Multiple cell upsets in the configuration RAM of a 7-nm FinFET SoC under heavy ions

This manuscript investigates the heavy ion effects on the novel 7-nm FinFET AMD (formerly, Xilinx) Versal SoC, namely on the configuration memory of the embedded FPGA. The experiment consisted in loading a program in a delidded device, irradiating in static mode and reading back the content. Several...

Descripción completa

Detalles Bibliográficos
Autores: Mayo, Jorge, Durán, Cristina, Franco Peláez, Francisco Javier, Cueto-Rodriguez, Juan, Kettunen, Heikki
Tipo de recurso: artículo
Fecha de publicación:2025
País:España
Institución:Universidad Complutense de Madrid (UCM)
Repositorio:Docta Complutense
Idioma:inglés
OAI Identifier:oai:docta.ucm.es:20.500.14352/115657
Acceso en línea:https://hdl.handle.net/20.500.14352/115657
Access Level:acceso abierto
Palabra clave:621.382.3
621.38
629.783
FinFET
FPGA
Heavy ions
Single event effects
Circuitos integrados
Electrónica (Física)
Electrónica (Informática)
3304.08 Fiabilidad de Los Ordenadores
2203.02 Elementos de Circuitos
3324.01 Satélites Artificiales
3307.90 Microelectrónica
Descripción
Sumario:This manuscript investigates the heavy ion effects on the novel 7-nm FinFET AMD (formerly, Xilinx) Versal SoC, namely on the configuration memory of the embedded FPGA. The experiment consisted in loading a program in a delidded device, irradiating in static mode and reading back the content. Several species and configurations were used to test the device. First of all, it was tested the efficiency of the XilSEM tool, provided by the manufacturer to fix erroneous bits. Second, this tool was disabled to find out the cross sections of the single event upsets of different multiplicities. Flipped upsets were classified using statistical techniques to reconstruct the multiple cell upsets. No multiple bit upset occurred and only 4-bit multiple cell upsets were observed in the worst cases, although it is clear that the cross section of the 2-bit multiple cell upsets is comparable to that of single bit upsets. Later, a strategy to carry out fault injection campaigns is proposed taking into account the knowledge acquired after the radiation tests, and the expected error rates in actual space environments is studied.