Go to…. Introduction Technology Amplifier classes Filter types Dedicated subwoofer amplifiers (built-in/"plate") Amplifier modules Stereo amplifier kits Monoblock amplifiers Electronic crossovers Linkwitz transform equalizers Filter modules Instruments

The technologies:

But first, a word about electronics technology… Someone wrote: "I see there are references to different "Amps". The CHIP, MOSFET and FET. What are the differences between these or are they basically the same thing? How do they compare to tube amps?" Without getting into the essentially religious debate over tube (valve) amps vs. solid state (SS) amps, here's a brief rundown to the players: TUBE/VALVE AMPS   Once upon a time, tubes were all there was. The first vacuum tube was a rectifier (AC to DC converter) invented in 1904 by John Fleming. In 1907, Lee de Forest figured out that by adding a third electrode, he could control the amount of current flow through the device. This was the first triode amplifier. A tube (also called a "valve", after its ability to control current flow) is therefore a transconductance device, using a voltage to control a current flow. From then until 1947, these were the only types of amplifying devices available. Throughout this period, improvements were made, chiefly correcting non-linearities and increasing power levels. These goals were often accomplished with the use of more electrodes, yielding tetrode (4 electrode) and pentode (5 electrode) tubes. The greatest weaknesses of tubes still remained their large size and their need for a cathode heater. The heater is essentially like a incandescent light bulb filament and like a light bulb, has a finite operational life before it burns out. Tube/valve amps use only vacuum tubes. Critics point out that, aside from the lifetime issues, their measured performance is often inferior to other types of amps, especially in even-order harmonic distortion. Tube/valve amp fans ignore the numbers and say that the amps simply sound better to them. If asked to explain it, many don't try. Of those who do, it's often pointed out that tubes often have lower odd-order harmonic distortion (which is more audible than even-order), and that tube/valve amps are typically designed with little or no loop negative feedback, which can greatly improve transient performance. Finally, most tube/valve amps do not sharply "clip" the signal when overdriven. Clipping generates massive amounts of all odd-order harmonic distortion, so the tube/valve amp's less abrupt clipping means that it behave better when over-driven. At that point, critics will again point out that it's also more likely to be overdriven since most tube/valve designs are not high power amplifiers. BIPLAR AMPS   The bipolar junction transistor (BJT) was invented at Bell Labs in 1947 by John Bardeen, Walter Brattain, and William Shockley. By 1956, when it won the Nobel prize in physics for the trio, it had already begun to revolutionize electronics. Requiring no heater and being very small, it overcame the two principle objections to vacuum tubes. Unlike tubes, a transistor works by using a current to control a current - in other words, it's a current amplifier, a fundamental difference from the operation of a tube. This fundamental difference led to a new generation of circuit topologies. One thing which became common was the use of loop negative feedback to control non-linearities. This simply means that a little of the output signal is fed back to the input to correct errors. However, since there is always a time lag between the input and output, loop negative feedback can cause problems with transient response. Most SS amps you see in common use today use BJT's. They're inexpensive and well understood. As time has passed, they've gotten to sound better and better due to both improved devices and improved circuit topologies. Variations exist which support specialty devices. Two of these have been used in prominent audio designs. One is the "ring emitter" transistor, essentially a BJT made by Sanken Electric using their own planar epitaxial structure. The second is the Insulated Gate Bipolar Transistor (IGBT), a hybrid between a BJT and a MOSFET (see below). "CHIP" AMPS   These are amplifiers on a chip - i.e. a power amplifier integrated circuit. They are simply an input, gain, and output power stage all fabricated on a single semiconductor substrate. This provides two significant benefits - it is extremely economical and all the devices are inherently almost perfectly matched. These advantages aren't free, though. The problem is that thermal changes in the output stage also effect the input and gain stages. Good design can overcome these problems to a degree, but they're always there. The most popular chip amps as of this writing seem to be the National Semiconductor LM3886 & LM3875 (50 W, and the most popular), SGS-Thomson TDA2050 (32 W), and Sanyo STK4050 (200 W). A Google search on "gainclone" will yield more on chip amps. FET AMPS   FET amps are amps which use junction field-effect transistors (hence JFET or simply FET). Sometimes only in gains stages, sometimes all the way to the output stage. As noted, BJT's are fundamentally current amplifiers. Also as noted, tubes (valves), are transconductance devices - and so are FET's. The controlling gate electrode of a FET is a semiconductor junction which is always reversed biased in normal operation. As the reverse bias voltage increases current flow is more restricted, so these are known as depletion mode devices because they work by depleting the number of available carriers (electrons and holes). In designing an amplifier, FET's work a lot like triodes. MOSFET AMPS   MOSFET amps also use FET's, but use Metal-Oxide Semiconductor FET's, (hence MOSFET). Thus, the term MOSFET is a structural definition. The gate electrode of a MOSFET is completely insulated by a layer of oxide from the conduction channel. hence another term in less common usage for the devices is Insulated Gate FET, or IGFET - more of a functional than structural definition. Since the gate is insulated, they can be either forward or reverse biased. Typically, it is forward biased in normal operation, with increasing forward bias voltage causing greater current flow. Operationally, this is known as an enhancement mode device since it enhances the number of available carriers (electrons or holes) in the channel. In designing an amplifier, MOSFET's behave a lot like pentodes. Done right, they can all be good. For example, the lowly AKSA kit amplifier has been praised for having a "tube sound", yet uses bipolar transistors. LC Audio also makes highly praised kit amps which rely on special "ring emitter" BJT's made by Sanken for its output devices. OTOH, many other highly praised SS amps use FETS or MOSFETS (e.g. Borbely, Marchand, et al). Finally, of course, tube/valve afficionados won't accept any SS design, but limit their debates to the merits of push-pull vs. single-ended circuit topology, triodes vs. pentodes, and transformer output vs. output transformer-less (OTL). The popular "SET" designation simply implies a Single-Ended Triode design.

