Amplifier Anatomy - Part 2

By: Patrick Quilter,
Chief Technical Officer QSC Audio Products, Inc.

Power amplifiers, those “unsung heroes” of the sound system, have traditionally been the old reliable that consultants and contractors learned to count on as everything else in the sound system became more complex. While the rest of the system demanded more and more attention, the power amps were something you just didn’t have to worry about.

Although power amplifier technology has more to offer and requires more thought, there are some basic principles that every leading amplifier manufacturer follows. Understanding how equipment designers are solving amplifier problems can give you a new appreciation for this basic piece of equipment.

In Part I, we described how an amplifier works, how power supplies affect amplifier function and which amplifier classes work best for various situations. In Part II, we will go into more detail about amplifier circuitry and design.

NEGATIVE FEEDBACK AND DISTORTION In Part I, we mentioned that transistors are not inherentlyperfect magnifiers. Most of the advanced circuit techniques we described involve re-assembly of the audio waveform in various ways. If we had no way of correcting errors, the result would typically be a rather garbled and harsh reproduction.

Fortunately, we have a powerful error-reduction technique called negative feedback.

You probably associate negative feedback with criticism. Actually, this perception is not so farfetched, but in electronics there is nothing bad about “negative” feedback. This technique is basically the same process we use daily when observing one’s progress and making mid-course corrections.

Let’s use driving as an example. Imagine if you tried to steer around a corner with your eyes closed.

Even if you turned the wheel about the right amount at about the right place before you closed your eyes, the car would soon leave the road because of uncorrected small errors. In the real world, you drive with your eyes open. You turn the wheel, observer how the car is turning, and then make small corrections to maintain the desired course. This is a basic example of how we use feedback to correct for errors and produce the desired result.

We can also use feedback in amplifiers. The output of real-world circuitry is distorted: The output is higher or lower than desired. We usually don’t know just what the errors are, or we could correct for them. We can correct for unknown errors by comparing the output to the input and telling the circuit to increase or decrease the output until they match.

To describe the actual process, let’s assume we want an amplifier with a gain of 10. The actual, imperfect amplifier has a gain that varies unpredictably from 8 to 12, resulting in up to 20% of output error. We attenuate, or reduce, the actual output of the amplifier by a factor of 10. If the amplifier were error-free, this reduced feedback signal would be a perfect match to the input signal, but the actual feedback signal varies around the input signal by 20%.

We have an accurate picture of just the error because we have compensated for the desired gain by the 10-to-1 attenuation. Now, here’s the tricky part. If we magnify the apparent error by amplifying the mis-match between the input signal and the feedback signal, then combine this information with the original input signal, we can get the main amplifier to reduce its own errors automatically.

We have a crucial choice, however. The error signal can be added or subtracted from the input. If we add the error signal, it just makes the error worse. This is called positive feedback, and it turns transistors into oscillators. It turns sound systems into oscillators, also. In an amplifier it leads to wild, runaway operation. If, on the other hand, we subtract the error signal, the error is always diminished. Thus, we have used negative feedback.

In practice, we reduce error by putting much more gain than we need into the amplifier. It we increase the gain of the amplifier by a factor of 10, the gain, even with errors, will range from 80 to 120. When we “close the feedback loop,” using the 10-to-1 attenuation, the error signals always a large positive value. The amplifier quickly reduces its own output until the feedback and input signals match up.

In effect, because the amplifier has extra gain, it is in a constant state of “holding back,” which makes it easier to hit the desired target. With enough extra gain, it is ultimately the accuracy of the feedback circuit itself, not the amplifier, that determines the accuracy of the final output.

There is a limit, however. All circuitry has a slight lag between input and output. If you increase the gain of the circuit too much, it will become too sensitive, and the combination of lag and feedback will cause hunting or oscillation around the desired value. This problem, called instability, limits the amount of feedback that can be used.

The only fundamental cure is to reduce the circuit lag by using high-speed components. There has been a lot of progress in this area in the last 20 years, and today’s transistors are about 10 times faster. This progress probably explains how solid-state amplifiers have gradually eliminated the harshness that some listeners heard in early amplifiers, which used relatively slow power transistors.

Feedback can be applied around all of the cascaded elements in the amplifier. This process, called global feedback, is popular because it corrects all internal errors in one swoop. Some designers prefer a process in which they sub-divide the amplifier into several cascaded feedback loops, called local feedback. Certain forms of instability are easier to eliminate with the local feedback technique, but internal signal levels must be higher when one feedback loop feeds the next. QSC amplifiers use global feedback because it is less expensive and is easier to assure accuracy with only one set of critical feedback components to worry about.

One notable effect of high feedback is on amplifier clipping characteristics. Without feedback, transistors approach saturation gradually, giving the effect of cushioning the impact. This type of clipping is called soft clipping. With feedback, the output signal is forced to stay on track until the last possible moment, resulting in an abrupt impact, called hard clipping, which sounds more fuzzy than soft clipping. Some amplifiers feature low-feedback designs to smooth out the clipping.

Because it is so hard to reduce distortion without high feedback, the trade-off is often between clean amplifiers with harder clipping and slightly mushier amplifiers with smooth clipping. The choice depends on personal preference and on how much the amplifier will be overdriven.

In any case, it is important to avoid sticky clipping. Poor feedback circuit design can make the amplifier track its input too far, and then snap back and ring without damping. The amplifier should enter and leave clipping cleanly with no snapping or chattering.

