New 32-bit microcontroller enables true single-chip DRM digital audio codec

With the explosive growth of digital audio products in recent years, many chips or chipsets have appeared on the market to meet the requirements of more advanced players. However, there are also some pitfalls for some products entering the digital audio market. Things are not as simple as choosing the appropriate processor hardware.

In recent years, stand-alone audio and multimedia players have dominated the market, but some consumers are also trying to connect their portable media players to their home or car stereo equipment, which has also prompted some manufacturers in the home or car stereo equipment market to use their high-fidelity (HiFi) systems in the digital audio era.

Using computer hardware
Some manufacturers have tried to use computer hardware in audio and multimedia players. Although these hardwares are very good at processing the huge data of computer multitasking, they bring great challenges to audio designers, mainly manifested in relatively slow speed, unpredictable task switching, and relatively poor real-time performance. Such computer hardware-implemented systems are subject to high power consumption and require high CPU running speeds to achieve uninterrupted playback. On the other hand, low integration is also a problem. Many systems require 3 or 4 chips, including MPU, SDRAM, NAND flash memory and audio codec. It is not realistic to further reduce the number of chips because modern wafer processes do not allow these technologies to be implemented on the same silicon chip.

Application-Specific Integrated Circuit Approach
Other manufacturers have taken a single-chip application-specific integrated circuit (ASIC) approach. In the past, microcontrollers only had enough processing power to decode digital audio content, so a custom MP3 or similar decoder had to be added to the existing microcontroller. This approach has long been considered an advantage by portable audio player manufacturers because a single IC means a smaller circuit board can be designed. Another advantage of this approach is that it reduces power consumption, allowing the use of smaller batteries. The reduction in power consumption is partly due to the lower clock rate required for the DSP, and the elimination of the peripheral memory bus. Although this approach has advantages in size and power consumption, it also has significant risks. Due to changes in technology standards and changing consumer expectations, a new product may become obsolete before it is actually available on the market.

The Perfect Audio Codec
Obviously, the ideal solution would be to use an off-the-shelf microcontroller with a CPU powerful enough to decode the audio content, eliminating the need for a custom audio decoder. This approach would have all the advantages of a single-chip ASIC, but with no hardware that could become obsolete, making the platform flexible enough to meet changes in consumer demand.

It is very clear that where there is a need, there is a solution. Atmel's newly released 32-bit microcontroller AT32UC3A3 can provide the required DSP performance and support innovative DMA schemes, which means that a highly predictable audio decoder can be implemented to achieve advanced audio quality.

Software Audio Decoding
The control center of the AVR32 microcontroller is the AVR32 CPU core, which is unique in that it can provide a wide range of DSP instructions that are only available on high-end CPUs and DSPs. Due to its high performance, it no longer requires customized audio decoder hardware to decode MP3 data streams, and the operating frequency is only slightly higher than 20MHz. Since the CPU can operate at a maximum frequency of 72MHz, the CPU has enough performance margin to handle more "heavy" audio formats such as AAC and AAC+. In addition, the CPU has enough performance to run the operating system, file storage, communication, etc.

In order to support the playback of encrypted audio formats, Atmel can provide a derivative product of the AT32UC3A3 series, which has a built-in 256-bit AES encryption unit, which can effectively improve the decryption speed of digital encrypted audio. However, this product with an AES encryption unit is subject to strict export restrictions in the United States, so it is not aimed at the traditional consumer audio equipment market.

High-Fidelity Playback
After the digital audio is decompressed, it must be converted to analog audio for playback on a series of speakers. For stereo output, the AVR32 microcontroller provides a high-fidelity stereo 16-bit streaming DAC that requires only a small external power amplifier to achieve the voltage signals required for the audio lines, headphones, and external speakers. 4-channel audio playback and full surround sound effects require an external audio codec, most of which are connected to the microcontroller through the IIS interface.

Flexible File Storage
The AVR32 microcontroller has enough flash and SRAM to store firmware, decode audio, buffer communications, etc., but there is not enough memory on the chip to buffer more than two seconds of audio content. The AVR32 microcontroller offers a wide range of memory options, and the three most popular memory options are SD/MMC cards, USB, and NAND flash. Audio devices can use any combination to store audio content. The SD card interface can support up to two High Speed ​​High Capacity SD cards, while the USB host interface can support not only regular USB memory sticks, but also media players, cameras, and mobile phones with USB interfaces. The NAND flash interface supports up to two memory chips, supporting single-level cell (SLC) and multi-level cell (MLC) ECC. For applications that only need one or two seconds of audio buffering, the on-chip SRAM is sufficient and no external memory is required.

High-speed communications
Another important feature of digital audio equipment is the speed at which music and other audio content can be transferred in and out. Processing a single digital audio channel may only require 200kb/s or less of bandwidth. But for applications with large audio libraries, consumers demand faster communication speeds to synchronize large audio libraries more quickly. With this in mind, the AVR32 AT32UC3A3 has a high-speed USB interface and an MMC/SD slot, and supports WLAN under SDIO. Other members of the product family, such as the AT32UC3A0 and AT32UC3A1, also support a full-speed USB interface, as well as a 100 Mb/s Ethernet port. Obviously, the AVR32 microcontroller is not just designed for mainstream consumer electronics devices, it also provides a wide range of traditional interfaces, including USART, SPI and I2C, while also having enough timers to run DC and stepper motors.

The AVR32 microcontroller
offers many high-speed communication interfaces, which means that a single SRAM can quickly become a system bottleneck. Atmel anticipated these situations and added no less than four SRAMs to the AT32UC3A3, two of which provide dual-port access to speed up communication and avoid contention. This ensures that the SRAM bandwidth does not affect the transfer speed of the system, and most importantly, this ensures that audio playback quality is not affected even when there is high-speed communication in the background.

The backbone of the AVR32 microcontroller is a multi-layer high-speed bus that enables the CPU and peripherals to share more data per cycle by allowing simultaneous access between multiple masters and slaves.

To develop this multi-layer bus, Atmel engineers have divided the main 128Kb SRAM of the AT32UC3A3 microcontroller into three modules, each of which is given its own independent memory interface to the bus. This clever approach allows three high-speed communication interfaces to access the SRAM simultaneously without waiting for data. Half of the SRAM is a high-speed SRAM embedded in the CPU and has dual-port access performance. The other half of the SRAM is a low-power SRAM, and the CPU and peripherals share the available bandwidth. To round out the design, the high-speed USB interface has its own 1.5Kb dual-port SRAM to ensure that the entire data packet can be transmitted efficiently without the risk of being affected by the main system.

The low-power
AVR32 microcontroller consumes less than 2.0mW/MHz, allowing the audio device to provide more than 150 hours of playback time from two AA batteries. In standby mode, with only the real-time clock running, the audio device can be left in a drawer for up to 9 years without running out of power.