NewsCompany newsIndustry News



What are the challenges of implementing import substitution in the six crowded chip start-up sectors?

In recent years, China's high attention to the integrated circuit industry, coupled with the introduction of support policies in various places, huawei and zte events as a catalyst, have made the integrated circuit industry become the focus of entrepreneurship, the start-up of various chip teams become the object of the capital market chase. By the end of 2019, there were 1,780 chip design companies nationwide, up from 736 four years earlier, according to the China semiconductor industry association design branch. But the domestic chip design field has been showing the industry pattern is "a group of small sampans", the lack of "aircraft carrier fleet", especially reflected in a few "entrepreneurship cluster" crowded track. According to the computing, communications, cognitive interaction, modeling, storage to classification, at present the most crowded startup circuit consists of MCU, bluetooth, SSD master, analog chips (power management and the signal chain), rf chip, photoelectric devices and chips these six areas, and these six areas also have market quantity bedding face is wide, the characteristics of domestic occupied low-end, if we can as soon as possible in the field of high-end import substitution, will have a strong effect. This paper will focus on these six chip product areas, sort out the status quo and enterprises, find out the main challenges to improve the localization rate of high-end products and realize import substitution.

In the previous part, we introduced the three main areas of digital chip, namely MCU, bluetooth and SSD.

1, the MCU

MCU microcontroller, the CPU frequency and specifications to do appropriate reduction, and memory, counters, USB, A/D conversion, UART, PLC, DMA and other peripheral interfaces, even LCD driver circuit are integrated into A single chip, chip level computer, for different applications to do different combinations of control.

Localization status:

The annual capacity of China's MCU market reaches about 40 billion yuan, but at present, the mainstream market occupied by domestic MCU is still 8, accounting for about 50%. Limited by embedded Flash technology, IP and simulation technology, domestic 32-bit MCU products in terms of performance and stability, and foreign enterprises still have a certain gap. At present, the total market share of domestic 32-bit MCU is less than 5%, and its products are mainly focused on the middle and low-end fields, while the self-sufficiency rate of high-end fields is almost zero. Most domestic electrical manufacturers and consumer terminal manufacturers preferred ST, NXP and other foreign manufacturers. Those with certain price requirements will choose MCU brands in Japan, South Korea and Taiwan before considering domestic brands. Due to the relatively low performance, most of the domestic MCU products are used in consumer electronics, toys and other fields.

Main challenges of localization:

First, the downstream manufacturers do not have a strong sense of substitution. In high-end sectors such as automotive, home appliances, industrial equipment, although the price of the MCU is very low, but due to the price of the cost of the whole machine of MCU chip used proportion is very small, the domestic MCU and the gap is not enough to drive prices of imported MCU high-end users to choose is not yet mature domestic MCU, so the industry user's attitude toward domestic MCU is very cautious. The domestic MCU has been locked in the low-end field for a long time, continuing to carry out low-level competition among domestic manufacturers, but has been unable to leap to a higher level to compete with imported MCU in the market of higher value.

Second, domestic MCU enterprises have serious shortcomings in ecological construction. With years of accumulation, foreign enterprises have built a complete ecosystem in product series, development tools, evaluation tools, software libraries, application support and know-how, and it is difficult to replace the application side. Most domestic MCU enterprises still stay on the development board, burner and base firmware library, as for the development environment (IDE), RTOS and middleware, still rely on the support of third-party higher level applications. In terms of Internet of things, professional library, industry application and university planning, domestic MCU enterprises are still far from international MCU giants.

Third, domestic MCU enterprises are too keen on quick success and quick profits. Most of the domestic enterprise is difficult to cast off the yoke of the ecological environment of ST MCU, and even many companies product directly was sold on the basis of software and hardware compatibility with foreign companies, to quickly enter the market, fierce competition in the market out of stock and price do this can be achieved under the condition of a certain product substitution, but the long run it will face the chips, hardware and software support in areas such as usurping the risk of intellectual property rights, also easy to cause the industry into a competitive, cannot upgrade to premium.

Fourth, the product diversification degree of domestic MCU manufacturers is not enough. International MCU manufacturers generally conduct business in the mode of IDM integrated component manufacturers. They often sell one MCU and simultaneously sell a number of analog chips, power devices, sensors and other products around the MCU that are far more valuable than the MCU itself, so as to achieve high profits for the company as a whole. But most domestic MCU manufacturers can only rely on MCU's own profits to hematopoietic development.

2, bluetooth

Bluetooth is a high capacity near field wireless digital communication technology standard. It is also the preferred short range wireless technology to connect different devices. With the evolution of bluetooth standard, the application of bluetooth also experiences different scenarios. Before version 4.0, bluetooth was mainly used for computers, mobile phones and other devices as the interface for data and audio transmission. As a standard device for mobile phones, bluetooth has established its dominant position in the Internet ecosystem. Version 4.0/4.2 has driven a huge wave of wearable, smart home and other businesses, and spawned a large number of Internet of things application companies and module companies based on connection and data such as huami. Application scenarios are embodied in various small interconnected devices, smart homes, smart buildings, etc. In addition to the qualitative changes in efficiency and user experience in traditional applications, the 5.0 and 5.1 releases have brought a broader range of applications to bluetooth, including networking and location services. Application scenarios are extended to smart city, industrial interconnection, automobile, medical and other scenarios.