PROTECTION CIRCUITRY The lower the impedance of the load, the greater the current drawn from the amplifier, and the greater the heat generated in the output transistors. If too many loudspeakers are connected to the amplifier, or if the ends of the loudspeaker wire touch together by accident, the load impedance goes very low, and the current flow becomes dangerously high. If the flow is not limited, the output transistors will burn out. Therefore, amplifiers need some kind of short-circuit protection. There are many ways to protect against short circuit, but the trick is that you can’t prematurely limit performance into normal loads. QSC amplifiers continuously monitor the load impedance. Loads above 2 ohms draw safe currents, and only normal voltage clipping occurs. Below 2 ohms, the maximum amplifier current is reduced if it exceeds the safe limit for more than a fraction of a second. This way, short peaks are permitted even into marginal loads, yet the amplifier is still protected against gross, sustained overloads.

Other common protective circuits include turn-on and turn-off muting, shut-down or muting in case of excessive temperature, protection against radio pickup (RFI), and dc fault protection, which shuts down the amplifier is a transistor loses control. The design of these circuits is just as important as the actual audio circuitry because they can make the difference between surviving an accident or winding up with burnt rubble.

SWITCHING AMPLIFIERS Remember those classes D, E and F that we promised we would talk about in Part I? As you have seen, heat in the output transistors is a major problem, and it is inherent in the way linear amplifiers operate. Whenever the amplifier is delivering only part of its power to the load, waste heat is created.

There is another way of converting dc power into audio power that reduces the inherent heat losses.

Because the losses occur when the output transistors are partially on, we avoid this state. We turn them fully on and send all of the dc power to the load, or we leave them fully off so that no power flows. In both cases, little or no power is wasted in the transistor.

To get the desired average power in the load, we rapidly switch the transistors on and off for the desired percentage of the time, with the time the transistors are on varying from 0% to 100%. The switching must occur much faster than the highest audio frequency if the averaging is to work correctly. This is the switching amplifier or class-D amplifier. (Classes E and F are special cases, as is class C, that apply mainly to non-audio uses.) How does on-off switching drive a loud-speaker? The magnetic field in the voice coil does not collapse instantly when the amplifier switches off. The loudspeaker continues interpolating a fraction of the waveform while the amplifier is switching.

Class-D amplifiers are only now becoming practical. The speed requirements for the switching transistors are 50 to 100 times greater than for linear audio amplifiers. The high-frequency switching causes radio interference, and many practical problems must be solved to attain the same audio fidelity that we expect with linear amplifiers. When the switching causes radio interference, and many practical problems must be solved to attain the same audio fidelity that we expect with linear amplifiers. When the switching is perfected, we can expect the heat sinks to be about one-fourth the size, reducing amplifier size and cooling requirements. When combined with the switching power supply, we will have a size and weight breakthrough comparable to the transition from tubes to solid-state. It will take a great deal of experience to overcome the reliability problems associated with complex new circuitry. This, too, should be an active area for development in the 1990’s.

MECHANICAL DESIGN We have treated amplifier design as an electronic problem, but the physical size, location and ruggedness of the parts is at least as important. You can inspect these features by eye at trade shows or on other occasions when the insides are on display.

The power transformer is the single heaviest element and must be mounted securely. Also, look for adequate clearance around it to allow for cooling and shifting in case the amplifier is dropped.

The power transformer is the single heaviest element and must be mounted securely. Also, look for adequate clearance around it to allow for cooling and shifting in case the amplifier is dropped.

The height of the chassis is another important variable. Low-profile chassis allow you to mount a lot more amplifier power in a given amount of rack space, but such a design pots much more strain on the rack ears. The rack ears should be well-connected to the chassis in such a way that overstress does not damage or loosen some other critical element, such as the faceplate. Rear support is strongly recommended. Standard-height chassis provide a greater mounting surface and allow for more internal space around components, such as the transformer.

Fan cooling creates noise in the chassis, but it dramatically reduces the size of the heatsink. Because of the increased power ratings of modern amplifiers, it is harder to find convection-cooled units. Class-G and -H amplifiers reduce heatsink requirements (as will class-D designs eventually) and may help bring back convection-cooled amplifiers. Meanwhile, high-quality, variable-speed fans can minimize the cooling noise. You should always check fan noise carefully if people will be sitting near the amplifier.

There are a number of details you should check. For example, external connectors, controls and displays should be recessed to protect them from external damage, and lockout covers are sometimes available for extra security. All connectors should be high-quality. Make sure input and output connectors are firmly mounted and will resist the strain of somebody tripping over the cabling.

Internal connectors also need to be rugged.

Sub-assemblies and circuit boards might shift as the amplifier bounces around, so excessive rigidity can be a problem. Card-edge connectors are frequently troublesome. Cable-type or shock-mounted connectors with some “give” are preferred.

Gold plating is frequently preferred on input connectors, but it might wear through after a few insertions.

The quality of the underlying plating is probably more important.

Internal connectors should also have corrosion-proof plating. Gold plating is best for small signal connections, but it can burn through at high currents. Heavy-duty connectors should use preciousmetal alloys or conductive oxide platings in which current flow actually improves the integrity of the contact.

You’ll find these principles used in QSC amplifiers and in amplifiers from other leading audio companies.

Various engineers have their own opinions on how each problem is best solved, but, in many cases, it’s basically a question of continuously refining a chosen approach. The basic principles here should help users appreciate what goes into modern amplifiers, often the unsung heroes of sound systems.

Courtesy QSC Audio. Used by permission.
Article in Sound & Video Contractor Feb. 20, 1993

David McLain | Loudspeaker Guy! | CCI SOLUTIONS
Be seen. Be heard.
PO Box 481 / 1247 85th Ave SE
Olympia, WA 98507-0481
Voice: 800/426-8664 x255 / Fax: 800/399-8273
Clearance Bin: