Electronic Fans Network Report (by Li Wanwan) With the continuous advancement of technology, there will be more and more cars with various levels of automatic driving functions. The higher the level of automatic driving of cars, the more demand for cameras in vehicles. , The demand for CMOS image sensor (CIS) is also increasing. Liu Qi, general manager of the automotive electronics business department of Haowei Technology China, introduced in a forum a few days ago how many image sensors a self-driving car needs, and roughly calculated based on the average selling price of image sensors and the annual global car production How big is the future automotive image sensor market?
Liu Qi said that the company discovered from the exchanges with a large number of domestic and foreign vehicle manufacturers that when the car reaches the level of L4 or L5 autonomous driving, the demand for the camera of the vehicle is mainly reflected in these aspects: the front view needs 1-3 eyes, mainly It is 1 eye, some high-end cars or vehicles with more advanced automatic driving functions may need 3 eyes; lateral perception requires 2-4 eyes; backward perception requires 1 eye; four surround view and APA (Automatic Parking Assist System) require 4 In the cabin, driver monitoring requires 1-2 eyes; in the future, a 1-eye camera may be needed to understand the status of passengers, including whether they have any problems after getting off the bus; in addition, according to national regulations, starting from 2021, vehicles must be mandatory to install driving records Instrument or event recorder, 1 mesh is required.
calculated in this way, it is estimated that every car in the future will basically need to be equipped with 11-15 eye cameras. Looking at it again, the annual global automobile production is about 80 million to 100 million, so the global automotive camera market is about 1.5 to 1.8 billion. The average selling price of automotive image sensors is US$4-5. In this case, the global automotive image sensor market will be close to US$10 billion.
According to data from the Japanese research company TSR, the global CIS market will grow from 17.2 billion US dollars in 2019 to 27 billion US dollars in 2023. Among them, the automotive CIS market will surpass consumer applications in the future to become the second largest CIS application market after mobile phone applications.
This is why many image sensor manufacturers and even some start-up companies want to enter this market.
Compared with mobile phones, what are the stringent requirements for automotive image sensors?
Automotive digital imaging technology has been widely used in the field of ADAS (advanced assisted driving system), but unlike the imaging of mobile phone cameras and digital cameras, automotive cameras usually need to face more strict condition.
On Semiconductor Application Engineering Manager Chen Li mentioned in a seminar recently that there are four challenges in the automotive imaging world: one is that the dynamic range of automotive image sensors must exceed 120dB, usually between 120dB-140dB; the other is temperature. The range is extremely demanding, and it needs to be able to reach minus 40°C and 105°C; third, there are relatively high requirements for low-lights, pedestrians or cyclists need to be detected on the streets at night, and almost in line with the European New Car Assessment (NCAP) ; Fourth, it is also necessary to solve the problems of traffic signal lights, LED street signs flickering and artifacts.
High Dynamic Range (HDR) challenge. What is dynamic range, that is, in the same scene, there are both low-light areas and high-light areas. The ratio of high-light to low-light is defined as the dynamic range of the scene. For images, it needs to be able to Both low and bright areas are shown.
For example, as shown in the figure above, for a vehicle-mounted camera, it is necessary to see the information inside and outside the tunnel at the same time. Obviously, the sensor on the left cannot see information outside the tunnel because of its low dynamic range.
LED lights flicker. LED lights are more and more widely used in daily life. LED lights are AC pulsed light sources whose frequency and duty cycle are variable. The lower the pulse frequency of LED lights, the smaller the duty cycle and the shorter the exposure time. The flicker problem will be more serious, because the image sensor is more likely to catch the LED off state when the frequency is low. The
LED light on and off frequency cannot be distinguished by the human eye, but the image sensor can. When the pixel is exposed to light, it will actually record the LED light off and on state, which reflects the objective world. The image seen by the human brain is supplemented by the brain The images are all lit. The requirement of the car is that the image output by the sensor must be consistent with the image seen by the human eye. This is the challenge of the LED light flashing.
In the actual scene, the LED lights blinkAnd high dynamic range often appear at the same time. For example, in the streets at night, many car headlights and advertising lights are LED lights. Sometimes LED lights are very bright and the contrast between light and dark is very sharp, which also brings high dynamic range requirements.