Amplifier classes:

There's a great deal of confusion, compounded by outright dogma, with respect to the various classes of amplifiers. While I'll try to skirt the dogma, a brief discussion of amplifier classes is appropriate. In this section, I'll list only the classes in common use for audiophile gear. There are other classes, useful in electronic engineering outside of the audio field, as well as "new", non-standard classes created by imaginative marketing departments. In general, amplifier classifications represent how much current is flowing at any given instant and are therefor related to the efficiency of the amplifier. Class A   In a Class A design, current is always flowing, regardless of whether or not there is a signal present. The power output device(s) handle both the positive and negative halves of the signal. When there is no signal (i.e. the signal level is zero), output devices idle at the halfway point. Class A operation is therefore quite inefficient and wastes a lot of power as heat. The advantage is that output devices are always operating in their linear range, so distortion is minimized, and maximum low-level detail is achieved. Many (most?) amplifiers use Class A designs in their front end and gain stages, even if their output stage runs some other class of amplification. Class B   Class B amplifiers are rarely used for audio and are mentioned primarily to help explain Class AB, below, and because of their importance in DIY amp design. Succinctly, a Class B amplifier uses different devices for the positive and negative halves of the signal. In the absence of a signal, all output devices are turned off. There is a cost, though. When a signal is present, the devices typically exhibit a turn-on delay and voltage dead zone. Low-level detail is therefore lost and distortion is generated each time the waveform crosses the zero level. With careful design, this can be overcome, but the advantage is arguable. Doug Self's "Audio Power Amplifier Design Handbook", which has become something of a "Bible" for DIY amp builders, recommends "optimum class B" for his vision of a perfect amplifier. Many who have built his designs agree that it's a great amp. Class AB   Class AB amps are quite similar to Class B in that they use separate devices and sections for the positive and negative halves of the signal. Where Class AB differs from Class B is that the output devices never turn off at the zero level. Their operating ranges thus overlap around the zero level, causing current to flow in each section. Since one definition of Class A is that current is always flowing, a Class AB amp is said to operate at Class A for low level signals and Class B for high-level signals. Class AB is the most common design approach for SS and high-power amplifiers - which can't afford the inefficiency of Class A. Class AA   Similar to Class AB, Class AA is not generally recognized as a proper amplifier classification, but as a variation on Class AB. In a proper Class AB amplifier, at some signal level, the amp operates in Class B with only one output leg active. In Class AA, the inactive leg never completely shuts off, but stays in Class A, but with a significantly lower idling current. Class D   Class D is a digital class. As such the output signal is always either on or off. Think of it as like Class A, but with super-high amplification. If the input signal is positive, the output is all the way positive. A soon as the input signal crosses the zero level, the output goes all the way negative. Rather than a clean waveform, the output is a rectangular wave. What good is it then? Consider now what happens if we add the input signal with a modulating ultrasonic triangular wave… During each cycle of the modulation signal, the output will make two zero crossings. If the input is a constant zero level, the output will be a square wave which is positive 50% of the time and negative 50% of the time. Similarly, if the signal is constant at the half-positive level, the output will be a rectangular wave which is positive 75% of the time and negative 25% of the time. If the modulation frequency is high enough (typically around 200 kHz), a simple analog filter will convert the output into a analog of the input. In engineering terms, this is called Pulse Width Modulation (PWM) and has been in common use in regulated power supplies for decades. This technology is most often used in autosound applications, and there's a good visual tutorial on the web. For home hi-fi, Class D amps are common in subwoofer "plate" amps. Class G   Class G is a variation on Class AB, where the power supply voltages to the output stage are switched to progressively higher values as more power is required. Originally seen in the audio field in power amps designed by Bob Carver, its use has become more common, especially in pro gear. The obvious advantage is greater efficiency than Class AB. Class H   Class H is a further variation on Class AB. Unlike Class G, where the power supply voltages to the output stage are switched among progressively higher discrete values, Class H actively modulate the power supply voltage(s) in response to the amp's power requirements. As expected, this further increases efficiency over Class G. Class T   Class T is a proprietary development of Tripath Corporation. Briefly, Class T is a variation on Class D which uses digital signal processing (DSP) technology to vary the modulating signal in response to changes in the input signal. In doing so, it accomodates any non-linearities and switching characteristics of the output devices. Class T is mentioned here since it was developed specifically as a high-end audio technology. Again, done correctly, there are no clear-cut distinctions which make one class of amplifier "better" than another. Many have noted that the difference between amps of similar specifications and measurements, operated within their designed power range, is often inaudible, regardless of the technology. Where differences do exist, they are more often differences in power supplies and/or differences in overload characteristics.

