原创 手机音频设计的演变 Part 1 ( English )

Within the cellular handset there are a variety of audio devices, such as amplifiers, analog and digital converters, microphones, and speakers. There are also several categories of each type of device. In addition, there can be a variety of audio generators in the form of Radio, TV, and music players.

Architectures
A review of some cell phone architectures supporting various audio functions is a good way to understand the audio devices currently used in cellular handsets.

Figure 1 below describes a basic cellular handset from an audio requirements point of view. The FM radio, music synthesizer, MP3/WMA decoder, and MUX functions are optional, depending upon the requirements of a particular model. The shaded area shows the optional elements.

The basic audio devices are, of course, the voice band codec, the microphone, and an earpiece speaker. For the silent ring indicator, it is common to use a DC motor with an off-center weight attached to the motor shaft to create the vibration effect.

Depending upon the earpiece speaker impedance and amplifier Gain and output power design, the earpiece speaker often doubles as the ring sound speaker.

The MUX block in Fig. 1 is not one single component, but indicates that some sort of switching and mixing may be needed if the outputs of the various audio generators must be usable on some or all of the audio output devices (speakers or mono/stereo headset). In this simple case, the MUX elements (if any are needed) are usually built using discrete analog switches. When used in front of the audio power amplifiers, the switching and mixing are done at relatively low signal voltage and power levels and the analog switches do not have to carry full output load currents.

Sometimes, analog switches are used on the output side of the amplifiers as shown in Figure 2 to route audio to the desired location. The disadvantage of this approach is that the analog switches must carry the full load voltage and current. The advantage of this approach is that it is easy to do.

Interfaces
There is a wide variety of signal paths used for audio in cellular handsets. In addition to the normal analog signal path such as amplifiers for microphones or output speakers, audio connections to the CPUs used can be analog or digital. In the analog case the data converters are often contained in the handset baseband CPU.

But as the semiconductor process for making CPU’s trends toward higher density and lower operating voltage, analog performance suffers. Therefore, the data conversion and amplifier functions are being dis-integrated from the CPU’s and is being replaced with a digital interface to an external data conversion function either as a separate device or contained in an audio subsystem.

The data conversion path for voice is typically a bi-directional Pulse Code Modulation (PCM buss) 64Kbps data channel using A-law and mu-law logarithmic coding schemes to get more dynamics from the 8 KHz voice samples than is available with linear coding. Another telephone voice standard called Adaptive Differential PCM (ADPCM) codes voice into 4-bit values and uses a 32Kbps data channel.

Some hardware music players/decoders have D/A’s providing analog audio output. Other hardware player/decoders use the uni-directional 2-channel I2S buss which has many data format variations for digital audio output. The output audio sample rate depends upon the (1) the sample rate of the original digital audio file, or (2) the input sample rate of the original digital audio file PLUS the output sample rate of the decompression algorithm used in the decoder to play it back. Software decoders output digital data directly on the I2S buss. Therefore, the digital music D/A’s must be able to determine and use whatever I2S format they receive. For devices using these types of interfaces, a control interface buss is also required which is typically I2C or SPI. Sometimes, alternate interfaces such as SSI or AC-97 are used for control and digital audio.

The hardware synthesizer block uses an 8 bit parallel data path, plus Chip Select, Read, Write, Address, and IRQ pins, similar to ISA.

To accommodate non-voice path digital audio playback and record, the trend is to create a bi-directional version of I2S.

Each of these sources has different signal characteristics and therefore must be handled differently in the cell phone handset. Additionally, cellular phones are being made with both a Communications Processor and an Applications Processor. Either one can be used for any of the audio functions listed, and in many cases, both processors must be able to fully handle the audio functions. This situation leads to a fragmented architecture for audio.

To help alleviate architectural roadblocks and in-efficiencies, companies in the cellular handset industry have created the Mobile Industry Processor Interface Alliance (MIPI) to formulate the next-generation handset design, which will set interface specifications for various functions in the handset, including audio.

