5 Performance

The overall quality of sound from a system depends on the “weakest link.” This includes everything from the quality of the drivers to the sample rate of a digital-to-analog converter (DAC). In evaluating an amplifier’s performance, one of the first major specifications to understand is THD+N. This references how much harmonic distortion, or overtones, as well as noise is added to the audio signal after amplification. In other words, this can measure how “true” the amplified signal is to the original, and the lower the THD+N, the better. Since THD+N varies with output power level and frequency, a common way to cite the THD+N of a device without a graph is to use a specific output power level and load at 1kHz. For example, 0.02% THD+N at 1 W into 4 Ω using a 1 kHz signal. See an example from the TPA3251 data sheet in Table 5-1.

However, the THD+N of a device is best conveyed through a graph. See Figure 5-1 for an example of a THD+N curve.

Figure 5-1 THD+N vs Output Power in AD Mode for TPA3221

The next important specification to consider is the signal bandwidth and frequency response of an amplifier. The bandwidth refers to the range of frequencies it can support. Standard Class-D amplifiers support 20 Hz to 20 kHz; however, high-performance amplifiers can support much higher for content from high definition (HD) sources. As ultrasound frequencies (greater than 20 kHz and inaudible to humans) are adapted into smartphones and smart speakers for hand gesture recognition and presence detection, higher bandwidth is becoming increasingly important. While the bandwidth refers to the range of frequencies supported, the frequency response indicates how well the amplifier performs throughout its bandwidth. This can be shown in various ways, such as frequency against the sound-pressure level (SPL) of an end system or frequency against THD+N. Figure 5-2 shows the frequency response of TAS5805M in PBTL using 12 V into 4 Ω.

Figure 5-2 THD+N vs. Frequency – TAS5805M in PBTL Configuration

The bandwidth should not be confused with the sampling rate or the modulation frequency, which is also measured in kHz. If an amplifier references a sampling rate, this is because it includes a DAC and sometimes a digital signal processor (DSP). While an analog signal is continuous, a digital signal is made up of thousands of samples of the signal per second. The resolution of samples per second is what’s referred to as the sampling rate or sampling frequency. In a digital amplifier, the bandwidth is limited by the sampling frequency. According to Nyquist’s theorem, also referred to as sampling theorem, the sampling frequency must be at least twice the desired bandwidth. Therefore, a system intending to play 40 kHz ultrasonic tones would need a sampling rate greater than or equal to 80 kHz, making an amplifier with 96 kHz sampling rate suitable.

The modulation frequency or switching frequency refers to how fast the pulse-width modulator of the Class-D amplifier switches, modulating the analog signal. Technically, this could affect the resolution of the audio signal, but for typical audio applications it’s always well above the sampling rate. The switching frequency is usually between 200 kHz to 1.5 M, or even as high as 2.1 MHz as seen in some industry leading automotive Class-D amplifiers, like TPA6304-Q1. While the switching frequency doesn’t usually impact the signal bandwidth, it does still play a role in electromagnetic interference (EMI) since the pulses emit high frequency energy. Modulation schemes and EMI will be addressed in more detail in a later section.

A few more common performance parameters include signal-to-noise ratio (SNR), dynamic range (DNR), and power supply rejection ratio (PSSR), all of which are measured in dB. The SNR conveys information similar to the THD+N: how true the reproduced amplified signal is to the intended audio signal. While the THD+N also reflects distortion, the SNR strictly measures the ratio between the average signal level and the average noise level, so the higher the better. The dynamic range is then the ratio between the lowest possible level of noise (noise floor) and the loudest possible undistorted signal, effectively the best possible SNR of the system in ideal circumstances. The power supply rejection ratio also speaks to noise, but in a different way. It refers to the ratio between how much a change at the input results in a change at the output. Basically, it quantifies how much the amplifier “rejects” noise at the input. If the ratio is really low, the amplifier is taking the noise and amplifying it to the output. If it is high, less of the noise is amplified to the output. Thus, a higher number here is once again better.