Outlook 2012: Bringing together the MCU and DSP
4 mins read
There has been a significant shift over the last 10 years towards combining MCUs and digital signal processors (DSP) within a single System on a Chip (SoC), thereby bringing the DSP's functionality and capabilities within the MCU.
The increasing number of connected devices on the market and a move towards acceptance of an 'internet of things', where more and more devices are being connected to the internet, has driven the demand for converged processors. Most mobile consumer devices include music playback functionally, with the market expanding from standalone MP3 players toward smartphones and tablets that have to integrate music applications to remain competitive.
Meanwhile, connected systems such as in car entertainment technologies, include increasingly sophisticated audio systems, as well as onboard computers, digital TV and physical media players, with DSPs traditionally been used within such devices.
Until now, separate processors have completed computing tasks within a device; for example, DSPs handled audio processing, while mcus tackled everyday computing and user interface requirements. But the two are now being combined and a number of such processors are beginning to appear on the market; for example, NXP's LPC4300, STMicroelectronics' STM32 F4, Texas Instruments' Stellaris M4F series and Freescale's Kinetis family.
DEFINING ROLES
Historically, DSPs have been used for applications such as audio signal processing, audio compression and decompression, image processing and motor and power control. As the number of mobile devices on the market has increased, DSPs have become commonplace in smartphones, MP3 players and audio hi-fi systems. These DSPs have been designed to focus on specific numerically intensive tasks.
Comparatively, mcus have many more applications as they are suitable for a variety of processing and control tasks within embedded systems. The MCU's low power requirements, when compared to applications processors, make them more attractive and they are becoming commonplace within most electronic devices. This ubiquity in the consumer electronics space has driven requirements with regard to bill of materials costs and energy efficiency concerns. In addition, compatibility with common interface connections is another advantageous feature of MCUs compared with DSPs.
Where digital signal processing functionality is combined with a MCU, there must be a sufficient balance between the low cost and power advantages of an MCU and the ability to process digital signals with the same efficiency as a dedicated DSP.
MCUs are increasingly being applied where a standalone DSP has traditionally been required, which is bringing a market shift away from DSPs.
With more consumer products incorporating multimedia functionally, digital signal processing is being incorporated within MCUs to ensure they are capable of handling the increasing processing requirements.
In 2012, we'll see many more combined MCU and DSP systems in which the DSP is replaced directly for the first time. In a combined processor, relatively modest digital signal processing tasks should be handled within the broad footprint and cost of a standard microcontroller. During 2012, OEMs will begin using these combined processors to eliminate the need for a standalone DSP, instead replacing it with a single combined processor, as a wider array of low cost DSP capable microcontrollers enter volume production.
A combined processor is usually, although not universally, called a Digital Signal Controller (DSC): NXP, for example, refers to the LPC4300 as a DSC. While this builds on the capabilities of the LPC processor range, it is currently the only LPC product to incorporate DSP functionality.
In the past, designers have had to create an MCU with a separate coprocessor block to handle digital signal processing. It is possible to combine the two within a single processor. An example of this is the ARM Cortex-M4 processor, which is capable of incorporating DSP capabilities within the MCU.
OPTIONS FOR OEMS
There are three possibilities for OEMs looking to incorporate an MCU with DSP capabilities. Firstly, where a product already includes an MCU, a Cortex-M4 processor allows DSP functionality to be built into an MCU without increasing costs for the OEM. Secondly, for those currently using a discrete DSP – a relatively expensive component with a difficult tool chain – this can be replaced by a Cortex-M4 based MCU, allowing developers to tap into the ARM ecosystem. Finally, those using separate MCUs and DSPs can merge them into one Cortex-M4 based chip.
ADVANTAGES OF COMBINATION
Most devices that require digital signal processing capabilities have a DSP sitting alongside the MCU. In such instances, there is a clear advantage to combining the two, as the need for a standalone DSP is removed.
Removing a dedicated DSP provides several potential benefits. Initially, there is a cost saving; there is one less chip to pay for. This compares favourably to the alternative of using a high cost dedicated DSP. Increased power efficiency is a secondary advantage. Provided the MCU offers sufficient power efficiency, combining the existing MCU with DSP functionality allows the power requirement of a standalone DSP to be removed. Having one processor means there is no need for communication between an MCU and a DSP, which saves power from the outset. Digital signal processing can then be handled by the MCU without increasing the power requirement of the unit significantly. The actual processing operation is also simplified, with the single unit
requiring one tool chain, rather than two.
APPLYING THE COMBINED MCU AND DSP
A combined MCU is highly suited to products that require audio functionality and playback, such as multimedia loudspeakers, iPod docks and in car entertainment systems. Audio will be one of the first market segments to benefit because the computational requirements are within the capabilities of recently announced devices.
In the software space, DSP Concepts has provided support for the Cortex-M4 within its Audio Weaver design environment. The tool combines a drag and drop signal flow editor, together with audio processing libraries optimised for the Cortex-M4 processor. DSP Concepts has been working on support for the Cortex-M4 for the past 12 months in anticipation of increasing demand for audio applications that can benefit from a combined MCU.
Recent trends towards streaming, connectivity and interactive user interfaces will increase the proliferation of microcontrollers with the audio
space. Many of these products require some form of audio processing and running the audio processing on the existing MCU has benefits for system cost and ease of integration. In the coming year, entry level and high quality audio processing tasks currently done by dedicated DSPs will migrate towards MCUs. These devices excel at peripheral support, connectivity and software device drivers, and recently added DSP features make running audio processing possible.
Audio processing is by no means the only application for a combined MCU. The lower cost, more power efficient units are applicable for motor control and power control applications in the industrial and medical sectors. As more systems emerge, the number of applications will expand as OEMs and designers become familiar with the new units.
It is clear that, through the combination of MCU and DSP, not only will we see advantages in overall bill of materials cost, but also realise significant benefits in terms of power consumption and PCB size.