With the entry of new players, how can the dominant automotive CIS manufacturers maintain their leading position?

In the past decade, CMOS image sensor (hereinafter referred to as CIS) is almost one of the fastest growing semiconductor products. Image sensor sales doubled from 2010 to 2019, according to ICInsights.

In the CIS market, in recent years, the growth rate of the automotive application segment has been evident due to the increase in the number of CISs that are equipped on average in bicycles with the increasing demand for automotive assisted driving and DMS driving records. In 2018-2020, the total sales of automotive CIS increased from 6% to 14%.

In recent years, as the growth of smartphones has slowed, automobiles have become a new growth point for CIS. Although Sony is a well-deserved overlord in the entire CIS market, its main revenue comes from the smartphone market. Coming to the market segment of automotive CIS, the market share of CIS manufacturers is another scenario. Sony and Samsung, the top two mobile phone CIS companies, do not have a high share in the automotive field. According to public information, Sony did not announce its entry into the vehicle-mounted CIS market until 2015, obviously starting late.

So, why is the market model of automotive CIS so different from smartphone CIS? What are the unique advantages of automotive CIS manufacturers? With these questions in mind, the Electronic Enthusiasts Network interviewed three automotive CIS manufacturers for some answers.

With similar technologies and different routes, vehicle CIS has put forward more comprehensive requirements.

Compared with mobile phones, security and other fields, the demand for CIS in automobiles is more diverse. In an interview, a OmniVision source told electronics enthusiasts that automotive is a more comprehensive application requirement. The scene used by automotive CIS. Applications are more diverse. For example, mobile phone CIS is mainly used for shooting, but for car CIS, except for reversing images. In addition to driving records, machine vision, such as ADAS, is also required. In addition, automotive CIS also needs to cover various scenes and lighting conditions, so it is a more comprehensive requirement for CIS.

Huang Chunhua, director of strategic marketing of the automotive perception department of ON Semiconductor's intelligent perception department, also has a similar view. He believes that automotive image sensors have higher requirements for safety and reliability, and the image quality is low in illumination. Dynamic range and functional safety are also higher than phone safety. Simply put, mobile phone security focuses on the visual perception of the human eye, and automotive applications pay more attention to machine vision.

Specifically, for ADAS, because ADAS is very important for the judgment of real-time images, there must be both WDR and motion artifacts. At the same time, there will be a lot of LED lights on the road. Due to how LEDs work, images formed during shooting may be lost, flicker, etc. The solution of LED flicker suppression (LFM) is also related to the realization of wide dynamic.

Relevant persons of Gekewei introduced to reporters that vehicle-mounted products have special use environments, so they also have special requirements, such as high temperature resistance, continuous work in high temperature weather; night driving requires shooting in a dark environment; high dynamic performance, enabling access to tunnels and other rapid environmental changes with better imaging quality. The product should be able to solve the problems of noise, black spots, photosensitivity and stability in special environments, and improve the imaging quality of the sensor.

At present, the application of automatic driving mainly requires rolling shutter with high dynamic and low illumination. According to Huang Chunhua, ON Semiconductor's latest AR0820AT, 8.3 million pixels, provides a high dynamic range of 140dB, up to 185 meters long, suitable for the most stringent lighting conditions, and can be safely and reliably used in autonomous driving applications. NIR+ technology enables the enhancement of near-infrared imaging, providing similar best-in-class low-light performance and excellent performance at night.

In addition, on this basis, CIS will also consider the security of network data, screen projection, machine vision requirements, multiple output, etc. in ADAS applications.

Aside from some technical differences, the vehicle rule requirements for a vehicle CIS are actually different from mobile phones. Security is different, and the entire system is different. Coway said: For Coway, vehicle CIS is a separate team and a separate process from product definition to development, production, testing, after-sales, and sales teams, with Tier1. OEM customers together to provide users with better services and create greater value.

Since automobiles have certain requirements for functional safety, it is necessary to estimate possible problems at the beginning of product design. Failure rate, consider how to reduce the probability, how to eliminate the failure.

