Tips for designing a smart phone based on WM8994

Nowadays, application-centric design has become the mainstream trend in smartphones , and peripheral functions such as various modems have been offloaded to various independent chipsets. There are many reasons for this, partly due to fierce market competition, including the desire to differentiate through leading performance, and the pressure of fast time to market faced by mobile phone manufacturers, as well as the different technical development ideas in the fields of multimedia technology and wireless chipsets.

For the various application processors, the trend is often away from the "standard-driven" development path and towards the development trend of wireless communication chipsets. The core capabilities required in these areas are different: Bluetooth, cellular or Wi-Fi chipsets need to meet their respective specifications, while requiring the lowest cost and competitive power consumption; in addition, in order to cope with the competition, application processors need to achieve a large degree of differentiation in their digital functions and performance. Increasingly fierce competition has stretched the capabilities of single-chip solution providers to the limit, and it is no longer practical to maintain their leading edge in all aspects of device functionality. Today, cost and time-to-market pressures are driving the industry to develop solutions with a more efficient division of labor, which in the long run is beneficial to both chip designers and consumers, because it allows chipset providers to focus on what they do best and eliminate many cost-wasting duplicate functions in the system.

In the past, these peripheral functions were often integrated into some application processor or modem in order to add value through integration. Time and cost pressures often led to less than ideal implementations, especially where specialized skills were required. Audio has long been a victim of over-integration, with duplicated and poor quality implementations due to the reliance on two or more chipsets, plus a host of “patch” components added by system integrators to support the mixing, switching, pop and click suppression, and power amplification required by their end products.

Wolfson Microelectronics' WM8994 audio-centric codec takes advantage of the latest application-centric modem architecture that integrates all mixed-signal audio functions, ensuring that low-cost processors can be used for successful system implementation and that audio signal path characteristics are consistent across a variety of data sources (Figure 1).

Figure 1: The WM8994 uses an application-centric modem architecture.

The WM8994 integrates a wide range of analog and digital I/Os to drive all types of audio transducers in the phone, while also interfacing with various processors, such as those used to handle Bluetooth transceivers, FM radios, cellular modems, and application processors. This high level of audio integration brings flexibility, consistency, and high performance to the phone, while requiring little planning at the design stage to maximize the benefits of the system architecture.

For most mixed-signal systems, attention to clocking schemes and good control of system noise, coupled with good power and ground schemes and a low-noise reference, will achieve optimal system performance. However, for complex audio systems, there are many other challenges. The following design tips will show designers how to take full advantage of the WM8994 in the challenging smartphone environment.

Ten tips for designing smartphones using the WM8994

1. Plan audio application scenarios in advance. For each application, first figure out which chipsets will be activated, then figure out where the signal comes from and where it goes? At the same time, what is the most efficient transmission path? Some unexpected events must be taken into account during the audio streaming process. For example, when a specific audio channel is activated, if an event occurs, what will be the result? Should the channel be terminated? Attenuated? Mixed with some other sound? Or re-find other transmission paths? How many small applications of this type will actually run at the same time?

2. Outline the audio clock scheme for the key application scenarios, and then also draw up clock schemes for those corner applications. From a "conventional" point of view, today's edge applications may be tomorrow's "must-have" functions. Make sure that each phase-locked loop (FLL), clock distributor and audio interface are configured properly so that the memory buffer in the processor is not overfilled or overempty due to non-optimal configuration. Since the WM8994 has sufficient clock programmability, it can cover most scenarios; but the question worth considering is where should the FLL reference clock come from for each application? Which port should be configured as the operating master mode? Connecting one of the multiple clock input pins to the always awake 32KHz clock can achieve power savings in many cases, especially in standby mode where a clock is needed to de-jitter the GPIO input, or in a simple low-power music playback mode. Using the audio interface bit clock and frame clock as the reference input of the FLL, there is no need for an additional high-frequency master clock input, thus achieving power savings.

The WM8994's sample rate converter allows the device to operate in two completely independent clock domains, allowing audio mixing and routing across these clock boundaries. Since the device's sample rate converter operates on a maximum of two full-duplex channels, there are few restrictions when the sample rate converter is connected to Audio Interface 2 and/or Audio Interface 3, which are typically interfaces to wireless chipsets that rarely require more than one or two concurrent channels.

3. Don't worry about speaker amplifiers that use Class D operation. The power savings of amplifiers in this mode are very large, and the power savings can usually reach tens or even hundreds of milliwatts. As more mobile phones with stereo speakers appear on the market, the power savings will naturally double. Imagine if the power budget saved could be used for other functions in the system. For those engineers who are not familiar with Class D amplification technology, the data sheets give specific suggestions on various aspects, including electromagnetic compatibility (EMC) design, how to choose speakers, how to improve efficiency, and how to achieve optimal circuit board design. In early mobile phone designs, people usually paid more attention to EMC; today, design engineers often pay attention to thermal design because in fact, in some cases, heat dissipation has a greater constraint on mobile phone performance. So design engineers began to pay more attention to power efficiency. Class D amplifiers have higher efficiency, which reduces the current surge on the power supply used by the speaker amplifier, which can cause the battery voltage to drop and the system to stop working earlier, shortening the battery life, especially when this surge occurs together with other current surges in the system. The above problems will be more serious. The reduction in battery voltage drop will help reduce distortion at high signal levels, and in some cases will provide more room to increase the maximum output signal of the speaker.

