Quantitative Analysis of Click and Pop in Audio Amplifiers

introduction

The special requirements of portable audio devices are key to product design. Why is product A better than its competitor, product B, and more desirable to use? Performance-wise, the frequency response flatness and THD+N between competing products are similar, making it difficult to tell which product performs better. The user interface can provide the main differences between the products, but this is largely subjective. Objective audio performance metrics can be used to compare products and explain why one product is clearly better than the others.

An important metric for evaluating audio performance is the "click" or other strange transient noise that occurs in the earphones (or speakers) when the device is turned on or off. As people's expectations for product performance increase, the absence of transient noise has become an important criterion for people to choose products, and therefore a key selling point for portable audio devices. Until now, the industry has still evaluated this click and pop on a subjective basis, and descriptions such as "low click" and "clickless operation" represent subjective judgments on the quantitative analysis of click and pop. However, user expectations are changing, and designers need objective metrics to evaluate click and pop.

This article describes a method for quantitatively expressing click and pop parameters that can be repeatedly compared across different products.

Characteristics of the Clicking Sound

Click-and-pop refers to audible transients that occur in headphones or speakers when an amplifier drives a transducer to turn on or off. In portable applications, where reducing power consumption is key to extending battery life, certain functional blocks are often disabled when they are not needed. This functionality can further accentuate the problem of click-and-pop. While an ideal component should not produce any audible output when the device is turned on or off, in practice, all audio amplifiers will produce click-and-pop. Depending on the sensitivity of the transducer used (speaker or headphone), the distance between the transducer and the ear, the amplifier's ability to handle transients, and the sensitivity of the hearing sense, click-and-pop may or may not be audible. Although there are many factors involved in determining the audio threshold, amplifier output specifications (independent of the audio transfer function) can be used to quantitatively compare product performance.

Table 1 lists factors that can cause amplifier signal transients.

Table 1. Factors that Contribute to Amplifier Transient Noise

1.	Powered up (power applied)	Category A
2.	Powered down (power removed)
3.	Brought out of shutdown (power applied previously)	Category B
4.	Forced into shutdown (power still applied)

Maxim divides audio tests into two categories to properly measure KCP measurements. Referring to Table 1 above, Items 1 (power-up) and 2 (power-down) are in Category A. It is generally assumed that Maxim products with shutdown (SHDN) have transient modes controlled by the shutdown pin (or register bit) when powered up under normal operating conditions. Category A does not represent normal use and is only relevant when measuring devices that cannot be shut down by software control. Items 3 and 4 (Class B measurements) are closer to normal use.

Figures 1 and 2 show (in the time domain) the transients of two different headphone amplifiers coming out of shutdown. The first is an AC-coupled headphone amplifier compared to the second DC-coupled headphone amplifier. The AC-coupled headphone amplifier has a larger transient when coming out of shutdown (Figure 1). This transient produces a noticeable low-frequency sound due to its slower turn-on process. (Note that the time scale is 100ms/div.)


Figure 1. Data shows the transient process of a well-performing AC-coupled headphone amplifier coming out of shutdown. The amplitude is large, and although this transient will produce a noticeable bass signal, the human ear is not sensitive to this sound.

The second transient, that of a DC-coupled headphone amplifier (Figure 2), appears to be buried in the oscilloscope's noise floor before the A-weighted filter. For this amplifier, most of the audio comes from the DC offset voltage generated when it goes from shutdown to full operation. Since the offset is only a few millivolts, the magnitude of the click and pop cannot be accurately determined with the unfiltered signal. After A-weighting, the click and pop generated by the DC-coupled headphone amplifier offset is extracted from the noise floor, allowing for a more objective measurement. (Note that the V/div scale of the filtered signal is not shown.)


Figure 2. Data showing the transient of a low-offset, DC-coupled headphone amplifier coming out of shutdown. Compared to Figure 1A, the amplitude is much lower (and therefore, subjectively much less noisy), and the amplifier fully turns on after 150µs.

There are two things to consider when analyzing this problem. First, how can the transient be measured objectively? Second, what criteria, if any, are used to measure the test results?

Click Test Method

Maxim uses Audio Precision's System 1 and System 2 (recommended) audio analyzers to measure click-and-pop (Figure 3), but similar test equipment from other manufacturers can be used. The recommended metric, KCP, is an objective measure of audio amplifier click-and-pop.


Figure 3. Headphone amplifier click-and-pop test setup. Note that the left and right channel input pins are AC-coupled to ground. The output load is a typical headphone impedance, and the shutdown pin is triggered with a square-wave generator.