Faced with these problems and challenges, various image sensor manufacturers need to find corresponding technologies to solve them. The manufacturers that can solve the problems and challenges of automotive image sensors faster and better will gain market recognition and gain more market share. .
The global automotive image sensor market pattern
is currently on a global scale. Manufacturers that can provide automotive image sensors include ON Semiconductor, Howe Technology, Sony, Samsung, Toshiba, BYD Semiconductor, and Siteway. Among them, ON Semiconductor is the largest supplier in the automotive field, with a market share of more than 60% of automotive image sensors, followed by Howe Technology, which accounts for 20% of the market share. Sony entered the market late and currently has a market share of about 3%. Toshiba, Samsung, BYD Semiconductor has small shipments, and Siteway is running into the market.
ON Semiconductor is the world’s largest supplier of automotive image sensors. On Semiconductor Application Engineering Manager Chen Li said at a seminar recently, “Our company’s global market share of automotive image sensors exceeds 60%. We will deliver image sensors in 2019. About 100 million.”
ON Semiconductor can have such excellent results in the field of image sensors, thanks to the company’s three previous acquisitions. First, it acquired Cypress’s CMOS image sensor business unit for US$31.4 million in 2011, followed by In 2014, it acquired the image sensor equipment manufacturer TRUESENSE for US$92 million in cash. In the same period, it also acquired the first automotive CMOS image sensor provider Aptina Imaging for approximately US$400 million in 2014. The three acquisitions established and consolidated ON Semiconductor’s Market position in the field of automotive image sensors.
It can be seen from the company's official website that ON Semiconductor can provide about 104 models of CMOS image sensor products, with a rich product portfolio that can meet the different needs of users. At the same time, in addition to a wide range of CMOS image sensors, ON Semiconductor is still In 2019, the product line was expanded to the millimeter wave radar and lidar market. Yan Kai, Industrial Marketing Manager of the Intelligent Perception Department of ON Semiconductor Greater China, believes that the history of diversified development is a necessary condition for stable growth of an enterprise.
In addition, companies also need to constantly solve industry problems and update technologies to continue to be recognized by the market. For example, for the challenges of high dynamic range and LED flicker in the automotive field, ON Semiconductor has also launched a new generation of Hayabusa platform, using super exposure technology , Broaden the dynamic range and solve the problem of LED flicker.
compares the LED flicker conditions in the two working modes. The above is the traditional Sensor working mode, which cannot suppress the LED flicker. It truly reflects the LED flicker status, so the 100 speed limit mark on the picture cannot be accurately identified, the following is super The exposure technology comes with an anti-LED flicker working mode, and the image recorded by the Sensor is similar to the image seen by the human eye.
Let’s look at the two mainstream technical routes on the market to achieve wide dynamics. One is time-domain multiple exposure fusion, and the other is large and small pixel fusion. The essence of the two technical routes is the same. They both take two or three photos and then merge them into one photo. The sensitivity is different between different exposures, so different brightness is collected to realize the expansion of the dynamic range. In fact, both of these two mainstream technical routes have problems. There is only one type of pixel for multiple exposures in the time domain, and multiple exposures have a lag in time, which will cause the problem of motion ghosts. The number of large and small pixel fusion exposures is only one, but There are two sizes. Because of the crosstalk between pixels and the difference in sensitivity, its wide dynamic and low-light sensitivity will have some limitations and more problems.
ON Semiconductor's new high dynamic range pixel technology video recording, namely super exposure technology, one exposure is equivalent to the dynamic range of the past two exposures. This super exposure technology can perfectly solve the previous time domain multiple exposure and large and small pixel technologies. Chen Li said, “The brand name of ON Semiconductor’s first-generation super-exposure technology is Hayabusa.” The
super-exposure technology is different from the design concept of traditional Sensors. The photosensitive area of traditional Sensor pixels is also a storage area.The innovation of Hayabusa technology is to separate the light-sensing function of the pixel from the storage function. The pixel is mainly responsible for light-sensing and has a small capacity. When it is saturated, the accumulated excess charge will be transferred to the external storage capacitor. In this way, Release the limitation of pixel capacity. Chen Li said, "The equivalent pixel capacity of the first-generation Hayabusa technology platform is 5 times that of the traditional pixel capacity, and the next-generation equivalent exposure technology can be expanded to several tens of times or even higher."
