Influence of Randomness During the Interpretation of Results From Single-Event Experiments on SRAMs

Abstract

After having carried out radiation experiments on memories, the detected bitflips must be classified into single bit upsets and multiple events to calculate the cross sections of different phenomena. There are some accepted procedures to determine if two bitflips are related. However, if there are enough bitflips, it is possible that unrelated pairs of errors appear in nearby cells and they are erroneously taken as a multiple event. In this paper, radiation experiments are studied as a special case of the urn-and-balls problem in probability theory to estimate how the measured multiple-event cross sections must be corrected to remove the overestimation due to the false events.

Inherent Uncertainty in the Determination of Multiple Event Cross Sections in Radiation Tests

“…Unlike Eqs. 4-6, already tested in [14], (7), (9) & (10) need to be verified. Thus, a Monte Carlo analysis was performed injecting random SBUs in a simulated 4-Mb memory.…”

“…Previous studies [8], [14] have proposed equations to estimate the expected number of false 2-bit MCUs, when geometric methods are used to group bitflips in multiple events:…”

“…If N BF were used, previous expressions would only provide a very pessimistic prediction that, at any rate, can be interesting enough to the researcher if the values allow ruling out the occurrence of false events. Let us note that (7) is a slightly corrected version of the one derived from the birthday-statistics test [8], [14] to exclude the central cell (T = 0). Recently, in [16] the authors needed to determine the ratio between false and actual 2-bit MCU.…”

“…This model has been widely used since its publication in order to test the accuracy of experimental multiple-event cross sections [5], [9]- [13]. Recently, the authors related the birthday statistics with the expected number of false pairs of events according to the classical "Urn-and-balls" problem [14]. Thus, the expected number of false MBUs and of false 2-bit MCUs can be determined, provided that the memory size, the word width and the number of bitflips are known.…”

“…In the literature, there are several methods [8], [14] to determine if the addresses (a i , a j ) are independent of each other or belong to the same multiple event. Some of them are: 1) MBUs: Bitflips are related if they belong to the same word.…”