Addressing the media late last month, Davide Bruno in charge of MEMS and sensors, Asia Pacific for Marketing Application team, STMicroelectronics shares about today’s reality for ST in automotive application for MEMS and sensors and what the future will be like in automotive for these kinds of products. Bruno covered the topic in three parts viz. opportunities in the automotive market for sensors, ST’s views on the way in which these sensors will be adopted in the automotive market along with a review of the market trends.
MEMS & Sensors in automotive application
Automotive market evolution: The 2019 TAM was in the range of $35 billion, which were split between the traditional automotive core electronics (around 65%), and digitalization and electrification (35%). 10~20 years ago, it would probably take 15 or 20 years or even more to change this partitioning. With the current speed of new projects, new developments and new technologies, we will move to more than 60% of digitalization and electrification and less than 40% will be for the traditional automotive core electronics. This will happen just in 3~5 years from now. In the automotive market, ST is already engaged in all of these applications from the well-known car electrification, with SiC device, then ADAS vision, ADAS radar, networking and 5G technology, positioning connectivity.
Due to the pandemic, the automotive market is impacted and there is a sharp drop this year. Car electrification due to the pandemic takes place much faster than before. We foresee that by 2024 the total portion of electric vehicles will be in the range of 37% to 40%. Two new trends are coming in strongly. Shared mobility – More than 10% will be shared mobility, this means traditional automotive model is changing. Many OEM and Tier 1 are rethinking the way their new cars will serve the consumer and the people.
The automotive sensor opportunity
Different kinds of systems and sub-systems including camera, LiDAR, long range radar, short range radar and ultrasound for an automotive are the mega systems, applications and functions. The sensors applied are – motion sensors for embedded safety, environmental sensors and Microphone. Different systems need to communicate with each other. It is about the fusion of all these sub-systems and sensors.
Automotive sensor growth drivers
ST has identified four mega growth drivers:
- Shared mobility and access control: Low power Accelerometer for Passive Key entry, Battery saving detect stationary state, Security detect no move.
- Road Noise cancellation: It is a system with a mix of sensors. ST has accelerometer for detecting the wheel vibration, microphone to sense and capture the acoustic noise inside the cabin. It applies not one microphone but many OEM solutions that consist of 4, 6 and 8 microphones in addition to the 4 accelerometers for the 4 wheels of a car.
- Connected Cars Telematics Proliferation IMU in V2X: ST discusses vehicles V2X/TBOX/CAR ALARM. ST has been a pioneer in these technologies. With automotive TESEO processors ST offers full set of kits for MEMS and inertial module unit. ST offers 6-degree of freedom device called 6DOF in which there are three axis accelerometers and three-axis gyroscope with the full integration into the navigation system. Car alarm has an inclination product. Insurance box can be used in shock detection.
- Driving Assistance: Includes fusion of systems which include radar, LiDAR and camera. In the assisted and advanced automated driving application: first installation of Radars and LiDARs, Camera orientation position monitoring or stabilization. In any LiDAR or Radar system, the proper inclination of the whole system in regards to the car plane is instrumental for the right measurement. Need for inclination sensor is to detect any misalignment both during the installation of the LiDAR/Radar itself and during the use in the normal operation of the car, while driving. Camera used for orientation or stabilization, combined with specific actuator. ST is a leader in the market for these kinds of products and OIS system.
Five levels of vehicle automation
Level 0: No Automation, in which driver fully controls the car. For cars at higher levels, if the driver is driving out of the lane or inside the lane, the car will give them some kind of information. ST believes one third (33%) of the cars produced in 2021 will be at the level 2 and level 3. So this is a big jump from level 1. To move to level 4 and level 5, we need a system roadmap. We need more fusion between different sub-systems including radar, Lidar, and so on. At level 5, the driver does not drive any more, and they just take the car to move from point A to point B without any knowledge about the car itself.
Level 2 started in 2015. The sensors from ultrasonic, LR radar, SR radar and LiDAR enables these functions at different levels. Level 3 starts in 2022 and level 5 in 2040. In Tokyo level 4 and 5 will probably come faster than other regions where the infrastructure is not already there or where the roads are not designed for the fully autonomous cars. In general, by 2040, the majority of cars will be fully automated. To reach level 5, we need more than 30 sensors to work, in intraoperative way and altogether.
The Three Waves: Adoption of Sensors in the car:
It is about what happened in the past few years and what will happen in the future. The 1st wave of sensor adoption in automotive started with the airbag in 1974 (Ford did a first experiment in 1971). We call it active safety. In the second wave car becomes fancier, has big display, the infotainment system took place and several more sensors were used. We call it non-safety application. The 3rd wave starts with active safety, because active safety is part of the autonomous car. It is this wave that we are now designing with all the Tier 1 and OEM car makers. ST is present in many of the first wave applications which include pressure sensor, driver-side airbag, the dual front airbag etc. In the second wave ST is present in the airbag market. For Non-safety, it involves telematics, navigation and passive key element. ST is recognized worldwide as the No.1 maker of non-safety applications.
