Image
image
image
image


Design Articles

Reducing EMI in Class D Audio Applications by Spread Spectrum Modulation Techniques

Class D audio amplifiers are a great solution for portable devices—except they’re noisy. This article shows how to design around that problem.

By Bill McCulley, Staff Applications Engineer, National Semiconductor Corp.

audio amplifier

The use of Class D audio amplifiers has become increasingly widespread in portable applications. As complexity, size and audio performance have improved, the Class D topology continues to gain market share. So compelling a benefit is efficiency that Class D amplifiers are now used extensively in portable designs worldwide, as battery life and small profile become key differentiators for end-users. The efficiency advantage is more pronounced when one realizes the typical audio application will normally operate at less than one-third of the total output power specified. Within that range, Class D amps operate at 80% or better efficiency, while Class AB less than 30%. With performance and cost improvements, the main challenge for Class D amplifiers remains the concern over radiated emissions, or electro-magnetic interference (EMI). EMI, or radio-frequency (RF) interference, is a growing concern for designers of portable applications. However, recent advances in Class D technology, primarily the use of spread spectrum modulators, have led to the introduction of truly filterless switching amplifiers with reduced RF emissions.

To better understand the source of EMI, a brief overview of Class D topology is helpful.  Class D amplifiers modulate an audio signal with a reference triangle or sawtooth waveform and produce an amplified signal, typically in the form of a pulse-width modulated (PWM) switching output.  While the modulation may vary, all Class D amplifiers continuously switch the output from rail-to-rail, at a frequency determined by the modulation frequency -- generally far above the audio range (commonly recognized as 20Hz-20kHz).  The duty cycle of the carrier square wave is controlled to represent an average which is proportional to the instantaneous value of the input signal. Typically the switching frequencies are more than 10 times the highest frequency of interest in the input signal.  In most Class D amps, a feedback path is also used along with an error signal to improve the total harmonic distortion and noise (THD+N), power supply rejection ratio (PSRR) and other performance characteristics.  Excellent references on Class D amplifier theory include books by Douglas Self [1], articles by Goldberg and Sandler [2], and semiconductor manufacturer application notes.

Practical limitations with Class D amplifiers are quickly evident.  High-frequency energy is present at the switching frequency and its harmonics along with the spectral components of the square wave.  Until recently, Class D amplifiers required a low-pass filter (generally a 2-pole Butterworth LC filter) to remove the high-current, high-frequency square waves, leaving only the audio signal. In newer Class D amplifiers a filterless approach uses the loudspeaker itself as an element of the low-pass filter. These newer “filterless” class D amplifiers have become very popular in portable designs. Unfortunately, this approach may end up producing more EMI than a traditionally filtered Class D would allow. 

class D amplifier
Figure 0: A basic switching or PWM (class-D) amplifier

It has been noted regarding Class D amplifiers that “great effort and ingenuity have been devoted to this approach, for the efficiency is in theory very high, but the practical difficulties are severe, especially so in a world of tightening EMC (electro-magnetic capability) legislation, where it is not clear that a 200 kHz high-power square wave is a good place to start.” [3]   

Portable design trends exacerbate the EMI problem.  As products become smaller, the components, traces and wires are all in close proximity, and proper printed circuit board layout becomes extremely difficult.  Due to space constraints, the use of filters has become almost out of the question. The trend towards louder devices drives the need for higher power and current, creating higher emissions.  Also, the merging of multiple wireless capabilities – Bluetooth wireless technology, Wi-Fi, wireless area network and others – into a single platform create a maze of opportunities for EMI to become a problem.  While EMI is an immediate concern internally within a product, also of concern are the RF emissions interfering with other external systems.  Most consumer systems need to pass some form of FCC requirements, which deal with unintended emissions from a product interfering with other devices that use the radio spectrum. 

There are numerous ways to deal with EMI.  One technique to reduce EMI within the amplifier is to slow down or soften the edges of the square wave, but the tradeoff is an increase in THD+N due to the decreased ability to accurately sample the incoming analog audio signal, as well as decreasing efficiency. The use of LC (inductor plus capacitor) filters can greatly reduce EMI, but LC filters are large and expensive, with size and cost increasing with output power levels.  PCB traces and wires essentially act as antennas with significant radiation occurring once the trace length reaches one quarter the wavelength of the signal they carry – so a general rule is to keep traces as short as possible.  Other options include routing PCB traces that carry high frequency signals between ground planes, and using insulated components and toroid inductors.  For filterless Class D systems, the trace and cable length connecting the amplifier’s output to the speakers will likely be the largest source of RF emissions.  Traditional practices such as placing ferrite beads in series with the loudspeakers close to the amplifier can be effective.  Ferrite beads act as an RF choke, attenuating high frequency signal components.  However, ferrite beads are effective over a narrow frequency range, and may not provide enough attenuation over the output noise bandwidth.  Shielding can also be used if PCB layout and filtering cannot reduce the EMI to an acceptable level.   The power supply is another possible EMI source. A Class D amplifier draws current in large, short duration pulses related to the output switching edges that appear on the power supply lines.  Power supply related EMI can be minimized with proper layout and bypassing techniques.