Localization status:

Bluetooth has become the best solution for short-distance connectivity to the Internet of things, and still keeps chips growing at around 400 million a year. Future bluetooth chip shipments will continue to maintain an extraordinary growth trend. It is predicted that the compound growth rate will still reach 8%~12% in 2018~2022, and the growth rate of low-power bluetooth will be 20% ~30%. In 2022, the overall shipment of bluetooth devices will reach over 5.2 billion. But the domestic rate of bluetooth chip is less than 20%. The international big factory is still at the high end, accounting for about 80% of the market, and the gross profit is more than 50%. A few domestic manufacturers have made breakthroughs in a few applications, but the market share is still small, low-end products are homogeneous, price war is obvious, gross profit is low, simple pass-through scenarios are more.

Main challenges of localization:

First, bluetooth chip is a great challenge to the team's comprehensive technical ability. Bluetooth in the field is not particularly complex SoC chip, however, from the perspective of communication chips, bluetooth chip is "the sparrow is small all-sided, important including processor, protocol stack, radio frequency, low power consumption structure, peripherals, etc., of course also includes many other like bus, memory, cache, Flash, the clock all aspects and so on, is a cross power, rf, digital, software, several big complex cooperation in the field of engineering, no step can around in the past, the design capacity is a very big challenge.

Second, low power consumption design requirements are higher and higher. In the future intelligent interconnection era, the biggest bottleneck of wireless connected devices lies in power consumption. Balancing the performance and power consumption of bluetooth chip is very critical, which requires rich experience and profound accumulation. The reduction of bluetooth power consumption is mainly realized through chip design and system design. At the beginning of the design, through the reasonable division of software and hardware, a reasonable low power consumption system scheme is obtained. On this basis, the chip design needs to consider the abnormal power design, power management design, power management design, micro-power watch-circuit design, etc. Specifically, to reduce the power consumption of rf, power management and system control. In terms of system design, it is necessary to develop peripheral software to adapt to hardware, optimize software code to reduce computational complexity, adopt low-power programming and effective peripheral power management design, so as to achieve the optimal balance between product power consumption and performance.

Third, the design capability of analog circuit is the gap of domestic bluetooth chip design. To resist external interference, the chip design of the manufacturer is the first step to ensure the stability of bluetooth chip. In particular, the circuit design of analog signal acquisition, analog-to-digital conversion and rf terminal determines the stability of high-end bluetooth chip products, which requires the company's experienced analog chip engineers to make a reasonable design of chip architecture, so as to maintain the stability of bluetooth transmission signals and optimize bluetooth performance. Therefore, analog circuit design capability is extremely important for high-end bluetooth chip design, which is also an important reason why TI, Dialog and other traditional analog giants have a high market share in high-end bluetooth chip products.

3. SSD master control

SSD can be divided into three parts: master chip, flash memory particles and others. The main control chip is the brain of the SSD, and its function is similar to that of the CPU in the computer. Its function is to properly allocate the load of data on each flash chip, and to take charge of the whole data transfer, connecting flash particles and external interfaces. The performance of different master controls varies greatly. In terms of data processing capability and algorithm, the reading and writing control of flash memory will vary greatly, which will directly lead to a large performance gap in SSD products. Poor quality control will not only affect the product performance, is more likely to be broken than particles earlier, affecting the product service life.

Localization status:

Orders for high-end SSD master chips are concentrated on Marvell and Microchip (acquired through the PMC acquisition) in the us, mainly for industrial and enterprise-class systems. Taiwan's SMI and Phison are the main players in the consumer segment. Among them, huirong's market share is as high as 30%. At present, there are 30 or 40 domestic SSD controller manufacturers, but the technical level is uneven, most of the domestic SSD main control core IP from abroad authorization/purchase use license, no own can modify the source code. Generally speaking, the SSD master in China still has a lot of shortcomings in technology accumulation. Most of the IP is in the hands of others. Therefore, it has no competitiveness in cost and is restricted in technology evolution and development.

Main challenges of localization:

First, SSD master control research and development difficulty gradually increased, the gap in domestic technology level widened. As NAND Flash enters the 3D process and the number of layers increases rapidly, SSD master control design becomes more and more complex and has higher and higher threshold. With the increase of SSD master controller r&d technical difficulties, need to pay attention to high-speed IP interface, NAND flash memory 28 nm particles adaptation technology, and more than 70 million above process logic gate chip design ability and encapsulation testing ability, on domestic cryptographic algorithms SM4 has an ambitious program of activities and performance requirements and system compatibility, high-performance LDPC error correction technology of continuous research and development ability, etc., you also need to consider the adapter firmware development, research and development of large-scale production difficulty such as the overall difficulty, for domestic SSD master pursued advanced manufacturers of products more difficult.

The second is the ability to match with NAND flash memory particles from different manufacturers. For independent SSD master manufacturers, the ability to quickly adapt to the latest NAND particles is critical. The NAND flash particle characteristics of different manufacturers vary greatly, and the NAND particle manufacturing process of the same manufacturer differs greatly. These features of NAND flash memory are only known by NAND granule manufacturers, and many of them involve core technical secrets that NAND granule manufacturers will not disclose to the public. Compared with large manufacturers, independent master control companies lack NANDFLASH, which means they lack the key parameters of the error correction algorithm. The product performance will be affected to a certain extent, and it is difficult to achieve mass production.

Third, capital and talent on the domestic main control of the constraints of manufacturers. The SSD master may contain several cpus, and corresponding SSD management algorithms are required for programmatic management. In firmware, including garbage collection, wear a balanced management, bad block management, data caching, power-fail protection firmware algorithm is of good or bad, directly affect the final development of SSD solution whether can production, is the most important index of the competitive, this requires engineers to write the firmware engineers, algorithm, and the talent in this aspect is lack in our country. In terms of capital, for domestic independent main control companies, almost every customer has to invest a lot of resources for technical support, including NAND adaptation, which makes these enterprises have to bear higher capital pressure.

Part of the content of this paper comes from network information sorting.

Source: Beijing international engineering consulting