Microphone for automotive
Microphone is a very interesting product for automotive. There is no automotive standard for microphone. Four kinds of applications drive the demand for microphones in automotive. They are the hands free call, and noise cancellation with a lot of microphones inside the car, the emergency call and the in-car call applications. The MP23DB01HP is the industrial grade product ST offers to partners and customers.
Inertial sensors for safety
ST talks about the penetration rate for inertial sensors for safety from 2018 to 2023. ST is in the airbag market, which is already quite saturated. The growing markets are the roll detection and ADAS/autonomous driving.
ST addresses these two applications at level 5 where we sit inside the car but we do not drive the car. For a sensor we need to understand the sensitivity, accuracy, stability and the linearity. Three important parameters are accuracy, stability and linearity. There is much of difference between a sensor which is an inertial module unit used in a smartphone than the sensor installed in car to achieve level 5 in the autonomous driving. The important lies in accuracy, linearity, and stability. At level 0, the sensor is not used. At level 1, the sensor in 30 sec can give an error of positioning of 50 centimetres. At level 3, 50 centimetres become 10 centimetres; and at level 5, there is no margin for errors, so there is no possibility for mistakes.
Today from level 0 to level 3, most of the OEMs and Tier 1 are using redundancy configuration for the inertial modules or sub-systems. Instead of using only one device, we put together several devices (all same) and operate in the way which can give the minimal error. You need to have software to synchronize all the signals and verify information. The future is to have one single 6-axis or x-axis system, where x can be 4, 5, 6 DOF (degree of freedom), called the functional safety. Depending on element/component of the subsystem, Functional safety is defined different by OEMs. It cannot be at the sensor level, but on the system level. There are 4 level – A, B, C and D for ASIL. The most stringent is ASIL D. Today, most of the 6 axis are ASIL B. The target is to achieve ASIL D, to make the overall system safer and simpler. ST is going to achieve the same in near future.
Inertial module unit:
ESC (electronic stability control) has been installed in our cars for many years now. ESC system corrects driving mistakes for avoiding accidents without being perceived. Today the car makers want to head us to rollover detection and stabilization. ST already offered customers 3 degrees, 4, and 5 degrees of freedom and different combination of DOF to address ESC, Roll over and stabilization.
Another important element of any MEMS sensor in Automotive is packaging. A ceramic package gives stability and better linearity. Ceramic package is the future for automotive for safety. Calibration: One of the questions is the cost. When we talk about the cost of sensors, we need to talk about the time needed to calibrate the sensor itself.
The aviation/airplane is fully autonomous at level 4 or at level 5. The pilot does really nothing. But to enable this function, you need a system that costs $150,000, which will take one week to be calibrated. This sensor is not different from a standard sensor. But the calibration means the sensor plus the ASIC, plus all the firmware, and the way it is calibrated one by one is the key point. The cost is 10,000 dollars. At entry level car, if the inertial module system costs 10,000 dollars, the car can be sold at $12,000 or $15,000. There is a need to bring this cost down. Today, ST sells million pieces in consumer applications, and have sold more than 17 billion sensors in the past year. ST calibrates the device in a few seconds. The consumer device costs is a single digit dollars and below it.
Today ST has the device called ASM330CHH. Its best competitor costs about 10,000 dollars, which requires 1~2 day calibration. For two parameters including offset and sensibility error, ST is already there. For many parameters, ST is even better. So providing affordable sensors for the inertial platform is the challenge.
Sleepiness, overdrunk or texting may be the cause of mortality while driving. This is a completely new family of image sensors with different Mega pixels options. It can recognize the degree of attention, eyes and facial expressions while driving. So when we are not able to control ourselves, the system will enable us to pull over, stop the car, and take a rest.
Key drivers for autonomous driving in the future: LiDAR: is compact, cost effective, not big and it is an enabler for the autonomous driving. ST provides mirrors in LiDAR. This is a real mirror made by semiconductor. It is in a very small form factor. This enables the LiDAR system to be small and cost effective. With semiconductor processor cost projection normally goes down. In the past it happened with gyroscopes, accelerometers and microphones. For the same ST cooperates with Leddar Tech with system made by MEMS, mirror and optics.
Davide Bruno also responded to media questions as below:
ELE Times: How ST is contributing towards the EV market in India and which are the automotive sensors showing strongest growth?
