Class D audio power amplifiers: Adding punch to your sound design
One of my first jobs as an engineer was at Empire Scientific Corp. in Garden City, NY in 1972. This was a high-end audio company that made speakers, turntables and phonograph cartridges. We had some direct-cut vinyl records with such incredible dynamic range that you could close your eyes and visualize the orchestra and the exact location of each instrument. When played through two Crown DC-300 amplifiers (Left and right channel) that, believe it or not, we used as Pre-amps into a Phase Linear Power Amplifier (First we had the 700B series and later the Series II).
Audio amplifiers have come a long way since then. Class D amplifiers were first conceived in 1958 and so much has been written about the different architectures that were cited just in 2016 which improve different aspects of their performance. I will share these recent techniques with you in this article.
Such recent design improvement techniques are Power Supply Rejection Ratio (PSRR) improvement, lower distortion, Electro-magnetic Interference (EMI) reduction, Intermodulation (IM) distortion improvement, quiescent current reduction, Total Harmonic Distortion (THD) reduction, and driving capacitive transducers in electrostatic loudspeakers.
The power transistor element
Audiophiles always said that the best sound came from vacuum tubes. There are schools of thought that say FETs are more closely related to the type of distortion and sound that vacuum tubes produce in audio than BJTs. An MIT report claims the opposite. The controversy continues as it has since the vacuum tube began to lose importance in most electronics designs.
If we look at the power element as a FET, such as Efficient Power Conversion (EPC) outlines in an app note1, we see that good sound needs good THD, Damping Factor (DF), and IM Distortion. Efficient Power Conversion’s eGaN FET designs in class-D audio amplifiers provide excellent sound quality due to their excellent switching abilities because of lower open loop distortion that helps lower THD and overall losses. This leads to a reduction in feedback which in turn reduces Total-IM Distortion and DF and improves sound quality. All of this over a wide range of output power which essentially enables better dynamic range performance in the audio realm.
Class D power amplifiers lower, power dissipation, footprint, and cost2
Since all linear power output stages will dissipate power in class A, B, AB, and D amplifier configurations, we look for the one architecture that will enable the most efficient power so that battery operation can be employed, with the smallest board footprint to reduce size and weight in a portable or wearable design, and the lowest cost to enable expansion into a wider variety of market applications than ever before. The class D configuration is the one that stands out as meeting all these requirements (Figures 1 and 2).
Figure 1: A linear CMOS output stage diagram (Courtesy of Analog Devices2)
Figure 2: An open-loop Class D amplifier basic diagram including a nearly-lossless filter design architecture in which the only intentionally dissipative element is the speaker (Courtesy of Analog Devices2)
So how does a class D amplifier work?
International Rectifier (now part of Infineon) explained this topology very well in a 2005 application note.
The class D amplifier is simply a switching amplifier or Pulse-Width Modulation (PWM) amplifier as opposed to class A, B and AB which are all linear amplifier architectures. Figure 3 shows a basic class D amplifier block diagram.
Figure 3: A Class D Amplifier detailed block diagram3 (Image courtesy of Infineon)
The class D amplifier essentially operates similar to a PWM power supply; their topologies are essentially identical in operation (Figure 5). Figure 4 shows the class D amplifier waveforms in action with its waveforms shown.
Figure 4: Waveforms shown here in a class D amplifier operation3 (Image courtesy of Infineon)
Figure 5: The strong similarity of the Buck power converter and the class D amplifier3 (Image courtesy of Infineon)
Eliminating the external low-pass filter while reducing EMI4
Maxim Integrated has employed filterless modulation in its class D amplifier designs (Figure 6).
Figure 6: The MAX9000 filterless Class D modulator topography (Image courtesy of Maxim Integrated)
There is almost always a compromise to be made in circuitry architectures when improving a particular parameter. In this case, the added possibility of radiated EMI from the cables emanating from the amplifier to the speaker. Enter Spread-Spectrum modulation to the rescue. By using a dithering or randomizing method for the switching frequency in the class D amplifier, this technique spreads out the spectral energy of the output signal which in turn reduces any high-frequency energy peaks at the output4.