Comparison of Single Event Upset Effects on the Clementine and Cassini Solid State Data Recorders—A Study in Data Mining

As part of an on-going “data mining” effort, the response to single event upsets (SEU) of the Clementine 2.1 Gb Solid State Data Recorder (SSDR) was compared to that of the two Cassini 2.5 gigabit (2.1 Gb usable for data) Solid State Recorders (SSRs) to see what lessons could be learned. Both systems were evaluated for their sensitivities to SEUs before flight. Estimates of the in-situ environments for the two missions allow evaluation of the ability of SEU models and ground tests to predict flight performance using actual data. The DRAMS that make up the solid state recorders, despite having different manufacturers, appear to have similar SEU cross-sections. This similarity has permitted a comparison of the effects of the ambient environments on the systems for two very different missions (lunar versus Saturn). Initial results from previous studies had revealed a nearly constant background upset rate for both systems of ~71 bit flips/day for the SSDR and ~280 for the SSRs due to the Galactic Cosmic Ray background. While there was no obvious correlation with a solar proton event recorded by Clementine on 20-21 February 1994 nor with trapped protons during its brief passage through the Earth’s radiation belts, the Cassini SSRs showed pronounced responses to both solar proton event and Saturn trapped radiation environments. This difference is explained here by applying the proton cross-sections measured for Cassini to the Clementine observations—the new results show that the protoninduced upset rates would have been too low to be observed by Clementine. This study completes the original Clementine SSDR analyses and, in the process, demonstrates agreement between the Cassini SSR upsets and the JPL SATRAD proton radiation model. Finally, the lunar orbit variations in the SSDR upset rates observed by Clementine are reevaluated using a new methodology—the pronounced lunar orbit altitude dependence is shown to fit the expected variation in GCR fluxes due to lunar shielding.