How to choose low-power MCU for smart hardware development

This article compares the typical low-power MCU series on the market and analyzes that the MCU series based on the ARM. Cortex M0+ core is most suitable for the development of wearable medical devices. Device developers should pay close attention to its development trends, combine existing market demand, product system construction and upgrade planning, and make reasonable analysis to select the MCU model suitable for their own products. Then, according to the needs of special medical monitoring tasks, the most optimized low-power strategy is formulated for the MCU system, so as to develop affordable and high-performance equipment.

　　Based on the characteristics of wearable medical devices, such as low cost, high performance, high integration and long battery life, the current mainstream low-power microcontroller (MCU) series are compared, and the analysis shows that the MCU series based on the ARM Cortex M0+ core is suitable for product development in this field. In terms of power consumption level, computing performance, peripheral integration and product cost, the MCU series based on the Cortex M0+ core of major semiconductor companies are further compared in terms of parameters, providing a guide for MCU selection for wearable medical devices.

　　In recent years, the market demand for wearable medical devices has been growing rapidly, and will become an innovative industry that drives economic growth. According to the "2012-2013 China Mobile Medical Market Annual Report" released by iiMedia Research, in 2012, the scale of my country's mobile medical market reached 1.86 billion yuan, of which wearable medical devices accounted for 420 million yuan, an increase of 20% over the previous year. It is estimated that by the end of 2017, the market scale of wearable medical devices in my country will be close to 5 billion yuan, showing a rapid growth trend in the next ten years. With the growth of market demand and the popularization of products, wearable medical devices are developing in the direction of low cost, high performance, long battery life and small size, which puts more stringent requirements on the control core of the device - the microcontroller (MCU). The trend of wearable devices requires the MCU selected by the device to have the characteristics of low cost, low power consumption, high computing power and high integration, otherwise it will be eliminated by the market and users.

　　1 Introduction to Wearable Medical Devices

　　Wearable medical devices integrate non-invasive physiological signal detection technology into daily wearable clothing and devices, with the advantages of simplicity, portability and long-term monitoring. This type of equipment can monitor the physiological condition of the human body for a long time anytime and anywhere, and has been widely used in chronic disease monitoring, home care and health care, sleep quality monitoring, etc., which is conducive to the early detection, early diagnosis and early treatment of chronic and hidden diseases.

　　1.1 Application of wearable medical devices

　　在市场和用户的追捧热潮下，各种穿戴式医疗设备的解决方案和新产品层出不穷，功能和性能也在不断提升。例如我国的迈瑞公司推出的MC-6800型动态血压监测仪，仅需将充放气的袖带绑在用户手臂上，就能在各种状况下进行24 h无创性动态血压监测。美国Medtronic公司推出的血糖实时连续监测系统（CGMS）可以连续工作3d，仅需将检测探头贴在患者腹部，每10s会对皮下间质液里的葡萄糖浓度进行测量，并将获得的数据通过无线方式传送到接收器上。美国SPO Medical公司推出的PulseOx 6000型“血氧手指套”能长时间工作500 h，仅需套在手指上即可实时监测用户的血氧饱和度和心率，可靠性堪比体温计或血压计。这些产品都体现了区别于常规电子仪器的显著特征：①非介入地检测生理信号；②通过无线或有线的方式连接用户、医护人员和数据系统；③续航时间长；④安全可靠。

　　1.2 Demand Analysis for Wearable Medical Devices

　　In order to meet the requirements of wearable medical devices in terms of power consumption, performance, and volume, the selected MCU needs to meet the following requirements: ① low cost; ② high energy efficiency; ③ high sleep efficiency; ④ high integration. In terms of cost control, you can consider low-power 8/16-bit microcontrollers or 32-bit microcontrollers based on the ARM Cortex-M series core. These chips have huge shipments and generally low batch prices. In terms of energy efficiency, you should choose an MCU series with low operating power consumption and high computing power. Low power consumption can improve battery life, and high computing power is conducive to running complex algorithms and data processing on the chip. In terms of sleep efficiency, you should choose an MCU series with flexible and diverse sleep modes, ultra-low sleep power consumption, and extremely short wake-up time. In terms of integration, you can choose MCU series with rich peripherals and superior performance, which is conducive to reducing size, reducing hardware costs, and improving system stability.