Amplifiers
Cell phones now require audio amplifiers suitable for multiple output speakers (earpiece, ringer, mono headset, stereo headphones or earbud type headphones, PTT, and speakerphone). These amplifiers all have different output power requirements.

During the last few years, audio amplifier output power for ringer speaker has risen from about 250mW to almost 2W. This is partly due to cellular handsets being used in all kinds of noisy environments, but also due to the fact that handsets are now so small that we stuff them into purses and inside coat pockets which are effective sound mufflers. Another reason is that speaker diameters have decreased, and more power is required to generate a sufficient Sound Pressure Level (SPL) from them. Features such as Push-To-Talk (PTT) and Speakerphone mode has pushed output power levels from about 550mW up to about 1.7W.

New phones on the market also offer stereo speakers for ringing tones and speakerphone functions, effectively doubling the system output power requirements as well as the power dissipation.

As the power requirements for the amplifiers have increased, the need for increased efficiency also increased. The industry has begun a shift from Class AB amplifiers to Class D amplifiers in order to save battery power as well as reduce heat within the phone.

Since the battery voltages remain relatively constant, a possible trend is to move the ringer and speakerphone speaker impedance to 4 , effectively doubling the output power to the speaker. This decision would require larger output transistors in the amplifier with the accompanying larger die size and cost.

To deliver high loudspeaker output power at cell phone battery voltage levels, the amplifier is operated in the Bridge Tied Load (BTL) mode instead of the Single Ended Output mode.

The BTL mode has the added advantage of not requiring output coupling capacitors, which would have a limiting effect on the low frequency (20Hz to about 200Hz) response.

Practically all audio loudspeaker amplifiers available for use with cell phones can achieve a closed loop frequency response that is flat to within +/- 0.5 dB. The overall system frequency response is then primarily set by the speaker characteristics.

An important parameter for audio amplifiers in handsets is Power Supply Rejection Ratio (PSSR). Typical amplifier PSRR values are specified at 1 KHz. For GSM phones, the PSRR should be specified at 217Hz (the GSM power modulation frequency) as well. PSRR values for audio power amplifiers should be greater than 60dB over the majority of the audio bandwidth.

Some baseband processors provide analog output signals in differential form, and audio amplifiers are being provided to meet that configuration. The differential configuration also improves PSRR.

When using Class D amplifiers, an important parameter is Output Noise, which should be less than 50 Vrms (using A-weighted measuring method) at the required gain setting.

Headphone amplifiers are discussed in the Headphone section.

Data Converters
When A/D and D/A functions are combined in a single device, they are called codecs (coder-decoder). Sometimes, D/A’s are mistakenly called codecs. Likewise, some software programs for audio and video are called codecs even though they only perform uni-directional data conversion. Initially, the audio data converters in a cell phone were primarily related to voice communication (Voice Band Codec) and usually had 12-14 bit resolution at an 8 KHz sample rate to support the normal telephone audio bandwidth of 300 Hz to 3.4 KHz. These devices performed A/D and D/A conversions, as well as band filtering and limiting required for voice coding in accordance with International Telecommunication Union (ITU) standards.

For voice recognition and commanding, voice band codecs are now available with 16 bit resolution and from 8 KHz to 26 KHz sample rates which extends the voice audio bandwidth to 11.7 KHz.

With the addition of digital music playback capabilities, the D/A converters have performance characteristics very similar to consumer audio D/A’s, e.g. 16 - 24 bit resolution and supporting multiple audio sample rates such 8, 11.025, 22.05, 44.1, 16, 24, 32, and 48 KHz sample rates.

A/D converters are primarily used for microphone input with characteristics suitable for voice communication. However, some handset makers allow analog audio input connections from FM radio or other sources which can be recorded in the handset. The converters in this case require capabilities basically identical to the digital music D/A converters.

Some handset makers are also requiring stereo A/D capability as well for recording in Stereo, but also for acquiring 2-channel microphone data for further processing to reduce echo’s and far field noise interference.