To begin the measurement, connect the device under test (DUT) output to a load or simulated load (dummy load). Load the DUT with the required SHDN and power supplies, and AC-couple all DUT inputs to ground. No input signal is required; input stimulus consists of control signals that switch the DUT between various operating or inoperative modes. Connect the DUT output to the analog analysis section of an audio analyzer.

Next, select the analyzer's A-weighted filtering (recommended) or unweighted 22Hz to 22kHz filter to limit the measurement bandwidth to the audio frequency range. Note that the fast, high-level transients on the oscilloscope do not indicate how much energy is present in the audio frequency band. The human ear has a limited frequency response to transients from speakers or headphones. Therefore, adding A-weighted filtering (Figure 4) is more useful for analysis because it enhances the frequency components that the human ear is sensitive to. Some audio analyzers do not have the option of A-weighting, in which case the bandwidth of the human ear's frequency response should be limited. A common bandwidth limit in audio test equipment is 22Hz to 22kHz, and the bandwidth-limited filter achieves a flat response at about 20kHz (usually the upper limit of the human ear).


Figure 4. Frequency response of an A-weighted filter. The frequency equalization is close to the sensitive range of the ear, so this parameter is often used for noise measurements. Note that the filter transfer function is unity gain (0dB) @ 1kHz, and the signals at both ends are attenuated.

Set the detector to peak reading (rather than RMS value) and set the detector sampling to 32 times per second. For the transients we want to acquire, RMS detection is not useful. The System 2 analyzer supports higher sampling rates, but 32 samples per second can obtain equivalent measurement options from the System 1 audio analyzer. (32 samples per second is the fastest acquisition setting in the System 1 model.) Disable the audio analyzer's auto-range adjustment circuit and manually select the setting that accurately tracks the expected peak signal amplitude. The range of the System 1 and System 2 analyzers is 1x to 1024x (0 to 60.21dB) with a step size of 4x (12.04dB). For accurate measurements, the recommended starting point for audio amplifier click-and-pop measurements is a 1X/Y range.

Drive the SHDN pin with a low-frequency square wave to allow for repetitive measurements. The SHDN cycle frequency is below the audio band and the period should be long enough to ensure that all turn-on and turn-off events are captured (some models have a longer turn-on delay). Maxim typically chooses a 0.5Hz period. The

analyzer's histogram option makes it easy to monitor DUT transients when transients occur between operation and shutdown. The peak voltage can be easily determined, and the histogram can be quickly reset during measurements. The peak voltage is recorded in dBV (dB relative to 1V). This specification is KCP.

The Importance of Test Equipment

The test method described above allows for comparison of similar devices and produces repeatable, objective results. It is best if the test equipment is able to maintain a linear response to any input. For example, the peak reading when testing a 1mV impulse response should be 40dB lower than the peak reading when testing a 100mV impulse response of the same pulse width. (See the Appendix for test transient calibration).

An oscilloscope with external filtering can be used for this click-and-pop measurement scheme. However, experience has shown that the typical click-and-pop levels of high-quality headphone amplifiers are in the millivolt range, which is difficult for most oscilloscopes to measure accurately. An oscilloscope can be used to test higher voltage devices such as high-power amplifiers.

Average value of repeated tests

Different parts of the same model may produce different test results. Therefore, multiple parts should be tested to equalize such differences before judging the performance of a particular part. For a properly designed DC-coupled headphone amplifier, most clicks and pops are proportional to the input offset voltage, which will vary from part to part unless it is equalized (or otherwise eliminated). When fully testing a particular part, measure transients in each operating mode multiple times to ensure consistent results. Then, calculate the average. Multiple measurements are recommended if the part is to be put into use. Test all channels of a stereo or multichannel product.

Establishing Absolute Voltage Levels

The absolute voltage level of the click-and-pop should be specified based on the actual application of the amplifier. For example, assume a device generates a -50dBV transient when it turns off. If the DUT is a 50W/8Ω power amplifier, the full scale is +29dBV. Thus, the ratio of the perceptible click-and-pop to the maximum peak voltage for this amplifier is:

-(+29 - (-50)) = -79dB

However, if the DUT is a 20mW/16Ω headphone amplifier, the full scale is approximately -1.9dBV, which will be less relative to the peak voltage ratio: -48.1dB.

Setting indicator levels

Although we have shown how to obtain an objective measurement of click-and-pop, one question remains: how accurately?

