ciullo says. "Thistends to mask the switching errors, which are averaged in with thenoise. This approach is somewhat similar to digital dithering. However,it requires DSP at the front end, which is expensive, and thetransistors are switched in the megahertz range, which presents its owndifficulties.
"Velodyne’s founder, David Hall, solved the problemdifferently, on the back end of the process," Ciullo continues. "He putan inductor between the two transistors. When they are ready to switch,the energy is temporarily stored in the inductor until the switch iscomplete, at which time the energy is released to the transistorthat’s on. This process is controlled by two more transistors.There is no voltage drop across the transistors when they switch, sothey are not stressed at all."
Interestingly, Velodyne’s design uses no power transformer.Rather, it rectifies the wall voltage and produces ±82 VDC rails.(Standard wall voltage is 120 VAC RMS; peak-to-peak is 164V.) Classes Aand AB can’t use rails at these extremes because they’d havetoo much heat to dissipate; the transformer steps the voltage down toreduce heat. According to Ciullo, Velodyne’s design is more than97 percent efficient; its 250-watt and 600-watt Class D amps need noheat sink or fan, and they are quite small compared with moreconventional designs.
Crown’s K series of full-range Class D amplifiers uses astrategy similar to Velodyne’s (that is, an inductor between theoutput transistors); however, the two companies have separate patentson their designs. Crown’s approach is called Balanced CurrentAmplifier (BCA). "This is an enhanced Class D topology," Whelan says,"that lets us make a 2,500-watt amp [1,250 watts per channel into 2ohms] in a two-unit package with convection cooling. No other topologycan do that and maintain a frequency response of 20 Hz to 20 kHz withlow distortion.
"BCA efficiency is around 90 percent," Whelan continues. "However,measuring efficiency with sine waves is a waste of time. The energydensity of a sine wave is very different from that of actual music; asine wave has much more energy than most music signals. For example, anamp that’s 90 percent efficient with sine waves might be only 40percent efficient with music." Perkins agrees with this assessment,adding that the efficiency of an amp changes dramatically when youmeasure it at the typical average level of most music, which is only 20to 30 percent of full output.
Class D amps have a potentially bright future as "digital"amplifiers. According to Ciullo, "Essentially, the output of thetransistors is ‘digital’ in that it oscillates between thetwo rails. In the near future, we’ll see Class D amps that accepta digital bitstream, decode it, and use it to determine when thetransistors should be turning on and off. This would keep the signal inthe digital domain all the way to the lowpass filter and would be avery noise-free design. You would have no chance to pick up analognoise anywhere along the way."
Power to the Speakers
Another hot topic in the world of power amps is the concept ofpowered speakers, in which a power amp is mounted within a speakercabinet. This subject sparks lively debate in the audio community.
Cave points out some of the advantages to this approach. "Thespeaker’s drivers and enclosure act as a system with impedances,pressures, and so on. If you drive it with an external amp, youdon’t know what kind of performance you’re going to get ifthe amp is not matched to that speaker. With a powered speaker, you canmatch the amp to the speaker in terms of power, impedance, andcrossover points, which optimizes everything."
Perkins agrees. "The main advantage of powered speakers is that youcan optimize their performance with the internal electronics," he says."You can dramatically change the sound of any speaker. For example, theMackie HR824 uses basically the same midrange and tweeter as anotherpopular powered speaker. The bass driver is similar, too, and theenclosure is the same size. But they sound totally different from eachother because of the electronics package."
On the other hand, some people point out the disadvantages ofpowered speakers. Subbiondo offers an example: "Instead of running onecable to your speakers, you have to run a signal cable and a powercable, which can introduce EMI into the signal cable. Also, ifyou’re flying the speaker in a live venue and it fails,you’re stuck. Powered speakers are well suited for monitoring,which is why they’ve seen so much success there. But whenyou’re powering many speakers with the same signal, it can be morecost-efficient to use a single, large amplifier."
Whelan sees an emotional factor, as well. "One of the beauties ofhaving a separate amp is that people have their favorite speakers andamps. Separating them gives you the flexibility to assemble your ownsystem and get the sound you’re looking for. When you put an ampin a speaker, you remove some of that flexibility."
Power amps are essential to any sound system, and their basicfunction is unlikely to change in the future. However, the relativemerits of the different ways they perform their function will bedebated as long as engineers devise new methods of amplifying an audiosignal.
With 30 years of experience in this field, Perkins recognizes twoschools of belief when it comes to power amps: "One school saysthey’re all the same; there are no differences. The other schoolsays that each design sounds different–and manufacturers use thisto tout their own products as better than the next company’s."
To which school does Perkins subscribe? "The reactive load of aspeaker is totally different from the resistive load on a test bench,"he says. "Because of that, you do have sonic signatures and differencesbetween products and manufacturers. Some of these differences aresubtle, and others can easily be discerned even by someone whodoesn’t care a whole lot."
Of course, the readers of EM do care. So if you’re in themarket for a power amp, listen critically to several models in yourprice range and take along recordings that you know well. Auditionamplifiers that use different topologies so that you can becomefamiliar with their sonic signatures. In addition, keep your purposefirmly in mind. As Cave says, "There are a lot of different designs,and you have to choose the one that is best suited to yourapplication."
Scott Wilkinson, a contributing editor to EM, knows that whatgoes in must come out.