Imagine this: If you drive the latest car, speeding on the highway at 75 miles per hour, while listening to popular music.
Suddenly, the engine management system or stability control system failed. Faced with this situation, not only may you encounter a serious (even fatal) car accident, but the reputation of the car manufacturer may also be damaged, if similar situations do not just happen to you.
As cars have evolved from purely mechanical equipment to highly integrated electronic systems for driving-by-wire cars in the past 25 years, the challenges faced by designers have also increased. They must add complex electronic equipment to each subsequent model, while maintaining high standards of quality and reliability, and meeting stringent low-cost and mass production requirements.
Traditionally, automotive developers have always relied on MCUs, ASICs and huge wiring harnesses to implement and control electronic systems and expand the functions of each generation of cars. Today, these solutions are approaching their technological limits, and as the complexity of automotive electronics has grown exponentially, their reliability has also attracted people's attention. In order to solve these problems, many designers began to turn to FPGAs as a flexible and low-cost solution for the next generation of automotive electronics design.
In order to ensure the normal functioning of various systems in modern automobiles, reliability data requirements must be put forward for the components. Although people have mastered most of the elements of component reliability, but in the process of selecting programmable logic devices such as FPGAs, some special issues must be considered.
Understand the destructive effects of cosmic rays
Specifically, technical decision makers must foresee the sources of failures affecting programmable logic systems. The bombardment of neutrons from space (cosmic rays) sounds a bit like a plot in the movie "Interstellar Travel", but in fact the errors caused by neutrons are harmful to many types of electronic devices. The firmware error caused by neutrons has changed from a pure trouble to a big problem. For example, if neutrons cause an SRAM-based FPGA configuration unit to be disrupted, it may disable the function. If this happens, it can cause the main system to malfunction. Looking to the future, this problem will be more serious, because future FPGAs will use deep sub-micron manufacturing processes, which will bring real challenges to FPGA-based automotive electronics design engineers.
Single event disturbance (SEU) caused by neutron bombardment is prone to occur in any type of volatile storage unit. The above-mentioned SRAM FPGA uses an internal storage unit to maintain the configuration state (or personality) of the FPGA. Such memory cells have more serious reliability problems. When the content is changed, we say that the device has a "soft error", because in this case, only the data is affected, and the function is not affected. Although correction data can be successfully rewritten to the device (EDAC (error detection and correction) and TMR (tunnel magnetoresistance) technologies can be used to correct SRAM data and registers, respectively), soft errors can still cause data loss or "system Unexpected failure".
If the SRAM FPGA configuration memory cell is damaged, we say that the device has "firmware errors" because these errors are not easy to detect or correct and are not temporary in nature. Once a firmware error occurs in the FPGA, the initial configuration data must be reloaded. In some cases, the power must be re-powered to clear the fault, and then reconfigure. In these configuration units, as long as one encounters an SEU caused by neutrons, the consequences can be very serious. If a configuration bit is disturbed and changes its state, it may change the function of the entire device, causing major data corruption or sending false signals to other circuits in the system. In extreme cases, if firmware errors are not detected for a long time, they will become "hard errors" and cause damage to the device itself or the system containing the device. A common example of this type of problem is: a firmware error caused by a neutron leads the signal to the wrong path, causing a short circuit.
For the use of SRAM FPGAs to implement mission-critical automotive electronic application systems, errors caused by neutrons have serious potential risks. Since the existing detection technology realizes detection by reading back the FPGA configuration at regular intervals, it is of no help in preventing errors in the system. In addition, the readback circuit capable of detecting corrupted configurations is inherently vulnerable to SEU or destruction. Furthermore, with the widespread development of this vulnerable FPGA technology, a new quality evaluation system that can check the immunity of automotive electronic systems to firmware errors caused by neutrons may be needed, and it will be added to the AEC-Q100 standard. , To supplement the deficiencies of JEDEC Standard No. 89. In addition, the current solutions for detecting and correcting FPGA firmware errors will increase the complexity of the system design, and greatly increase the size of the circuit board and the cost of materials, thereby increasing the cost of the bill of materials for discovering errors caused by neutrons.
The firmware error caused by neutron may have a great impact on the FIT (failure in time) rate of the entire system. Such firmware errors are difficult to detect and almost impossible to diagnose, causing thorny problems for maintenance and repairs, and increasing maintenance costs. Among the three mainstream FPGA technologies (anti-fuse, Flash, and SRAM), only anti-fuse and Flash technologies can resist the effects of soft errors and firmware errors caused by neutrons.
Example: Automotive electronic system implemented with SRAM FPGA
This example analyzes a system installed in the floor of the cab. Someone used SpaceRad 4.5 (a widely used radiation effect prediction software program) to calculate the density of neutron rays at an altitude of 5000 feet in Denver, Colorado. According to published radiation data, for a SRAM FPGA with a density of one million gates using 0.22 micron technology, the probability of SEU occurring per day is 1.05E-4.
If the manufacturer uses a million-door SRAM FPGA in the occupant sensor and airbag control module, the SEU occurrence probability of 1.054E-4 is converted to the SEU occurrence probability per system per day, which is 4.38E-06, or the FIT value of the system Is 4,375. This means that if the manufacturer adopts this million-door SRAM FPGA security system in 500,000 vehicles, the probability of occurrence of this SEU of 1.054E-4 is multiplied by the number of vehicles/systems on the road, and the number of vehicles/systems on the road is obtained. A total of 52.5 SEU events will occur (assuming that the vehicle has been running). This is equivalent to an SEU every 27.4 minutes. Even assuming that these vehicles only travel for two hours a day on average, two SEUs will still occur every day. Since these failures are firmware failures, they will continue until the SRAM FPGA is reloaded (usually a power cycle or forced reconfiguration is required).
In the current semiconductor technology, soft errors in devices have received great attention. As device sizes continue to shrink, everyone recognizes that such soft errors will become a big problem; these errors may greatly reduce the availability of the system. Therefore, in many applications, people are strongly required to avoid soft errors, so that the usability of the system can be maintained at an acceptable level.
When choosing an FPGA, the most important thing is to evaluate the total cost of ownership of each programmable system, and choose those suppliers that have essentially reliable core technologies, rather than second-class quality supplies designed for lower-level applications. Commercial grade products provided by vendors.
If SRAM FPGA is used to design, designers must add circuits for detecting and correcting configuration errors, which will increase system cost and complexity. Fortunately, designers have other options. Radiation test data shows that FPGAs based on anti-fuse and Flash technology are not prone to configuration data loss due to SEU events caused by neutrons. This makes them particularly suitable for applications requiring high reliability.
Now, imagine a slightly different scenario: you are driving your latest car on the highway at 75 miles per hour, listening to popular songs in your ears. Knowing that the engine management system uses a non-volatile Flash-based FPGA instead of an SRAM-based FPGA, you can continue to push and accelerate and enjoy a comfortable and worry-free journey.
5W Wall-Mounted Power Adapter,5V1A Universal Travel Charger Adapter,5V1A Portable Travel Charger Adapter,5V1.2A Powe Adapter For Security Monitor
Guangdong Mingxin Power Technologies Co.,Ltd. , https://www.mxpowersupply.com