마이크로칩 8월
This page was machine-translated and may differ from the original. View original

MCU processing speed improvement with LEA modules

Google 우선 소스Published2017.03.17 09:44
Low-power coprocessor that triggers interrupts without CPU intervention
Integration of vector processing tasks and low power consumption possible with LEA modules

Internet-connected devices are very diverse, ranging from industrial sensors used for factory process monitoring and intelligent fire detection systems to home Wi-Fi connected thermostats, video doorbells, and barbecue thermometers.

Many fields seek to add connectivity features, but implementation is difficult due to hardware complexity and cost. However, the recent emergence of easy-to-use and cost-effective wireless connectivity modules and ICs has made this feasible.

One of the growing trends as nodes become interconnected is the demand for higher performance and localized data processing.

For example, consider a wireless glass breakage detector. In such detectors, a microphone or vibration sensor is used to capture an audio signature, and the signal is then passed through an analog-to-digital converter (ADC), typically at the end of the analog chain. Then, there are several options: the entire captured data set can be sent to a central hub for signal processing, or the data can be processed locally. This signal processing typically involves digital filtering and frequency-based signal analysis to determine whether the input signal is a valid trigger.

While technologies exist to process such audio data locally, current solutions tend to implement high-performance multi-chips, adding system complexity and costs, and, more importantly, increasing energy consumption. Since these detectors typically require a single battery to last for several years, they necessitate implementing an ultra-low power system or requiring a very large battery. Of course, rechargeable batteries are also an option, but in this case, it is necessary to ensure sufficient access to allow for intermittent recharging or to verify the existence of a permanent power supply to recharge the battery.

Figure 1: Simplified LEA block diagram

This type of application is one of the main reasons we developed the MSP430 Microcontroller (MCU) Low Energy Accelerator (LEA) in a short period of time. The LEA module is a fixed-function vector math engine installed in various MSP430 devices. The LEA module primarily performs signal processing, matrix multiplication, and other operations that require significant time and energy for computation during application execution.

An LEA module, which runs without CPU intervention, is essentially a low-power coprocessor that performs tasks and triggers an interrupt when a requested function is completed. It can execute concurrently with the main CPU to perform multiple tasks in parallel, or it can run while the CPU is in low-power sleep mode and trigger an interrupt upon the completion of a specified task.

The LEA module operates with a predefined set of instructions, each of which is optimized to achieve maximum performance with minimum energy consumption. These commands can be easily used with the MSP DSP library, which consists of more than 50 dedicated functions that are specifically optimized and written for the MSP430 MCU to enable the implementation of signal processing functions selected by the programmer.

When the LEA module is used in the device, the compiler automatically utilizes it to optimize performance beyond the optimization techniques already implemented in the MSP430 MCU C code. Developers can use the newly developed filtering and signal processing TI Design reference designs to directly verify how significantly processing speed has improved compared to an MCU that does not support the LEA module.

This TI Design reference design is based on the new MSP-EXP430FR5994 MCU LaunchPad development kit. The kit includes built-in programming and debugging interfaces, EnergyTrace++ technology to help developers optimize code for maximum energy efficiency, user LEDs and push buttons, and a 40-pin header that allows the LaunchPad kit to be used with various types of compatible BoosterPack plug-in modules.

Figure 2: Comparison of MSP430 MCUs with and without LEA

To help developers better understand the additional performance that LEA modules can provide, TI has published a benchmarking application note on the signal processing capabilities of the new low-energy accelerator. This application note covers enhanced signal processing performance while maintaining ultra-low power consumption in a 16-bit MSP430 FRAM MCU compared to an ARM® Cortex-M0+ based MCU.

The first device equipped with the new LEA module is the MSP430FR5994 MCU. This device is the latest version in a broad family of 16-bit ultra-low-power microcontrollers and includes 256KB of FRAM (ferroelectric random access memory), 8KB of shared SRAM, analog peripherals, and the LEA vector math accelerator module. The LEA module, which provides very low active power consumption of approximately 120uA/MHz and an RTC mode of less than 500nA, is a significant help in improving processing performance without additional energy consumption.
Figure 3: MSP430FR5994 Block Diagram

The MSP430FR5994 device also includes a 12-bit 200K sample ADC, a 16-channel analog comparator, an AES256 hardware accelerator, and up to eight serial ports of UART, SPI, and I2C, as well as a real-time clock with alarms and a calendar.

In conclusion, if you need to integrate FFT, FIR, IIR, or other vector processing tasks into current or future embedded designs while simultaneously achieving low current consumption, the LEA module can be the peripheral device you are looking for. Operating at ~67uA/MHz is more energy efficient than simply clocking the MCU at tens of MHz to achieve the same processing performance.
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.
김지혜 기자