Davide Bruno: Thank you for the question on India market. Regarding the MEMS sensor, there is not much difference from one country to another. India is very aggressive to push the adoption of electric cars. This will boost the overall demand of electric cars, with the local players designing more energy friendly cars. This is an opportunity for ST. We address this one with the same technology as we address in another country. In India, we have presence in two-wheeler market like e-bikes and motor bikes. We use 6-axis technology to prevent the rollover of motor bikes. We will address the rest including the shared transportation and the commercial vehicles on the national mobility plan with the same technologies used in other countries but it could be possible to customize the final product if we consider different form factor or packaging.
ELE Times: Recent advances of sensor technologies have been powered by high-speed and low-cost electronic circuits, novel signal processing methods, and advanced manufacturing technologies. The environment for these sensors continues to be increasingly challenging with respect to robustness, reliability, quality and cost. Please discuss the background of innovative sensor technologies developed by ST recently.
Davide Bruno: Robustness and reliability are somehow linked with each step from the concept of the product and its end-use (if consumer, if industrial or for automotive applications). The design or manufacturing of the product, which include testing at different levels – wafer or final product level, ST has more than 20 years of experience with fluidic MEMS and motion sensor MEMS. We can surely say, (17 billion parts sold so far), that ST MEMS technologies is robust and reliable. The ASIC portion of the specific product (an accelerometer, gyroscope, or pressure sensor) is produced in standard technology and consider the nature of signals that are involved when we measure from an acceleration to a pressure, we do not phase big challenge in respect of high-speed, neither constrains for novel signal processing.
Moving toward Automotive applications, we need to define different methodologies at design stage of both mechanical and ASIC part to follow automotive design rules. This will imply that we might need to design and work with equipment makers to have equipment able to achieve some specific geometric specification. Here we work with key tools manufacturing to keep challenging this specific step of manufacturing. Also, we need to define a different flow for testing. Automotive customers require a very low PPM; we need to implement dedicated tests, for example burn-in test, or hot and cold wafer test or narrowed final test, that are not used or needed for consumer MEMS where a relaxed PPM is allowed.
From the perspective of cost, if we compare the cost of the gyroscope 10 years ago with its cost today, they are quite different. Semiconductor is the key for the cost reduction. These technologies are not new but in ST we are working to improve different technologies (for motion sensor or for environmental sensors) to achieve better performance (new equipment) and a more reliable way (new test methodologies). A significant effort is put in the methodologies of testing. Testing can result in a very low yield (too much screening) with a consequent cost increasing. So, a strong know-how of the technologies, the variation in production of specific parameter and ultimately the definition of test-methodology that guarantee high level of quality and an affordable final production cost are surely the challenges. We are answering with our new product introduction in the market. We do not necessarily need to reinvent the new technology to address sensors for automotive, the challenge is to reduce the cost of the system by increasing the accuracy, enhance linearity, and having a better stability all of these aiming a PPM level in line with the automotive customer request.
ELE Times: Could you give an overview about the significant developments of methods, structures, manufacturing technologies, and signal processing characterizing today’s sensors and sensor systems for automotive industry?
Davide Bruno: Thank you for these challenging questions. In terms of methodology, we have no new methodology beside what already answered, new methodologies in testing aimed at reducing the PPM. For structures, speaking about the mechanical parts, with the 8-inch fabrication, fab that we have, we constantly work with all key equipment maker to reach geometry able to reduce mechanical failure, reduce noise, reduce sensitivity to external factor like temperature and increase the stability. This is possible to be done because ST has its own fab for MEMS and we mastered the knowhow of MEMS fabrication in its single step. This is also the key for automotive. If you talk about automotive, you do not want to change the technology every day. The OEM and Tier 1 are very scared to use new technologies because they have not tested the technologies on the field, and they have no knowledge of different conditions, and do not understand how these technologies will work within different environmental conditions. The point is that ST has been proved in the market, the design of the mechanical part is ST own IP. The technologies we have developed in the market in the past are the key advantages to succeed in automotive when, as said, we are able to innovate in testing methodologies and challenging new reliable geometries starting from a solid and proven and successful technology.
For technology, while it is reliable, stable and in the field for many years. ST was the 1st MEMS vendor to introduce the machine learning core among different sensor vendors. ST anticipated the news of each sensor to be “independent” and partially autonomous from the main central processing unit. By doing so, in wearable applications, we could improve the battery life by 1000 times, because what was before done by using the main CPU/MCU is now done in the sensor itself. For what we have said talking about smart cars, the smart sensor, with machine learning capability or AI capability, will be soon coming and we are already there with our experience from consumer or/and industrial product then we will enhance more in Automotive this existing technologies.