Huang Chunhua also said that the factory standard of ON Semiconductor is 00PPM. In order to make the intelligent perception technology of car cameras absolutely reliable, ON Semiconductor has analyzed more than 4,000 failure modes to ensure absolute reliability.

However, despite the different needs, many technologies are interrelated compared to security and cell phones. Gekewei, who started to deploy non-mobile phone image sensor business as early as 2014, said that due to the sufficient reserves of CIS technology, the expansion from mobile phones to vehicles and other non-mobile phone fields is relatively smooth.

When discussing the similarities between automotive CIS and mobile phones, security and other fields, Howe told reporters: For example, the requirement for dark effects. Requirements for small pixels and so on, cars and mobile phones. The safety is similar, but the automotive CIS is more comprehensive and requires perfect technical reserves.

For example, CIS on a mobile phone usually has 100 million pixels, and the pixels are getting smaller and smaller, and the line is very aggressive. However, in an automobile, the size of the pixels, the resolution requirements, the dark-state performance, and even the size of the entire solution need to be comprehensively considered. Howe believes: At present, the mainstream ADAS system has reached 8 million pixels, and it may increase to 12-15MP in the future, but it will not endlessly pursue resolution like mobile phones.

With the entry of new players, how can the dominant automotive CIS manufacturers maintain their leading position?

In recent years, major CIS manufacturers have increased their investment in automobiles due to the growing demand in the automobile market. For example, Samsung started to enter automotive CIS at the beginning of this year, and Sony showed a Vision-s concept car at last year's CES, equipped with its own various CIS solutions, to show its determination in the automotive market.

Compared with other competitors, ON Semiconductor and OmniVision, which currently occupy the top two positions in the automotive CIS market, have long been deployed in automotive CIS. For example, ON Semiconductor's intelligent perception department was mainly continued through the acquisition of APTINA. APTINA launched the world's first automotive AECQ100 certified CMOS image sensor in 2005. OmniVision also launched its first automotive CIS product in 2005.

As for its advantages in the automotive sensor market, Huang Chunhua said that ON Semiconductor's image sensors occupy more than 80% of the market share in the field of advanced driver assistance systems (ADAS) and autonomous driving, and have the smallest share and smallest packaging in the product. The advantages of energy efficiency, weight and performance, with the technical advantages of high dynamic range, global shutter, compliance with vehicle safety integrity levels.

It can be seen that ON Semiconductor has been deeply involved in the automotive market, and accumulating user experience in automotive applications and product reliability are the main advantages for them to stay ahead.

From the perspective of the automotive CIS market, although it is currently developing rapidly, the absolute number is far from mobile phones and security.

Howe revealed: From the perspective of technical investment, the most basic technologies of CIS in the fields of mobile phones, security and automobiles are similar. Technological investments in the field of mobile phones and security can be quickly applied to the automotive field.

Compared with the later challengers, Gekewei also had a certain position in the field of mobile phone CIS. It is precisely because of the technology accumulation in the field of mobile phone CIS that Gekewei has more confidence in the automotive sensor market. Gekewei told reporters: At present, the company's CMOS image sensor products are mainly used in the field of mobile phones, occupying a major position in the mobile phone sub-camera market. With our self-developed patented technology and process design, ADRF6755ACPZ sensors for the automotive market are mainly concentrated on 1.3 million and 8 million pixel products. Gekewei can quickly expand to the automotive sensor market based on its experience and technical record in mobile CIS. In addition, the BSI fab we will build in the future can make our research and development faster, thereby accelerating the research and development of non-mobile phone products and keeping up with the changes in the market.

On the other hand, due to the relatively large number of mobile phones and security, the resource guarantee of the upstream supply chain will be more advantageous.

Due to the long cycle of the automotive industry, the competition model of the automotive CIS market in the future - any new player entering the automotive industry needs to face a relatively high threshold, so there must be long-term planning. Long-term technology investments and product planning can be market-based.