Hayabusa image sensor platform is used in automobiles Vision and perception system solutions currently include four products: AR0147 and AS0149, AR0233, AR0323, ranging from 1.3 million pixels to 3.5 million pixels. These four sensors have common architecture and pixel performance, the same dynamic range and LED flicker suppression capabilities. In addition to providing customers with a variety of options, platform-based products also have a very important feature. Since the performance of the four Sensors is the same, when customers use one of them for algorithm learning and training, if they change to other models, whether it is Need to reduce costs or upgrade performance, the workload of algorithm learning will drop a lot, and many system parameters can be reused.
Howe Technology
Howe Technology is the world's second largest supplier of automotive image sensors. "We have been working in the automotive industry for 15 years, and the total shipments in the automotive field have exceeded 800 million." Liu Qi introduced at a forum a few days ago, "Howe began mass production of the first automotive image sensor in 2005. It has played a guiding role in some new technologies in the automotive industry. In 2009, we mass-produced the first-generation megapixel technology, and in 2018, we mass-produced the first-generation deep-field technology. This technology can improve the single-pixel sensitivity of the image sensor itself. "Z1z
Howe Technology has a major advantage in the field of image sensors is that the company has a deep layout in various fields such as mobile phones, security and automobiles. Electronic enthusiasts previously wrote in an interview with Wang Song, Senior Vice President of Howe Technology China, “In addition to mobile phones, automobiles and security are the other two growth points of image sensors. Technologies are learning from each other and merging. For Haowei, which spans the three fields of mobile phones, security, and automobiles, different products and technologies are more interrelated and common, and this advantage will be more obvious in the future."
In December, Howe Technology released two new automotive-grade image sensor platforms-8-megapixel front-facing cameras OX08A and OX08B. When the new product was released, Howe stated that the OX08A HD camera provides the industry’s best high dynamic range (HDR), and the integrated HALE (HDR and LFM engine) synthesis algorithm integrated on the chip, the pin-compatible OX08B is LED flicker suppression ( LFM) performance sets a new benchmark.
As mentioned above, high dynamic range and LED suppression capabilities are required by cars. ON Semiconductor has introduced super exposure technology for this purpose. Howe Technology also uses its own unique DCG technology to increase the dynamic range of image sensors. Howe said that clarity and dynamic range, as well as integrated ASIL-C functional safety, are critical for forward-looking vehicle applications such as advanced driver assistance systems (ADAS) and Level 3+ autonomous driving, because it can be used in various Accurate detection of humans and objects in the distance under light conditions is directly related to driving safety.
In addition to the two new products launched at the end of last year, Howe also launched the world's first automotive wafer-level camera on June 2 last month-OVM9284, which is a megapixel CameraCubeChip automotive wafer-level camera Module, this product has the lowest power consumption in automotive camera modules-more than 50% lower than the power consumption of neighboring competitors.
According to reports, OVM9284 integrates Howe Technology’s image sensor, signal processor and wafer-level optics in a single compact package, thereby reducing the complexity of dealing with multiple suppliers, and speeding up development time while increasing Improved supply reliability. In addition, unlike traditional cameras, all CameraCubeChip modules are reflowable. This means that they can be mounted on the printed circuit board simultaneously with other components using automatic surface mount assembly equipment, thereby reducing assembly costs at the same time.Improved quality at the time.
Currently, samples of the OVM9284 module are available now, and mass production is expected in the fourth quarter of 2020.
Sony
Sony’s market share in the global image sensor field exceeds 50%, but its market share in vehicle image sensors is only 3%. Public information shows that Sony announced its entry into the vehicle image sensor market in 2015. You must know the global market. Howe, which ranks second, launched its first automotive image sensor in 2005, 10 years earlier than Sony. Although
has been six years since the announcement of its entry into the automotive field, at present, Sony, which has advanced image sensor technology, has not developed as smoothly as expected in the field of automotive image sensors. From Sony’s performance in the last fiscal year It is not difficult to see that the operating income of on-board sensors only accounts for a few percentage points of the overall sensor department, which seems to be only used by Toyota's Lexus and others.