Although reduction techniques done “after the fact” can be effective, it is best to start with an amplifier that generates less interference.  Compared to previous Class D topologies, a spread spectrum device, offers such an opportunity.  The spread spectrum techniques is not a recent development, dating back over a half a century, with some of the earliest efforts occurring in communications systems and military radar applications.4 In the past decade, spread spectrum modulation techniques have become popular in other applications – particularly in clocking circuits.  The benefits the spread spectrum technique has brought to those applications can also be seen when applied to Class D amplifiers.

figure 1
Fig 1: Noise and noise floor - before and after spread

A spread spectrum modulator adjusts the switching frequency of the output bridge in a band around a center switching frequency (for example, a 300 kHz center frequency with a +/- 30% frequency spread). As long as the frequency variation remains random, the actual method for varying the frequency can range from a simple sweep to uncorrelated jumps in carrier frequency.  The spread spectrum modulation scheme has some key advantages: high efficiency and low THD+N are maintained, while the radiated noise and EMI is reduced.  It is important to note the total amount of energy is not decreased.  Peak energy decreases, however, the total energy remains the same, and is spread over a wide frequency band, as shown in Figure 1.  Noise exists over a wider bandwidth but the peak noise at any one frequency will be less than what is generated by a fixed frequency device.

figure 2
Fig 2: Noise energy – FFT

By varying the frequency of the switching waveform randomly over a spectrum range, the wide band spectral components are flattened. Figure 2 illustrates the impact on noise energy, as seen through the use of Fast Fourier Transform (FFT).  On the left, the FFT of a fixed frequency amplifier shows the higher peak energy concentrated at the harmonics.  On the right, the FFT of a spread spectrum modulated amplifier shows overall lower peak energy and decreasing harmonics, resulting in a higher noise floor.

figure 3
Fig 3:  Class D Audio Amp with Spread Spectrum Modulation (LM4675)
The benefits of spread spectrum modulation techniques are primarily in two areas: improved EMI performance due to lower radiated noise peaks, and the reduction or elimination of the EMI filters commonly seen in Class D applications.  For example, consider the performance of a recently introduced spread spectrum Class D audio amplifier, shown in Figure 3.

FCC and CE (European compliance engineering standard) Class D radiated emission standards are shown below and apply to any digital consumer device that is not intended to transmit.  All consumer electronic products must be certified before sale in the United States and Europe.


Frequency Range (MHz)

FCC Class B Limit (μV)

0.45-1.705

48

1.705

48

Frequency Range (MHz)

CE Class B Limit (dBμV)

0.15-0.50

56

0.50-5

56

5-30

60

Initial EMI tests conducted on the part with a 2-inch speaker cable and no filter components, demonstrated excellent EMI performance during FCC class B testing, as shown in Figure 4.  The red line is the FCC Class B limit.  The noise spectrum must remain below this line in order to meet FCC emissions requirements.

figure 4
Fig 4: Radiated Emissions 30-1000MHz, 2” Speaker Cable (Horizontal
Polarization) with Spread Spectrum Modulation

Spread spectrum modulation techniques offer significant benefits to Class D audio amplifier applications. The reduction of RF emissions and the simplification of costly EMI-reduction strategies such as LC filters greatly reduce the handicaps that traditional Class D topologies once faced in the portable design space.  Relevant applications that can benefit include any portable devices subject to FCC/EC regulations or other EMI-related regulations such as Mil-Std-461.  In addition, any portable devices that require reduced system noise such as communication devices, music players, radios, and speakerphones will likely benefit from spread spectrum techniques.

EMI is a system-level concern.  The smart system designer will use all tools at his disposal to create a well-performing product, starting with the building blocks of his design, the components. The use of devices featuring spread-spectrum modulation can play an important part in reducing the EMI signature of a portable system design.

References:

  1. Douglas Self, “Audio Power Amplifier Design Handbook”, 4d Ed, 2006
  2. Goldberg and Sandler, “Noise Shaping and Pulse-Width Modulation for All-Digital Audio Power Amplifier” Journal of Audio Engineering Society, Feb 1991, p. 449
  3. Douglas Self, “Audio Power Amplifier Design Handbook”, 3d Ed 2002, p.35.
  4. Robert A. Scholtz, “The Origins of Spread Spectrum Communications”, IEEE Transactions on Communications, Vol Com-30, May 1982, pp 822-854.
This article originally appeared in the January, 2007 issue of Portable Design. Reprinted with permission.

 Share
Digg Reddit Del.icio.us Stumble Upon Facebook Twitter Google BlinkList Technorati Mixx Windows Live Bookmark MySpace Yahoo Bookmarks Diigo


Insert your comment

Author Name(required):

Author Web Site:

Author email address(required):

Comment(required):

Please Introduce Secure Code:


image
image