Full-featured and complete GSM/GPRS cellular audio recording/playback solution

The complexity and high density of cellular telephone circuits present a challenge for system designers, and establishing a high-quality audio recording/playback channel that meets vendor specifications can be a difficult task. New multimedia features added to the new model, such as cameras, ringtone generators, MP3 players, and voice memos, often require a greater degree of product change. This is not just about adding some new components, but also making major changes to the layout of the printed board, which can cause poor grounding and new noise problems.

This article refers to the address: http://

The noise and interference on the analog audio channel in a cellular telephone is typically due to demodulation or shared/bad grounding of the radio to audio band.

When a high-energy RF signal from a telephone antenna is received, an audio circuit having a relatively low bandwidth in the telephone erroneously demodulates the RF transmission signal. This can degrade the noise background of the audio channel. A certain technique and structure can be employed in the audio amplifier circuit to minimize this deteriorating effect, and placing the suppression component in close proximity to the input pin is an inexpensive remedy. Low-value grounding capacitors are often used because designers typically choose the lowest capacitor impedance based on the RF carrier frequency.

Integrating all of the commonly used analog audio input/output functions into a single IC is a very effective solution that minimizes the effects of shared/bad grounding. This is actually the most problematic and troublesome grounding problem being transferred from the printed board layout engineer to the IC manufacturer. In addition to the necessary analog audio input/output capabilities, this IC must also provide a digital audio interface that is sufficient to support the voice band and any multimedia functions such as application processors. The IC should also provide zoned shutdown control for different units to maximize battery life.

The following focuses on some of the analog/digital audio issues that arise in a single-chip solution. We will discuss the MAX9851, a technical solution that simplifies the design of GSM/GPRS cellular phones.

Analog audio - reduce microphone noise

High-gain audio circuits, such as microphone amplifiers (microphone amplifiers), are most affected by poor grounding. This is especially true for single-ended circuit configurations in which a small voltage difference between the microphone amplifier reference ground and the signal source ground (in this case, the GND pin of the microphone) is amplified into the signal path. In complex products like cellular phones, the ground plane of the audio portion is often shared with other circuits, and since the copper ground plane is not "zero ohms" (which we often think), this can cause performance degradation problems. Therefore, if any current flows through this finite resistor, a small potential difference is created at the ground plane.

Grounding problems can be solved with a fully differential input microphone amplifier. This method has been adopted by the MAX9851. The actual use of the differential input is to remotely sense the GND pin of the microphone. After remote sensing, the AC voltage difference between the CODEC reference and the MOSFET is presented as a common mode signal to the mic amplifier. This voltage difference is attenuated by the amplifier's common-mode rejection ratio, thus significantly reducing its equivalent noise contribution to the signal path. The only cost of this design is the need to lay a printed circuit board between the microphone and the CODEC and add a coupling capacitor.

The MAX9851 can also be switched to an external stereo microphone input instead of an internal microphone. This type of input usually comes from car hands-free or other external headphones. In this case, the EXTMICGND pin "Kelvin senses" both L and R channels, using the input CMRR of the amplifier to eliminate ground noise, as described above. The printed circuit board layout of EXTMICGND should extend to the GND end of the car hands-free socket or headphone jack for best results (Figure 1).

The microphone bias circuit also introduces significant noise into the signal path. Most of the bias voltage noise is presented directly at the input of the mic amplifier. A more reasonable mic amplifier design, as integrated in the MAX9851, should provide a low noise bias voltage that is adjusted to match the output noise level to the microphone amplifier input noise level.

Analog Audio - Stereo DirectDrive? Headset and Receiver Output

To play a compressed music file close to the sound quality of the CD, a high quality headphone audio playback circuit is required. Signal-to-noise ratio (SNR), linearity, and bandwidth are significantly higher than the basic 300Hz to 4kHz voice channel. Low frequency expansion can be problematic because the headphone driver typically has a series capacitor to prevent the headphone amplifier's DC bias from entering the earphone. A typical stereo earphone has a typical impedance as low as 16 Ω, and it forms a high-pass filter with the series capacitor, which attenuates the low-frequency components. To extend the low frequency response, for example down to 100Hz, two 100μF DC blocking capacitors are required for 16Ω stereo headphones.

These two series capacitors can be removed using Maxim's DirectDrive technology because the output of the amplifier is referenced to 0V. The low frequency components in this case are limited by the de-dc filter (digital source, as designed in the MAX9851) or by the input coupling capacitor on the analog source input such as a line or a mic. Another advantage of the DirectDrive design is that it completely eliminates the cause of clicks and pops when it leaves or enters the shutdown mode. Since there is no series capacitor, there is no need to charge or discharge the capacitor, and no net current flows through the headphones during the on/off process.

The stereo headphone output of the MAX9851 also works in bridge mono mode (Figure 2) to accommodate different headphones and accessories. The same socket can be used for either stereo headphones or for mono headphones (Mike Plus Switches and Speakers). In this mode the output is still referenced to ground and there is no DC voltage on the headphone cable. Therefore, the chance of a short circuit fault is much smaller.