Despite this, Sony still has its own expectations and thoughts on the automotive image sensor market. Sony President Kenichiro Yoshida said before, “The main trend in the past 10 years has been mobile terminals. In the future, it will be mobile travel. In the era of interconnection, automation, sharing, and electric power, the sensors that play the role of the car’s eyes will be the most important parts of the car.” Terushi Shimizu, head of Sony’s semiconductor department at
, said, “We don’t want to be involved in simple things. Price war', our technology is better, so we hope that our sensors can be used, nothing more." In the long run, after the development of true driverless cars, Sony hopes to expand its entertainment business by providing video games, movies and music to users in the car.
Although Sony’s current sales of automotive sensors are still small, some analysts have found that Sony is making progress in the automotive field. In 2018, Toyota began to use Sony’s image sensors in its advanced driver assistance systems, including BMW and Mercedes. European automakers including S.A. and Audi use Sony's image sensors to test their self-driving cars. Sony is also actively negotiating with emerging automakers such as Tesla, which plans to produce self-driving cars that rely only on cameras and radar. Kun Soo Lee, senior principal analyst at
IHS Markit in Tokyo, said: "Sony has the technology and production capacity to replicate its high market share in the mobile market (image sensor)."
Smartway
Smartway in the image sensor field The performance is also very eye-catching. The global market share in the security field has been ranked first for many years. In March 2020, Ms. Chris Yiu, Chief Marketing Officer of Siteway Technology, said in an interview with electronic enthusiasts, "The company will continue to consolidate the security field in 2020. In order to achieve further growth, we will continue to expand other application areas, focusing on automotive electronics, to achieve further growth."
Recently, we have seen many actions of Smartway in the automotive field. In mid-June, Smartway announced its success. After acquiring Shenzhen Allchip Microelectronics Co., Ltd. (Allchip), Siteway said that this acquisition will further expand Siteway's product line and accelerate the company's business layout in the automotive field. It is necessary to know that Anxin Microelectronics has made great achievements in car camera CMOS image sensor products. It is reported that its chip car camera and other miniaturized video surveillance applications are very competitive, and it has a variety of SOC series images independently developed and mass-produced. Sensor products.
In June this year, Siteway also joined a general-Ouyang Jian, as the deputy general manager, responsible for the company's strategic resource integration and the construction of the upstream and downstream ecological cooperation system of the industry. According to the data, Ouyang Jian was as early as 20 years He began to devote himself to the field of digital imaging before, and has been cultivating in this field for many years. At a forum not long ago, Ouyang Jian said that Siteway’s consistent advantages in the security field can bring better night vision imaging performance to in-vehicle products, combined with a number of in-vehicle electronics related technologies that we have released, such as LED Flicker suppression technology, PixGain HDR technology, etc. have laid a solid foundation for us to enter the field of automotive electronics.
can foresee that Siteway's development in the field of automotive image sensors is worth looking forward to.
Summary
In addition to the above-mentioned ON Semiconductor, Howe Technology, Sony and Smartway, in the field of automotive image sensors, Toshiba,Samsung and BYD Semiconductor both ship products.
In the domestic image sensor field, Geke Micro's performance is also relatively prominent. Recently, reports show that Geke Micro's image sensor shipments have ranked second in the world. According to the report, according to Frost&Sullivan statistics, according to the shipment caliber, Gekewei achieved 1.31 billion CMOS image sensor shipments in 2019, occupying 20.7% of the global market share and ranking second in the industry.
It is also worth noting that Gekewei is also the only image sensor company in China that, in addition to designing its own chips, also invests in building factories on its own. In March this year, Gekewei announced the investment in the construction of 12-inch CIS integrated circuits in the Shanghai Lingang New Area. The special process research and development and industrialization project, the project started on July 7, and the application of Geke Micro Technology Innovation Board was accepted on the same day.
has also reported that Gekewei has plans to enter the field of automotive image sensors, but has not seen relevant actions and progress recently.
In addition, more intelligent image sensors will also be the future development direction. Sony has recently launched image sensors with AI processing functions, but these two products will be used in retail and industrial equipment. As the future automotive market becomes more intelligent Demand, it is not impossible for Sony to launch an image sensor with AI function for the automotive field. In fact, Howe Technology has already used its AI technology to improve the wide dynamic performance of automotive image sensors.
