From Flash to MRAM: Meeting the needs of modern FPGA configurations

Emerging avionics, critical infrastructure and automotive applications are redefining expectations for field-programmable gate arrays (FPgas) at a time of rapid technological development. Fpgas previously relied primarily on flash memory to store configuration bit streams. This method is suitable for many mainstream FPGA configuration applications. However, as technology advances and the need for greater reliability and performance increases, there is a need for more diverse configuration storage options. The catalyst for this shift is the diverse needs of applications and industries that are now pushing the limits of FPGA applications, demanding further improvements in data integrity, system durability and operational efficiency.

Modern applications require more advanced features

1. Increased durability and reliability: Applications such as advanced driver assistance systems and advanced interconnect avionics require components that can withstand harsh environmental conditions and have high durability. Flash memory, while reliable under certain conditions, has limitations in durability and therefore cannot meet these stringent requirements.

2. Faster configuration times: In time-critical environments such as real-time sensor data processing or high-reliability communications, the need for fast configuration is critical. Traditional flash memory can cause startup time delays.

From Flash to MRAM: The key to mission success

When designing circuits or applications for FPGas, a hardware description language (HDL) is needed to describe how functions inside the FPGA should be routed. HDL code is compiled into an FPGA configuration file, a bit stream, using FPGA development software such as Lattice Radiant™. The bitstream contains binary data that tells each logic element inside the FPGA (triggers, gates, etc.) how to connect and perform digital functions. After the bit stream is generated, it is stored in a non-volatile memory device. During the power-on process, the configuration bit stream is loaded to the FPGA. Once the bitstream is configured, the FPGA begins to perform any number of programming tasks, such as data or signal processing, control functions, and protocol bridging.

Magnetic random access memory (MRAM) is an emerging non-volatile memory technology that uses the magnetic properties of materials to store data. Unlike conventional flash memory, which relies on charge storage, magnetic random access memory uses magnetic tunneling junctions to represent binary data as the direction of a magnetic state. This approach has several advantages, including lower power consumption, higher durability, and faster read and write speeds. In addition, the non-volatile nature of MRAM ensures that data is retained even when there is no power supply, making it a reliable and efficient alternative to flash memory. The scalability of MRAM and its ability to seamlessly integrate with CMOS processes further make it a strong contender in the pursuit of energy-efficient storage solutions.

While traditional storage technologies such as flash memory perform well, new applications are driving the need for more reliable configuration storage that requires greater stability and performance in harsh environmental conditions. For example, in network edge applications that require high durability or high performance, MRAM can handle a large number of high-speed read/write cycles over OTA to support continuous data updates without going through an erase cycle or using flash file systems or dedicated controllers.

In automotive applications, MRAM operates efficiently over a wide temperature range and under harsh conditions. In mission-critical transportation and avionics applications, MRAM is critical for storage systems' setup and operational data logging. In space applications where data reliability is highly required, MRAM has the ability to resist strong radiation, simplify on-orbit reprogramming, and limit radiation-induced errors.

Future-proof applications with MRAM-enabled FPgas

Lattice FPGas, including Lattice Certus™-NX, CertusPro™-NX and Avant™, use reliable process technology, optimized architecture, and proven design technology. Background reconfiguration, built-in hard erasers to detect and correct errors, fault-tolerant IP, and a variety of targeted tools to help reduce reliability risks. With these FPGA devices, users can benefit from low power FPGA architectures and fast and secure bitstream configuration/reconfiguration.

To enhance the programming experience for customers, Lattice is working on updating EDA tools to support native MRAM programming. We have been working with prominent MRAM product vendors including Everspin Technologies and Avalanche Technology to demonstrate hardware interoperability and native software integration.

Lattice's newly released Radiant tool provides direct access to and programming interface to MRAM bitstreams, supporting a variety of data rates and storage capacities. It supports devices with MRAM capacity up to 256 Mb and fully supports SPI, QSPI, and xSPI. Lattice's latest FPgas now support X8 data widths operating at 160 MHz, achieving the industry's fastest configuration times. Combining this high-performance FPGA configuration interface with MRAM brings inherent design advantages to mission-critical systems.Lattice Radiant supports MRAM SPI flash memory.

Using MRAM to store FPGA configuration bitstreams is not only a technology upgrade, but also a strategic move towards a highly reliable system for the future. With the increasing demand for their electronic components in various industries, FPGA systems that support MRAM have become the best solution for zero-tolerance applications.