4. Wherever possible, digital connections should be used to replace analog connections to the modem. Although the WM8994 can also support analog voice data channels, this analog solution will consume more system-level power and be more susceptible to crosstalk, which in many cases will lead to PCB rework and increase the additional cost and board area overhead of passive components in the signal path (Figure 2).

Figure 2: Using the WM8994 will significantly improve audio performance.

For example, when a voice call to a wired headset is transmitted using a pulse code modulation (PCM) link between the codec and the modem, connecting to a digital data stream will allow activation of Class W mode on the headphone output, which will save power for the codec; it will also save power for the modem by turning off the typically power-hungry internal DAC. Using the digital connection during a handset call or in hands-free mode also allows the speaker to be tuned using the integrated parametric equalization (EQ), and control of the EQ settings can be implemented by only one device (usually the application processor), rather than relying on two or three processors to duplicate the same tuning function.

5. Where possible, the benefits of digital microphones should be considered at the system level rather than at the device level. Abandoning long analog microphone wires will help reduce costs and save space. Today, microphone arrays are being widely used to reduce noise, and as the number of microphones in mobile phones increases, there is a high probability that one of the many microphone signal channels will be close to a noise source. Although noise immunity can be improved through differential connections and careful wiring, the need for high-quality capacitors in the signal path is as high as doubled. Using digital connections will improve noise immunity, and for this reason, the WM8994 provides up to 4 digital microphone channels. It must be noted that in order to obtain the best performance, it is critical to select microphones with the best matching characteristics when using microphone arrays.

6. Make sure that the serial interface of the application processor connected to the WM8994 can support the highest sampling rate required by all audio applications. Also, check the number of TDM channels required to support large processing such as microphone array processing or multiple concurrent audio digital streams.

7. An effective power management architecture should be selected. In fact, the WM8994 only needs two external power supply voltages. Usually, one is directly connected to the battery and the other is connected to the 1.8V power supply. Since the 1.8V power supply powers the charge pump, which in turn powers the headphone amplifier with a ground reference level, it is best to use a high-efficiency switching power regulator for this power supply. For example, a low-dropout linear regulator (LDO) can be used to power it to ensure that the low power consumption performance of the headphones is not affected.

8. When setting up voice calls for handsets and wired headset handsets, use the digital sidetone feature wherever possible; the wind noise filtering and low latency of the digital channel can provide a more natural surround sidetone compared to analog sidetone, which contains distracting low-frequency mechanical and wind noise components, or uses software to implement sidetone in the baseband processor, which introduces some delays and causes some changes in the sound.

9. Use an analog volume control on the headphone output, or at least within the first few dB below full scale. This can provide an additional 4 dB of noise performance improvement, which is invaluable with today’s high sensitivity headsets. As the acoustic efficiency of headphones has improved dramatically in recent years, the requirements for the signal-to-noise performance of the digital-to-analog converter (DAC) and the noise performance of the headphone amplifier have become more stringent; insufficient signal-to-noise performance in the DAC can result in an audible tone or “hiss” when using the headset. If an analog volume control is used, the WM8994 can provide up to 100 dB of signal-to-noise ratio into the load, which will reduce the noise at the headphone by an additional 4 dB. Typically, this solution provides up to 10 dB improvement in noise performance over many codecs on the market. For the user, this improvement can be the difference between being able to hear the noise and not being able to hear it at all.

10. Careful selection of series resistors for headphone outputs. In many countries there are clear regulations on the maximum output power of headphones in order to avoid hearing damage, so the design engineer must keep this in mind when choosing the resistor value. The resistor value should be small enough so that the output voltage attenuation at the line load (such as Hi-Fi input) can still support 1Vrms voltage at full scale, which is generally desired; on the other hand, the resistor value must be large enough so that for smaller loads, after appropriate attenuation, it can still meet the regulations on headphone power in those countries. For the WM8994, due to its high output power rating, even with a relatively large series resistor, its headphone output can still drive headphones with very low distortion levels. 1Vrms output can be achieved even when the total load is required to be as low as 32Ω.

Future-proof audio performance

The audio hub solution developed in recent years can easily integrate high-performance application processors and more cost-effective modem systems by integrating all required audio function sets into a single chip. This solution eliminates duplication in the system, minimizes the interdependence of components, and provides a stable platform for the development of future products.

Based on the rich signal channels and available mixing functions on the WM8994, a simple solution is provided for various new audio applications in the future without the need for heavy software development. The device supports a large number of digital and analog channels, which will enable mobile phone design engineers to achieve further design innovations without adding new audio components to the system. If the above basic principles of beneficial system design can be followed, it is entirely possible to achieve excellent audio performance in a wide range of applications.