The receiver output is also powered by the charge pump in the Direct Drive design so that the output is still single-ended, allowing the negative terminal of the speaker to be connected to GND (0V). The output still has nearly the same voltage swing as the more typical BTL (differential) output because the inverting charge pump provides a negative supply rail with an absolute value almost equal to the applied AVDD. The peak-to-peak output of the final output to the receiver speaker is almost 2 x AVDD.

Analog Audio - Class D Speaker Amplifier

The MAX9851 incorporates Maxim's third-generation Class D technology to drive 8Ω (or 4Ω) speakers. The advantages of Class D (switched) amplifiers over Class AB (linear) amplifiers are primarily in efficiency. Class AB amplifiers dissipate a large amount of power in the output components unless the amplifier is driven to the clipped state. However, since the output components of the Class D amplifier operate in the switching state, their heat consumption is much smaller, so battery life can be extended. If the cellular phone is frequently used in speaker mode, or supports push-to-talk (PTT) mode of operation, the extended battery life will be very significant.

However, special care is required when using Class D structures, especially when used in products where the core functionality is RF receiving/transmitting, such as cellular telephones. The fast switching edges created by the operation of high efficiency Class D amplifiers can cause RF emissions problems, especially when the printed circuit board wiring and speaker leads are long. To address RF emissions, the stereo Class D speaker amplifier in the MAX9851 uses the company's proprietary EMI suppression technology (actively limiting radiation) to significantly reduce RF emissions from speaker leads/wiring at the expense of slight efficiency degradation. Exquisite IC construction techniques minimize the interference of the Class D switching output stage with other sensitive low noise analog circuits in the CODEC.

The stereo amplifier operates directly from a single-cell Li-Ion battery that is not voltage-regulated and delivers 1W to an 8Ω speaker when operating from a 4.2V supply (Figure 3). If you use a lower impedance speaker, you can also output higher power, but it is difficult to find 4Ω in small-caliber speakers commonly used in cellular phones.

Digital audio - general architecture, signal flow

To implement the call—the basic function of this GSM/GPRS cellular phone, the system is required to provide an 8 kHz (or 16 kHz) sample rate ADC/DAC channel with 16-bit depth in both directions. In the MAX9851, this input/output channel is fully synchronized to the MCLK input of 13MHz (or 26MHz, corresponding to fs = 16kHz) to ensure no missing or repeated samples. The S1 digital input/output interface works in GSM voice mode (Figure 4) and provides an interface to this basic function. The S1 digital interface can operate in master or slave mode.

Many mid- to high-end phones also typically require DAC functionality with higher bit counts and higher sample rates. For example, use it to play WMA/MP3 music or generate ringtones for WAV files. Integrating digital-to-analog conversion for this function with a converter for speech helps increase integration and centralizes all data conversion tasks into a single "point source." Such integration is very useful in product design. If you try to combine two functions in the analog domain, you will encounter problems such as ground loops and differences in audio levels.

Therefore, integrating voice and multimedia data with a single converter seems to be an ideal solution. The main difficulty with this approach is that all voice conversions must be synchronized to the GSM/GPRS rate (controlled by the MCLK input). For multimedia playback, an unrelated sampling rate is usually required: for example 44.1 kHz or 48 kHz. The MAX9851 solves this problem by processing digital input data using a sample rate conversion (SRC) algorithm that allows a single DAC to simultaneously convert voice and multimedia composite data.

In slave mode, the sample rate of the input GSM voice data must be kept accurate (controlled by MCLK). The internal digital PLL is locked to the LRCLK input of the S2 digital interface to accurately play back non-synchronized multimedia audio data (average of multiple samples). In Master mode, the voice data is still accurately aligned to the required MCLK integer divide, but the S2 LRCLK data rate is only an approximation with a slight fs error and usually does not have a significant impact. S1 or S2 inputs can support sample rates from 8kHz to 48kHz.

The S2 digital input/output interface of the MAX9851 supports I2S and other popular small variants. When not working in GSM voice mode, the S1 interface can also be programmed to support I2S, providing maximum interface flexibility for multi-function high-end phones.

Digital Audio - GSM Filter

As shown in Figure 5, there is an additional filter on the S1 digital input/output interface that can be enabled in GSM voice mode. These digital units are a highly efficient implementation of the rigorously regulated low pass and high pass filters. They suppress energy near the Nyquist frequency and the low frequency band. This filter has proven to be beneficial for noise and signal leakage specifications, making it easier for phones to pass tests and certifications. Figure 6 shows the frequency response of the filter.

Conclusion

The above examples only highlight a few of the issues that cell phone system designers/architects must address. The design cycle of such terminal applications is getting shorter and shorter, and the integration of performance is constantly improving and changing, and almost every model is different. Therefore, it should be meaningful to invest a certain amount of time and choose a core chipset with good engineering planning, flexible application and comprehensive functions.

Low-noise analog circuits interface with multiple playback/recording systems at different sampling rates, and control of this portion of the circuit is only part of the overall design task. It is also important to integrate the following features in one scenario: