Semiconductors in Automotive Industry
For automobiles, the huge push for improvements needs semiconductors. At present, the amount of automotive semiconductors accounts for more than 15% of the total semiconductors market, which is the fourth largest field after communications, computers, and consumer electronics. Automotive semiconductors start from car radios and speakers, to anti-lock braking systems, electric power assist, tire pressure monitoring, navigation, etc., and have been upgraded to the current intelligent driving, etc. With the continuous improvement of modern automobiles, the application of chips is becoming more and more extensive.
As for performance, automotive-grade semiconductors have higher requirements than industrial-grade and consumer-grade semiconductors (get more info from KUKE). In addition to the general specifications of the automotive industry, car semiconductors also have more stringent automotive-level standards. For example, the failure rate requirement is one in a million, while the consumer grade is only one in a thousand. What’s more, its demand is higher than military-grade and aerospace-grade semiconductors. Therefore, automotive semiconductors have the characteristics of high technical barriers and huge demand.
What Semiconductors are Used in Cars?
What should we refer to distinguish automotive chips? According to the different technologies involved, they are mainly divided into power IC, IGBT, CMOS, SOC, etc. According to the classification of devices, it can be divided into MCU, ASIC, ASSP, analog devices, discrete components, memory, micro devices, optoelectronics and sensors, etc.
Table 1: Car Chips Types
| Types |
Main Chips |
Applications |
| Microprocessor |
8-bit, 16-bit, 32-bit, 64-bit |
Engine electronic control system, electric motor control, lighting control system |
| ASIC/ASSP |
Programmable, semi/full-custom integrated circuits |
Assisted driving system, air conditioning control system |
| Analog and Discrete IC |
Power Management ICs, Rectifiers |
Power management system, engine electronic control system, vehicle infotainment system |
| Memory |
ROM、RAM |
Assisted driving system, engine electronic control system, infotainment system |
| Sensor |
Image/acceleration sensor, gyroscope, MEMS sensor |
Assisted driving system, automotive electronic control system, infotainment system |
From the perspective of application fields, the semiconductors of traditional fuel vehicles are mainly concentrated in the electronics fields such as body and chassis safety. With the continuous development of automobile electric intelligence, the IC demand in power, assisted driving, infotainment and other fields is rapidly increasing. For new energy vehicles, power semiconductors are the core components to realize electric energy conversion. The newly added three-electric system (battery, e-drive and e-control), OBC (on-board charger), DC/DC, charging piles, etc. requires a large number of inverters, transformers and converters, so the consumption of power devices such as IGBTs and MOSFETs have increased significantly, and they are power semiconductors.
IGBT
Cooperating with high current and high power, the demand for IGBT of new energy vehicles is strong. For example, IGBT modules have become the standard configuration of inverters in e-control systems to convert DC power into AC power to drive motors. In the on-board OBC, the IGBT rectifies the input AC to DC to charge the new energy power battery, and the on-board air conditioner is also equipped with an IGBT single tube/module. All of above-mentioned ICs include IGBT power switches, DC/DC converters, current sensors, ripple capacitors, and microcontrollers, among which the IGBT cost accounts for as much as 44%. However, only a small number of IGBT single tubes are used for engine ignition in fuel vehicles.
Despite this, the downstream demand for e-vehicles and energy storage has increased rapidly in the past two years, and meanwhile the supply shortage is still affected by factors such as the Covid-19 situation and geopolitical conflicts, production cuts and other factors. Large suppliers Infineon and STMicroelectronics have also become an important factor in the chip shortage due to long deliver times.
ECU
ECU(Electronic Control Unit) is the brain of the car, also known as "driving computer", and MCU is its core, which plays a controlling role. ECU is used to control one or more subsystems of a car. ECU is composed of microcontroller (MCU), memory (ROM, RAM), input/output interface (I/O), analog-to-digital converter (A/D), and drivers. The role of it is to monitor various vehicle operating data (such as braking, shifting, speed, heading angle, position, etc.) The pre-designed program logic calculates the data sent by various sensors, and after that, re-sends each parameter to each related actuator to execute various predetermined control functions.
Under the trend of electric intelligence, the number of traditional distributed EE architecture ECUs is increasing, and the system is complex. Bosch's five domain divisions came into being at this time. With the development of automobile intelligence and informatization, more and more ECU chips are used. They are connected together through CAN and LIN buses. However, the increase of ECUs makes it difficult to arrange wiring harnesses, software maintenance and upgrades, and low efficiency of information communication between modules.
MCU
Microcontrollers (MCU) are widely used in many car semiconductor fields such as instruments, anti-theft devices, chargers, tire pressure gauges, hygrometers, and sensors. At present, the average number of bicycles equipped with more than 20, mainly divided into the following three categories:
8-bit MCU: It is mainly used in lower-level system controls such as air conditioners, wipers, sunroofs, window lifts, low-level instrument panels, junction boxes, seats, and door control modules.
16-bit MCU: It is mainly used in the control of chassis systems such as engines, gears and clutches, electronic turbine systems, power transmission systems, suspension systems, steering wheels, torque, and electronic brakes.
32-bit MCUs: It is mainly used in the control of dashboard, body, multimedia information system, engine, intelligent real-time safety system, power system and X-by-wire system, etc.
Analog Chips
In automotive applications, analog chips play an auxiliary role as a bridge and power supply, covering every corner of the five domains of the vehicle. Take some subdivision modules in the five domains as examples to illustrate the use of analog products.
1) Chassis domain: Taking the body dynamic stability module (i.e. ABS/VSC) as an example, the analog chips that need to be used include driver chips, interface chips, etc.
2) Body domain: Taking the interior light module as an example, the required analog chips include LDO, interface chips, etc.
3) Power domain: Taking dual-clutch transmission as an example, the required analog chips include interface chips, PMIC chips, analog switches, etc.
4) ADAS domain: Just care about the telematics control unit.
5) Smart cockpit domain: Taking the instrument panel as an example, the analog devices used include LDO, PMIC, interface, surge protection, etc.
Take DC-DC and gate drive as an example: 1) DC-DC application parts are all over the car, and the increase mainly comes from the low-voltage scene in the infotainment field and the high-voltage scene in the power field. 2) Gate drive Let us take the body domain as an example. Various applications involving motors, such as seat position adjustment, require the participation of gate drive to play a driving role.
SoC Chips
The self-driving SoC chip is based on CPU+XPU to meet the computing requirements. Processor chips can be divided into general-purpose and special-purpose. The former includes CPU, GPU, DSP, etc., and the latter includes FPGA, ASIC, etc. As the computing and control core unit of the SoC chip, the CPU has strong scheduling, management, and coordination capabilities, but it is not good at handling calculations repeatedly, and it is difficult to meet the high computing requirements of the autonomous driving SoC chip. Therefore, the SoC chip requires one or more Multiple XPUs (GPU/FPGA/ASIC) to have AI calculations.
GPU, FPGA, and ASIC have their own advantages and disadvantages
1) GPU: It is good at image processing, supports a large amount of data parallel computing, but has weak management and control capabilities, high power consumption.
2) FPGA: It belongs to semi-customized chips, its energy efficiency ratio is better than CPU and GPU. It has the advantages of short cycle time and low cost of trial and error in the early stage, but it does not have cost advantages after mass production.
3) ASIC: a dedicated chip designed based on a specific algorithm, which needs to be redesigned after the algorithm is iterated, which may lead to high sunk costs and a long development cycle. Long, but the computing power and energy efficiency ratio are high, and the cost after mass production is lower than that of FPGA.
The data processing of the smart cockpit is more complex, and the SoC chip is the core computing unit. The smart cockpit covers multiple parts such as vehicle infotainment system, human-computer interaction module, car networking module, and HUD system. At this stage, multi-modal interaction is a trend of smart cockpits, that is, on the basis of in-car and out-of-car perception, the effective integration of voice, emotion, gesture, face recognition, positioning and other functions can be realized through multi-modal technology. In this evolution process, the number of cameras in the car has increased, also their resolution has been improved. What’s more, 3D information introduction, model optimization, and the upgrade of various software, hardware, and algorithm have led to a significant increase in the complexity of data processing. So a single ECU operation is not powerful enough. In short, a system SoC chip integrating multiple processors such as CPU, GPU, and NPU is required for data processing operations.
Traditional manufacturers of smart cockpit SoC chips: Renesas, NXP, TI, etc dominate the automotive MCU and ECU chip market, which have rich experience in automotive-grade chips, but are weak in innovation during product development iterations. Their product mainly used in mid-to-low-end car models, which occupy a large market share at present. Compared with former manufacturers, when it comes to consumer-grade chip manufacturers: Qualcomm, Samsung, Intel, etc., have natural advantages in the field of high-computing, advanced-process automotive-grade chips. Their software service ecosystem is also better, and the product iteration speed is fast in high-end car models.
Table 2: Smart Cockpit Chip Products & Suppliers
| Band |
Product |
NPU |
GPU |
CPU Process |
Time to Volume |
| Qualcomm |
820A |
4 TOPS |
320 GFLOPS |
14 nm |
Yes |
| Qualcomm |
8155 |
10 TOPS |
1142 GFLOPS |
7 nm |
Yes |
| Qualcomm |
8295 |
30 TOPS |
1720 GFLOPS |
5 nm |
2023
|
| Samsung |
Exynos V910 |
5.7 TOPS |
1205 GFLOPS |
8 nm |
Yes |
| Samsung |
Exynos V920 |
30 TOPS |
/ |
4 nm |
Yes |
| Intel |
A3960 |
/ |
216 GFLOPS |
14 nm |
Yes |
| MediaTek |
MT2712 |
/ |
133 GFLOPS |
28 nm |
Yes |
| Renesas |
R-Car H3 |
/ |
288 GFLOPS |
16 nm |
Yes |
| NXP |
i.MX 8M |
/ |
128 GFLOPS |
16nm |
Yes |
| Texas Instruments |
Jacinto 7 |
8 TOPS |
166.4 GFLOPS |
16 nm |
Yes |
| Huawei |
Kirin 990A |
3.5 TOPS |
640 GFLOPS |
7 nm |
Yes |
| Allwinner Technology |
T7 |
/ |
/ |
28 nm |
Yes |
CIS
As a mainstream imaging chip, CIS(CMOS image sensor) will benefit the high accuracy of vehicle cameras. The main hardware structure of the car camera includes optical lens, image sensor, signal processor, serializer, connector and other devices. From the perspective of cost structure, the image sensor accounts for about 52%, which is the most important part of the vehicle camera module. In addition, the lens group and module packaging account for 20% and 19% respectively. Image sensors are mainly divided into two categories: CCD and CMOS. At present, CIS accounts for more than 90% market share for vehicle cameras.
On-board memory chips are widely distributed, and DRAM and NAND are the mainstream products. Vehicle-mounted memory chips are distributed in automobile body domain, chassis domain, cockpit domain, power domain, and automatic driving domain, and support the storage functions of applications such as ADAS, IVI, instrument panel, interconnection, and black box. In addition to carrying the system alone, the memory chip is also packaged inside various main control chips (MCU, SoC) for caching, reading and processing data information to improve the efficiency. Automotive memory chips are divided into volatile type (including DRAM and SRAM) and non-volatile type (including NAND, NOR, and EEPROM).
Under the trend of intelligent networking, massive data puts forward higher requirements for on-board storage. Under the trend of automobile intelligence and networking, the application of ADAS system, smart cockpit, and Internet of Vehicles technology will generate a large amount of data, which will put forward higher requirements for on-board storage. Taking the ADAS system as an example, the research and development of the ADAS platform needs to collect a large amount of road test data when the vehicle is driving, including camera, radar, GPS and other data. On the verification and simulation, the whole process will generate a large amount of process data. The road test of L2 vehicles will generate about 2TB of data in one hour, and the data volume of L4-L5 road tests per hour will reach 16-20TB, and a single road test will generate 8-60TB of data, so in the entire development cycle will reach EB level. The caching, reading and processing of massive data will put forward higher requirements for the read and write performance, capacity and reliability of the storage system, and the demand for on-board memory chips will therefore increase rapidly.
Top 10 Automotive Semiconductor Companies
* Infineon Technologies (1999-, German)
* NXP Semiconductor (2006-,Dutch)
* Renesas Technology (2002-, Japan)
* Texas Instruments (United States)
* STMicroelectronics (1987-, Dutch)
* Bosch (German)
* Analog Devices (1965-, United States)
* Microchip Technology Inc. (1989-, United States)
* ROHM (1940-, Japan)
Chips in Future Automotive Components
Development Prospects
Since the development of automotive semiconductors, 2021 is the best year ever for automotive chips. First of all, the demand for chips has increased significantly stimulated by automotive electrification, intelligence, and networking. Second, due to the production capacity can’t meet demand, tight supply and demand have led to high chip prices. In addition, according to statistics, currently about 25% of the global analog chip market is used in vehicles. According to requirements, analog chips can be divided into two categories: power management and signal chain. On the one hand, in the high-voltage working environment brought about by the electrification of automobiles, so power management chips used to realize power distribution and control are in the first place; on the other hand, signal chain chips connect the real and digital worlds, which are the basis for electronic systems to achieve automation and intelligence.
Senior automotive semiconductor companies still dominate the main market. With stable products and high market awareness, they may be one or several parts in the automotive semiconductor industry chain, for example, TI and ADI occupy half of the global market, but also there are opportunities in subdivided fields.
Overview of Automotive Semiconductor Industry Chain
Automotive Semiconductor Product Chain List
1) Upstream Manufacturing
Chip design
Wafer Fabrication
Packaging inspection
2) Midstream Devices
MCU
Analog IC
Logic IC
Memory
Diode
Transistor
Thyristor
Sensor
Phoelectron
3) Downstream Products
Radar
Electronic control unit
Vehicle controller
Camera
Assisted driving
The higher the intelligent level of vehicles, the more control chips are required, and the larger the capacity of on-board storage, the demand for automotive semiconductors has surged. That is, the incremental market for automotive semiconductors has opened. In the past, the models of many companies started to expand from a single category, but from the above table and analysis, we can see that newcomers in the industry should combine their own advantages and exploit the gap between upstream and downstream to open up a broader space for growth.
English
