Commercial off-the-shelf programmable System-on-Chip devices for space applications
Εμπορικά διαθέσιμες συσκευές προγραμματιζόμενου συστήματος-σε-chip για διαστημικές εφαρμογές
View/
Keywords
Programmable System-on-Chip (SoC) ; Field Programmable Gate Array (FPGA) ; Space applications ; Single Event Effect (SEE) ; Radiation effects ; Reliability ; Memory scrubbing ; Heavy-Ion irradiation ; Fault tolerance ; Proton irradiation ; Fault mitigation ; Space electronicsAbstract
The recent advances of the semiconductor industry have enabled the integration of complex components and system architectures into a single silicon die. Nowadays, state-of-the-art programmable circuit devices, such as field programmable gate arrays (FPGAs), include not only programmable logic fabric but also hardwired components, such as hard-core general-purpose processors, dedicated processing blocks, interfaces to various peripherals, on-chip bus structures, and analog blocks. These new heterogeneous devices are commonly referred to as all programmable system-on-chip (APSoC) devices.
One of the major concerns about radiation effects on APSoCs is that radiation-induced errors may have different probabilities and criticalities in their heterogeneous hardware parts at both device and design levels. For this reason, this work performs a deep investigation of the radiation effects on APSoCs and the correlation between hardware and software resource sensitivity in the overall system performance. Several static and dynamic experiments were conducted on various hardware components of an APSoC.
This thesis presents a comprehensive study on the single event effect (SEE) susceptibility of all major components in a commercial off-the-shelf (COTS) AMD Zynq-7000 APSoC. The work encompasses detailed radiation experiments, in-depth analysis of observed SEE phenomena, the SEE upset rate estimation for three different orbits, and the development of an efficient mitigation strategy to meet the high-reliability demands of space environments.
Initial efforts targeted the characterization of the programmable logic (PL) of the Zynq-7000 under ultra-high-energy and very-high-energy heavy ions at CERN Super Proton Synchrotron North Area (SPS-NA) in Geneva, Switzerland and GSI Helmholtz Centre for Heavy Ion Research in Darmstadt, Germany respectively, and high-energy protons at Paul Scherrer Institute (PSI) – Proton Irradiation Facility (PIF) in Villigen, Switzerland. Memory elements within both the design and configuration layers were evaluated, with calculated cross sections for each memory type. Novel or under-investigated SEE behaviors, such as single-event resets and functional interrupts, were revealed, which can evade traditional redundancy-based protection methods like triple modular redundancy (TMR). Multi-bit upset (MBU) and multi-cell upset (MCU) patterns were characterized across memory types, highlighting limitations in built-in error correction and informing the design of more robust scrubbing and interleaving schemes.
The study was extended to the processing system (PS) of the Zynq-7000 using proton at PSI-PFI and heavy-ion irradiation at GSI. SEE sensitivity of SRAM-based arrays (e.g., OCM, L1/L2 caches) was assessed under static and dynamic workloads, and application-level effects were quantified by executing processor benchmarks under varied cache configurations. These findings enabled accurate modeling of real-world error impact at the system level.
A key outcome is the introduction of a robust configuration memory scrubbing technique. The proposed mixed 2-D coding scheme integrates internal error correction code (ECC) with an interleaved parity code to detect and correct both single bit upsets (SBUs) and multiple bit upsets (MBUs) with minimal latency and overhead. Two implementations, an external microcontroller-based scrubber and an on-chip hardware scrubber, were validated under irradiation and compared favorably to state-of-the-art methods.
Finally, upset rate predictions were performed for typical space orbits using a dedicated software tool for space environment and radiation effects on electronic devices such as OMERE, demonstrating the applicability of the Zynq-7000 APSoC in radiation-prone missions. This work lays the groundwork for future APSoC reliability strategies and advances the state of knowledge in SEE effects and mitigation across heterogeneous SoC architectures.