Consider the following question. After measuring two headphone amplifiers using the above method, you obtain repeatable Class B click-and-pop suppression results, with the first amplifier having a KCP of -59dBV and the second having a KCP of -61dBV. Is the second amplifier really significantly less noisy than the first? Or are both results acceptable? The measurement is objective, but the definition of "acceptable" is still subjective.

An acceptable, detectable level of click-and-pop suppression depends on many factors: the efficiency of the headphone/speaker being tested, the typical distance from the ear to the transducer, the SHDN cycling frequency, and the background noise level during listening.

Although many factors affect the establishment of an acceptable click-and-pop level in many applications, it is possible to establish a reliable benchmark for the specification. Note that in the test results for Class B click-and-pop of Maxim headphone amplifiers (Table 2), all tests were performed with a 32Ω load resistor, and each KCP value represents the average of four samples at each port.

Table 2. KCP Values ​​for Headphone Amplifiers (A-weighted, 32 times/sec, peak voltage, 32Ω load)

Part Number	KCP		Comments
	Into SHDN (dBV)	Out of SHDN (dBV)
MAX9750C Headphone Amp	-55.8	-47.9	+3dB gain setting
MAX9760 Headphone Amp	-57.4	-56.2	Unity gain, 15k resistors, 220µF output capacitors
MAX4410	-69.9	-77.8	Unity gain, 10k resistors
MAX4299	-59.1	-49.4	Category A (no SHDN)

The above data is the result of Maxim's testing of KCP performance. To ultimately eliminate the subjective factor in amplifier performance testing, Maxim recommends that other semiconductor suppliers adopt this method and the defined KCP parameters. For more information on this test, please visit the following link

1. Maxim Audio Product Information
2. Maxim Audio Discussion Group
3. For audio test and measurement standards visit: Audio Precision World Network.

A similar article appeared in the March 2005 issue of EDN.

1 Other manufacturers of similar equipment include Rhode & Schwartz (audio analyzer) and Prism Sound (dScope).

Appendix. Calibration Equivalent Equipment

The objective test approach to obtain click-and-pop performance in this application note uses Audio Precision's System 1 and System 2 audio analyzers. If a System 1 or System 2 analyzer is not available, the following approach can be used.

KCP performance measurements can be made using equivalent test equipment from other manufacturers. Figure A shows a general test setup of an audio analyzer and DUT.


Figure A. Click-and-pop test setup using equivalent test equipment from other manufacturers

Before recording test results and making direct comparisons, the test setup should be calibrated. In addition, it is necessary to verify that the total energy recorded by the equivalent analyzer is, in fact, linear with the input amplitude. Only in this way can the energy of the click and pop be accurately recorded, especially when fast rising transients are present in the audio band. A simple calibration requires a function generator and an equivalent analyzer. (See Figure B for an example.) Calibrate as follows:

1. A 0.5 Hz square wave of known amplitude is applied to the input of the equivalent audio analyzer.
2. Set up the analyzer to detect the peak voltage after A-weighting.
3. Record the peak voltage readings for various input signal amplitudes.

Figure B. Test setup for calibration of an equivalent audio analyzer. Calibration must be performed to ensure that the total energy recorded by the equivalent analyzer is linear for a variety of input amplitudes.

Table A below shows the calibration results for a System 2 Audio Precision audio analyzer set to A-weighted, sampling 32 times/sec. An input signal with the 1X/Y auto range set to 1mVP-P to 40mVP-P produces a 6dB weighting factor. This 6dB weighting factor is related to the A-weighted limited transfer function of the Audio Precision analyzer. For input signals greater than 40mVP-P, the calibration results become nonlinear for this particular setup. This range is appropriate for most amplifiers.

Table A. Audio Precision System 2 Calibration Results

VIN (mVP-P)	VTHEORETICAL (dBV)	VREADING (dBV)	A-Weighted Calibration
1	-60.000	-66.295	6.295
5	-46.021	-52.391	6.370
10	-40.000	-46.186	6.186
20	-33.979	-39.883	5.904
40	-27.958	-34.120	6.162
60	-24.437	-32.140	7.703
80	-21.938	-30.791	8.853
100	-20.000	-28.747	8.747

This calibration can be applied to an equivalent analyzer to ensure accurate click-and-pop performance measurements. In addition, by determining the same calibration values ​​and the appropriate input signal range, the click-and-pop performance specifications of the two amplifiers can be accurately compared using an equivalent audio analyzer.