　　2 Comparison of Typical Low-Power MCU Series

　　Major semiconductor companies such as Freescale, ST, NXP, SiliconLabs, Atmel, TI, Microchip, etc. have launched low-end and mid-range MCU series suitable for wearable medical devices. Table 1 and Table 2 compare the typical low-power MCU series of 16-bit and 32-bit. 8-bit MCU is not included in the comparison list. This is because 8-bit MCU is no longer suitable for the development trend of wearable medical devices, and its market is being eroded by MCUs with ARM Cortex-M series cores.

　　表1重点比较了16 bit/32 bit内核的性能差别，32bit的内核在运算效率方面全面超越16 bit 的内核，意味着当穿戴式医疗设备需要在片上执行数据处理和复杂算法时，Cortex-M系列内核的32 bit MCU更具优势。表2则将典型的低功耗MCU展开能效对比，可以发现16 bit MCU在低功耗方面的优势已不明显，以低功耗著称的MSP430系列在运行功耗和休眠功耗方面跟Cortex-M系列32 bit内核的STM32L系列相差无几。而32 bit MCU在休眠状态下的唤醒时间也能做到了10 μs以下，在休眠效率、快速响应方面有良好表现。

　　Table 1 Performance comparison of typical low-power core architectures

　　Note: (1) The core performance test results (CoreMark Scores) are based on the data published by the EEMBC organization.

　　Table 2 Energy efficiency comparison of typical low-power MCUs

　　Note: (1) For the test reports of specific MCU series models in Table 1, the selected models have similar on-chip configurations and the Flash capacity is 64 kB;

　　(2) At room temperature +25 oC, all peripherals are turned off, and the program runs from Flash; the MCU power supply voltage is 3.0 V except for 3.3 V for PIC24 and 3.6 V for Nano120; the test results of each model are the best configuration under the current main frequency;

　　(3) Test standard for sleep power consumption: the on-chip main clock and all peripherals are turned off, the RTC is turned on, and the RAM is retained.

　　From Table 1 and Table 2, it can be seen that the 32-bit MCU of the Cortex-M series core has achieved the same power consumption level as the traditional 8/16-bit MCU, and has obvious advantages in computing efficiency, making it more suitable for wearable medical devices that have high requirements for tasks and algorithms.

　　3 MCU selection analysis based on Cortex-M0+ core

　　3.1 Comparison of Cortex M series cores

　　The low-power members of the Cortex-M series include M3, M0 and M0+, which are designed by ARM for applications that are cost-sensitive and have high requirements for energy efficiency. When the traditional 8/16-bit MCUs were becoming increasingly weak in terms of performance and functionality, ARM launched the low-cost, low-power, and high-efficiency Cortex-M0 core in 2009. The Cortex-M0 core defeated the traditional 8-bit MCU with its excellent performance and successfully entered the low-end MCU market. Taking this opportunity, ARM appropriately launched the upgraded version of M0, M0+, in 2012, further optimizing and adding energy efficiency and functionality, and providing faster task processing capabilities with ultra-low energy consumption.

　　From the data in Tables 1 and 2, we can see that the ranking of the three core performances is M3>M0+>M0, and the ranking of the operating power consumption is M3>M0>M0+, that is, the energy efficiency of the M0+ core is higher than that of the M0, and the computing performance is second only to the M3. Since the M0+ has an advantage over the M3 in terms of price, it is more suitable for executing low-cost, high-energy-efficiency tasks. In summary, it can be seen that the MCU with the M0+ core is the most suitable for those devices that have strict requirements on power consumption, complex computing tasks, and need to control costs.

　　3.2 Mainstream MCU series based on Cortex M0+ core

　　Major MCU manufacturers have integrated and optimized the Cortex-M0+ core based on their own advantages, and each has its own strengths in power consumption, performance, and peripherals. Table 3 lists the mainstream MCU series with M0+ cores on the market and analyzes them in combination with the needs of wearable medical devices.

　　Table 3 Mainstream MCU series based on Cortex M0+ core

　　Note: (1) Both ST and NXP have established product lines covering all cores of the Cortex-M series. The Chinese market for Cortex-M series MCUs reached US$168 million in 2012, with ST ranking first with a market share of 35%, while NXP ranked second with 32%;

　　(2) Silicon Labs acquired Energy Micro, a company specializing in low power consumption, in 2013. The Zero Gecko series launched later absorbed the advantages of the previous EFM32 series' ultra-low power consumption.

