Automotive manufacturers have long desired a reprogrammable technology able to be applied in a single platform targeted at multiple product models, with the flexibility to adapt and change during lengthy product lifecycles. For example, the ability to design a single engine control module or safety system that can be integrated into both economy and luxury models enables automotive OEMs to streamline design and leverage engineering resources across their entire product line.
But the critical nature of automotive under hood (power train) and safety systems has historically relegated field-programmable gate array (FPGA) technology to non-system critical applications like telematics and infotainment. Manufacturers have been forced to use hard-wired ASICs and application-specific standard products (ASSPs) in system-critical applications in order to ensure firm-error immunity and extended temperature operation—until now. Programmable logic technology has finally evolved to the point where it can reliably and cost-effectively serve the needs of under hood and safety systems in automotive applications. Still, there are a number of considerations which must be taken into account when applying the technology in a design.
Following are some issues that can impact FPGA architecture selection:
Power consumption and thermal issues
All FPGAs are not created equal when it comes to power, yet power is a main consideration when choosing an FPGA solution for an automotive application. By selecting a device with ultra-low power, designers can set aside their concerns related to thermal reliability and runaway.
Clearly, under hood applications expose electronic components to extremely high temperatures. Components used under the hood often must be qualified to operate in a junction temperature range of -40 to +135C (AEC-Q100, Grade 1). Thus power consumption of the device itself can further exacerbate the effects of high temperatures on devices, leading to overheating and failure
The use of flash memory, however, has changed the power landscape for FPGAs, dramatically reducing leakage current to enable ultra-low static power—a key factor in eliminating thermal reliability and runaway concerns. In fact, some flash-based automotive-grade FPGAs are capable of delivering static power as low as 40 mA at 135C.
Firm-error immunity
Nonvolatile flash memory is the key technology for delivering firm-error immunity in system-critical designs. And firm errors don't just affect the quality of the automobile; they can have potentially life-threatening consequences.
Programmable logic devices, such as FPGAs and CPLDs, have traditionally relied on SRAM (static RAM) for configuration memory, making them prone to neutron-induced (i.e. cosmic ray) errors which can cause thousands of failures in time (FIT). Moreover, neutron-induced errors rise exponentially with fluctuations in temperature and altitude. Nonvolatile memory solutions do not suffer from neutron-induced errors and are therefore firm-error immune.
Electromagnetic interference (EMI)
As the use of sophisticated electronics in automobiles continues to grow at astounding rates, so does the issue of EMI. Most all automotive electronics today must meet stringent electromagnetic emissions standards. The inconveniences and critical failures caused by EMI can range from windshield wiper motor interference on the in-cab entertainment system to static charge on, or failure of, the backup assist display.
The ability to quickly and easily tweak designs to eliminate EMI makes a case for reprogrammable FPGAs. Other FPGA benefits include low power; high levels of integration; analog phase-locked loops (PLLs) to generate high-speed, on-chip clocks from low-speed external clocks; and programmable I/O slew rates.
Speed
Automotive electronics designs require lightning-fast response times. Safety-critical systems and closed-loop engine control modules can benefit from the fast response times offered by an FPGA versus an interrupt-driven microcontroller. FPGA response times, which can be below 100 ns, may be several orders of magnitude faster than even a high-performance microcontroller (MCU). These inherently fast response times may allow for lower overall clock speeds—which can further help to reduce both EMI and power consumption.