Filter types:

In dealing with crossovers, you will hear a lot about various filter types. The job of a filter is to pass a certain range of frequencies and block another. There are three fundamental types of filters:        Low pass Just as its name implies, a low pass filter allows only frequencies below a specified design point to pass, blocking all frequencies above that point. High pass Just as its name implies, a high pass filter allows only frequencies above a specified design point to pass, blocking all frequencies below that point. Band pass A band pass filter allows only a range of frequencies above and below the specified design points to pass, blocking all frequencies outside that range. The range of frequencies passed by a filter is called its passband.The range of frequencies rejected by a filter is called its stopband. Filters are only a specific kind of tuned electrical circuit. In dealing with filters, three characteristics define how it will operate:        Passband ripple Ideally, the frequency response within the passband will be flat. However, many types of filters exhibit ripples, or ringing, in the passband. The passband ripple is usually specified as the amplitude of the ripples in dB, peak-peak. Stopband ripple Ideally, all frequencies within the stopband will be suppressed. In the real world, the output only approaches zero asymptotically. Also, as with the passband, the stopband may exhibit ripples, where the output actually nulls at one or more specific frequencies. Slope The slope of a filter defines how rapidly it makes the transition from passband to stopband. Ideally, the drop-off would be infinitely steep, but this is achievable in the real world. Typically, filter slopes exist at multiples of 3 dB/octave. This makes sense since an octave represents a doubling or halving of frequency, while a 2:1 factor represents a change of 3 dB. Both as shorthand, and for engineering reasons, you will often see filters described as Nth-order. To make the conversion from order to dB/octave, simply multiply by 6. For example, a 1st-order filter rolls off at 6 dB/octave, a 4th-order filter rolls off at 24 dB/octave, etc. There are a number of standard filter topologies which exhibit useful and repeatable characteristics. You will see these used over and over in various designs. The most useful for loudspeaker design are: Filter class Frequency Response Square Wave Response Comments Butterworth Often referred to as "maximally flat" response, a Butterworth filter exhibits the widest and flattest possible passband. Rolloff is the steepest of any filter without significant passband ripple. Impulse response exhibits minimal, critically-damped overshoot. Any tuned system with Q=0.707 will exhibit Butterworth response. The popular Linkwitz-Riley crossover topology is built using Butterworth sections. Bessel Often referred to as "minimum phase" response, a Bessel filter exhibits the smoothest possible response at the cost of slower transitions between the passband and stopband. A Bessel filter exhibits absolutely no impulse overshoot. Any system with Q=0.5 will exhibit Bessel response. Also note that although a Butterworth filter will always exhibit a more extended F3 than a Bessel filter, the Bessel filter will exhibit a more extended F10. Cauer (elliptical) Called an "infinite slope" filter by Joseph Audio who uses them extensively, a Cauer, or elliptical, filter exhibits the steepest possible slopes, but at the cost of significant ripple in both the passband and stopband. Among the DIY community, Jason Cuadra has done some significant work on Cauer filters. His designs are available here. There are an infinite number of other possible filter topologies. Since a filter is only a kind of tuned electrical circuit, many of the same characteristics are used to describe loudspeaker enclosures and other types of non-electrical tuned systems. One popular in designing bass speaker alignments uses a Q of 0.577, midway between Butterworth and Bessel alignments. This is called a "minimum delay" alignment since it results in the lowest group delay characteristic of any filter topology. Many DIY-er's feel that a system Q around 0.6 yields a "tighter" feel to the bass and greater perceived extension than Butterworth due to a combination of the alignment's extended F10 plus room gain. Another type of filter not shown above is the Chebyshev filter which has a Q≥0.707. It exhibits ripple in the passband, but unlike Cauer filters, not in the stopband. A Chebyshev low pass filter is often used for subwoofers - designed with only 1-3 dB of passband ripple, it provides both low pass filtering plus some bass boost.

©1998-2007 by Bob Stout, all rights reserved

Tested with: .

Support: