Switched-On Amps: Power with a Pulse (Feb. 1995)

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting

Departments | Features | ADs | Equipment | Music/Recordings | History

by Bascom H. King

Switching amps have captured the imagination of many audio pioneers over the past 20 to 30 years. The siren song of this technology has been-and still is-high efficiency, small size, and the potential for more linear operation (or at least a different type of linearity) than that of conventional solid-state amps. Internally, switching amps handle signals entirely as on/off pulses, using pulse-duty-cycle modulation (PDM), or pulse-width modulation (PWM). Therefore, these amps are often referred to as "digital"; they are also some times described as Class-D amps. Their downside is the considerable difficulty in keeping r.f. generated by the switching action from radiating out of the amplifier and interfering with one's FM tuner, TV set, and possibly other electronic equipment.



How do switching amplifiers work? Fundamentally, they consist of three main blocks: An analog input-to-duty-cycle modulator, a switching output stage, and a power supply (Fig. 1). Assuming a bipolar (positive and negative) power supply, the output stage can be in only one of two states: Fully positive or fully negative.

Imagine that the output is switching between these two states at a rate of 500 kHz. With no incoming audio input, the waveform will be a 500-kHz square wave with equal positive and negative time intervals. As shown in Fig. 2, this produces a 50/50 duty cycle: The waveform spends half of its time at the positive rail and the other half at the negative rail. Under these conditions, the net average (or d.c.) output is zero. Now, suppose that the duty cycle were to be proportional to the incoming audio signal's amplitude (a proper supposition, as that is the function of the modulator block pre ceding the out put stage). As the incoming audio wave form goes positive, the duty cycle changes so that the time spent positive is correspondingly greater and the time spent negative is correspondingly less, which produces a net positive output voltage. When the incoming audio waveform goes negative, a similar effect takes place, producing a net negative output voltage.

Of course, with real audio signals, all this takes place on an ongoing basis, making the average output of the switching output stage an amplified replica of the incoming analog audio waveform. It is necessary to remove the 500-kHz switching components, so they will not get into the actual output to the outside world; this is the job of a high-level, LCR low-pass filter. The cutoff frequency of these filters is typically around 50 kHz.

Another way of thinking of all this is as a sampled data system with a 500-kHz sampling frequency. This is a simplified explanation, of course. In actuality, most designs allow the switching frequency to come down for the signal peaks of each polarity as the signal modulation approaches full scale, so as to more effectively allow the duty cycle to approach 0% and 100%. Furthermore, audio-frequency negative feedback is inherent in the operation of most designs, helping to maintain overall input/out put linearity.

Historically, I believe Infinity Systems had the first actual switching amp on the market, back in 1976. This unit, informally called the Swamp (for switching amp), had some very impressive characteristics-among them some 300 watts per channel into 8-ohm loads, a switching frequency of 500 kHz, use of a switching regulated power supply, and the use of one pair of fast bipolar transistors per channel in the switching output stage. The designers of this circuit originally thought that it could be small and light, with virtually no heat-sinking. Over time, they learned this really wasn't so, and the final product ended up with a very considerable bulk of extruded aluminum heat-sinks on the rear of the unit. The nemesis of Infinity's Swamp was a propensity for failure that was not consistent from unit to unit. Some units would be quite robust, while others would fail at the blink of an eye. This characteristic persisted de spite considerable and ongoing efforts by the designers (I was later one of them) to improve reliability.


Infinity came up with a new design, nicknamed Swamp 2, whose audio circuitry was similar to the original Swamp's but which had a non-switching power supply. Only a few, if any, of these units were sold.

Infinity shortly got out of the switching-amp business.

Around the time Infinity was working on the Swamp 2 (1977 to 1978), Sony came out with a twitching amp, the TA-N88, which used two pair of V-FET devices for the switching output stage. This de sign also used a switching, regulated power supply. It was a nicely made unit and sounded pretty good, but it was on the market for only a year or two.

In car audio, there have been a number of switching power amplifiers. Around the early to mid-1980s, a deluxe sound system for the more expensive GM cars appeared. These systems, designed and built by Bose, utilized a switching design for the power amplifiers.

The amplifiers didn't use a power converter to generate a higher bipolar supply for the output stage; instead, they used a full bridge output stage operated directly off of the battery volt age. This, in conjunction with nominal 2-ohm speakers, produced some 25 watts of power per speaker position. (Today's Bose car amps feed 0.5-ohm loads, to deliver 100 watts apiece.) I have heard that Alpine has also used a similar approach in some of the name-brand car systems that they have been involved with. Yamaha also had a switching amp for cars on the market for a while, in the late '80s.

In the sound-reinforcement field, I understand that a few switching amplifier designs have come and gone over the years. Currently, Peavey and others have switching amplifiers intended for this market.

Infinity Systems has recently gotten back into the switching-amp business, this time with aftermarket amplifiers for car systems. These products have been designed by my good friend and mentor, W. M. (Mack) Turner of Arnoux Corp. I have known about the development of these amplifiers from before Mr. Turner's relationship with Infinity was established. He calls his design "past stable," which really defines a topology whereby the circuit, from a linear systems feedback point of view, is unstable at high frequencies and thus oscillates. Oscillate it does, at about 500 kHz, producing an overall pulse-duty-cycle, frequency-modulation scheme that produces a highly linear overall input-to-output relationship. Infinity is currently selling these car amplifiers in various power levels and configurations.

As a separate effort, Arnoux is readying two models for home hi-fi use, under its own name: The Model 7B, a 60-watt/channel stereo unit, and the MB-300A, a 300-watt mono piece. I have had a lot of listening experience with various versions of the smaller stereo amp, and I must say they sound very natural and satisfactory. One thing I like about the 7B is that it consumes so little operating power, some 7 to 8 watts; I happily leave it on all the time so that it always sounds its best when I use it in my system. The larger MB-300As have an obviously more powerful sound and rival some of the best linear designs in overall reality of sound reproduction. Look for these amps at selected dealers soon (and possibly direct from the factory in Santa Barbara, California).

I wouldn't be surprised to see more activity in switching amps in the near future. Their newfound reliability makes them practical, and their high efficiency has strong appeal in this energy-conscious era.




(adapted from Audio magazine, Feb. 1995)

= = = =

Also see:

Build a Class-A 100 Watt Mono Amp--Part 1 (Jan. 1995)


Prev. | Next

Top of Page    Home

Updated: Saturday, 2019-07-13 12:18 PST