Radiation-Hardened FPGA

A programmable chip that a proton can punch straight through without scrambling a single bit of logic — because every storage cell was designed from the transistor up to survive it.

An FPGA (Field-Programmable Gate Array) is a chip that engineers can configure in software to act like any digital circuit — a signal processor, a flight computer, a high-speed communications interface — without custom silicon. In space, this reconfigurability is enormously valuable: the same chip can be reprogrammed in orbit to fix bugs or add capability. The problem is that space is filled with high-energy protons, heavy ions, and electrons that can flip the bits inside ordinary chip memory, corrupting the FPGA's configuration and causing catastrophic failures. A rad-hard FPGA adds a suite of hardening techniques — at the transistor level, the circuit level, and the architecture level — to guarantee that configuration upsets, single-event latchups, and accumulated dose effects cannot corrupt or destroy the device over a mission spanning five to fifteen years.

Radiation damage falls into two categories. Total Ionising Dose (TID) is the cumulative energy deposited by all radiation over the mission; it degrades transistor thresholds and eventually causes parametric failures. Single Event Effects (SEE) are caused by individual particles: a Single Event Upset (SEU) flips a stored bit; a Single Event Latchup (SEL) can latch the chip into a permanent high-current state and destroy it. Three principal hardening approaches are used, often in combination. First, process hardening: Microchip's RTAX antifuse-based FPGAs and RTG4 flash-based FPGAs use intrinsically radiation-resistant cell structures — antifuse connections are permanently blown and cannot be un-blown by a particle; flash cells are engineered to tolerate cumulative ionisation without threshold shift. Second, circuit hardening by design (RHBD): Xilinx's Virtex-5QV uses dual-node interlocking latches where each storage cell has two nodes so far apart that a single particle track cannot strike both; only a simultaneous upset of both nodes can flip the cell, an astronomically unlikely event. Third, Triple Modular Redundancy (TMR): critical logic is triplicated — three copies of every register and voter circuits resolve any single-copy disagreement by majority vote. Some FPGAs implement TMR in hard silicon; others offer it as a synthesis option applied by the designer.

In plain terms

Imagine a whiteboard that you can write any program on, and it stays exactly as you wrote it even when someone shoots tiny invisible bullets through it every few minutes for ten years. Normal computer chips would have their writing smeared all over the place by those bullets. A rad-hard FPGA is a special whiteboard where every letter is locked down so tight that even a cosmic ray particle can't move it.

Also called: rad-hard FPGA, space-grade FPGA, RTG4, Virtex-5QV, RTAX, RHBD FPGA, SEU-immune FPGA.

Why it matters

Where it lives

Key numbers

Surprising facts

A short history

The need for radiation-tolerant programmable logic became acute in the 1980s as satellite designers sought the flexibility of gate arrays without the permanent configuration of ASICs. Early approaches used antifuse technology — Actel (now Microchip) introduced the first antifuse FPGA in 1987, and by the mid-1990s space-qualified antifuse devices were flying. The challenge with antifuse parts was one-time programmability: a design flaw caught after launch could not be corrected. Xilinx addressed this with the Virtex-5QV, introduced around 2008 as the first 65 nm rad-hard SRAM FPGA with 1 Mrad TID tolerance and genuine post-launch reprogrammability. Microchip (through Actel/Microsemi acquisitions) responded with the RTG4 in 2015 — a flash-based device that is both reprogrammable and immune to configuration upsets without scrubbing, combining the best properties of both approaches. By the 2020s, Microchip's RT PolarFire at 481 K logic elements and 10 Gb/s SERDES was pushing performance toward what terrestrial data-centre FPGAs achieved a decade earlier, but qualified for deep space.

Sources

Related