Due to the semiconductor shortage in 2021 everyone realized that cars these days integrate a lot of electronics. The average number of computer chips per car has increased a lot in the last decade.
If you think about it, future cars actually combine several trends in one system:
- High-bandwidth (5G) communication with car maker, other cars, infrastructure, internet
- Huge amounts of data collected during every ride
- Broad set of sensors, including new radar, lidar
- Further electrification, electric cars require innovation on power conversion
- Artificial intelligence for increasing levels of autonomous driving
Originally the network for electronics in cars was so-called ‘flat’. To reduce the total (electric) wire length and associated weight experts have proposed domain and zone oriented approaches. These can also reduce the complexity when additional applications are added and can help to reduce the number of ECUs.
Different types of car electronics
There are 2 different types of semiconductor products in cars.
- [A] A decade ago, the majority of the car electronics was produced on mature, BCD processes. Electronics were used in harsh ‘under-the-hood’ applications close to the engine. Compute performance was limited. Microcontrollers could easily handle the workload.
- [B] More recently car makers started using more advanced CMOS and FinFET processes for infotainment and ADAS systems. These high value and high margin applications drive a lot of innovation turning the car into a datacenter on wheels. For these newer systems high-speed connections are required for the transfer of video and sensor data to the high end processors. The car industry is debating about the best interface for these links: SerDes or Ethernet (link).
High voltage requirements for type A electronics
The car battery is a ~12V device. But car electronics must be able to sustain much higher voltage (transients).
A typical LIN interface for instance consists of a combination of a mostly digital (controller) function and a mostly analog (transceiver) block. The controller can be easily instantiated in any logic process and thus integrated with other digital functions or programmed in an FPGA circuit. The transceiver or PHY on the other hand requires high voltage tolerant pads. Despite a 14V nominal operation for the car battery these LIN pads must be able to sustain 40V or more.
Severe reliability requirements for LIN interfaces and associated power rails
- Above standard HBM, MM requirements
- Transient latch-up immunity: -27V to 40V
- ESD under powered conditions: 0V to 18V
- IEC 61000-4-2 system ESD
- ISO 7637-2 load dump pulse
- EMC IEC 62132 DPI
To handle these types of requirements, engineers design products on so-called BCD processes. It also requires special ESD protection concepts (link).
Innovation for old interfaces
It is clear that the new applications require high-speed interconnects that are not possible with the initial, low-speed interface types. But there is also innovation possible for the old interface types, used in the first application space (A).
One of the decade-old interface types is based on the LIN (Local Interconnect Network) standard. From Wikipedia: “LIN is a serial network protocol used for communication between components in vehicles. It is a single wire, serial network”. In average, modern cars now use between 10 and 50 such LIN interfaces.
System Basic Chips – SBCs
Due to the high total number of chips required every year combined with a standardized protocol and lots of competition between several vendors, the discrete LIN transceiver market is turned into a commodity. In high volume they come at $0.3 a piece. The current vendors produce those LIN transceiver ‘discrete’ on fully paid-off fabs in specially tuned processes.
But something is changing. Car makers and ECU developers demand higher integration, fewer chips. Automotive IC designers are thus trying to integrate the transceiver with the digital logic into SBC’s (System Basic chips) that combine LIN, CAN transceivers and other functions like power management, wake circuits, logic/MCU on one chip, sensor integrations.
The benefits for an integrated approach are clear
- The total cost is lower than the combination on PCB/system of the different parts
- Reduced Bill of materials, improved yield/quality
- Improved electrical parameters like power consumption, reduced duplication of internal blocks (reference voltage, monitoring circuits)
At Sofics we also receive a lot of support requests lately for such an integrated solution. IC design companies with limited automotive experience look for third IP providers for the analog blocks like LDO, and LIN/CAN transceivers. While the control logic for LIN/CAN is easy, the physical layer is not straightforward due to the reliability requirements. Therefore, our engineers developed a LIN transceiver IP block that can be integrated along other functions on the same die. This enables fabless companies without prior automotive expertise to join in on this automotive growth opportunity.
Contact us to discuss your LIN transceiver SBC product.