　　The above-mentioned Cortex M0+ core MCU series can provide a variety of options for wearable medical device developers, and the specific MCU model should be determined according to the actual needs of the device. In the same series, the MCU's highest main frequency, core efficiency, and power consumption are consistent, and the difference between specific models lies in the on-chip resources. As shown in Table 4, the STM32L0 series is divided into 3 main product lines, and the differences are reflected in some special integrated peripherals, such as DAC, USB controller, and LCD controller. The proper selection of these highly integrated MCUs can help reduce the number of external chips, which can reduce system costs and power consumption. Therefore, the type, quantity, power consumption, and performance of on-chip integrated resources are important reference factors for determining MCU selection.

　　Table 4 Three product lines of the STM32L0 series

　　3.3 Low power consumption strategy of MCU system

　　The MCU series with Cortex M0+ core combines low power consumption, high performance and flexible sleep mode, providing an excellent platform and electrical foundation for the development of wearable medical devices. However, how to minimize the overall average power consumption of tasks while maintaining high performance will be an important task for device developers. The low power consumption strategy of the MCU system determines the performance and battery life of the device. The formulation of the strategy needs to start from the following four aspects:

　　(1) Reasonably control the MCU's clock system, select a clock frequency suitable for system operation for specific tasks, and quickly complete complex tasks to gain more sleep time;

　　(2) Select the appropriate sleep mode and sleep time;

　　(3) When entering sleep mode, turn off unused peripherals and clocks;

　　(4) Optimize the task time slice to minimize the average power consumption.

　　Figure 1 shows the low power consumption strategy of the dynamic ECG recorder designed based on the Zero Gecko series in Table 3. The theoretical current consumption of the MCU system task is shown in Figure 2. Among them, the MCU mainly switches between three modes: operation mode 1 (EM0_1), operation mode 2 (EM0_2), and deep sleep mode (EM2). Usually, the MCU works in EM2, the high-frequency clock and peripherals are turned off, and the current consumption is IEM2; when the timer is interrupted, the MCU wakes up from EM2 and enters EM0_1 to run at high speed at f1 main frequency. At this time, the current consumption is IEM0_1, and the A/D is started to sample the ECG signal. After the sampling is completed, the data is temporarily stored in the RAM; if the cached data volume does not reach the threshold, the MCU will directly enter EM2 and wait for a timer; if the cached data volume reaches the threshold, the MCU switches to a higher f2 main frequency to enter EM0_2, and the current consumption reaches IEM0_2 in a short time, and the cached data is processed and stored on the SD card. After the storage is completed, it enters EM2. In the running mode, two different main frequencies f1 and f2 are used, which are determined by the different requirements of the A/D sampling task and the SD card storage task for computing power, so as to optimize the average power consumption of the task.

　　Figure 1 Low power consumption strategy of the Zero Gecko series-based Holter recorder

　　Figure 2 Theoretical current consumption curve of the dynamic electrocardiogram recorder when performing different tasks

　　4 MCU Selection Cases for Wearable Medical Devices

　　Monitoring blood oxygen saturation is an important means to understand the cardiovascular physiological status of the human body. A wristband blood oxygen saturation monitor is designed with the following design goals: based on the measurement method of reflective photoelectric volume pulse wave, to achieve non-invasive and continuous detection of blood oxygen saturation of human arterial blood; to process and analyze the pulse wave signal, and calculate the two important physiological parameters of heart rate and respiratory rate; when the user's blood oxygen saturation or heart rate exceeds the normal predetermined range, an alarm will be automatically sounded.

　　Figure 3 Functional block diagram of wristband blood oxygen saturation monitor

　　The system function is planned according to the design plan and goals. The functional block diagram of the wrist-worn blood oxygen saturation monitor is shown in Figure 3. The special requirements of this device for the MCU are:

　　(1) High energy efficiency, i.e. low operating power consumption, ultra-low sleep power consumption and high computing performance;

　　(2) Low-power ADC, sampling accuracy not less than 10 bits, pulse wave sampling frequency set to 200 Hz;

　　(3) USB controller, which is required to burn programs or communicate with the host through the USB interface.

　　Considering the device's requirements for MCU performance, power consumption, and peripherals, MCU selection can be done in three steps:

　　(1) Based on the analysis of different cores in the previous article, choose the low-power, high-performance Cortex-M0+ core;

　　(2) Based on a horizontal comparison of the Cortex M0+ core MCU series, the STM32L0 series with an integrated low-power 12-bit ADC was selected to meet the needs of long-term sampling;

　　(3) Considering the model with USB controller, STM32L052C8 can be selected as the main controller of the device to achieve the best balance in performance, power consumption, cost and size.

　　In the actual MCU selection, it is necessary to analyze specific issues specifically and make the most appropriate decision based on the actual needs of the existing MCU series and equipment.