⚡ Single Event Upset (SEU) Rate Estimator

Upsets per day from memory size, orbit & device class · MTBU · EDAC sizing notes

Ad placeholder

Inputs

Results

Per-bit rate used
– upsets/bit·day
System upset rate
– upsets/day
Mean time between upsets
–
Upsets per scrub interval
–
–
Ad placeholder

📖 How to use

  1. Enter the total sensitive memory and the orbit; the per-bit rates are typical heritage values (galactic cosmic rays + trapped protons, incl. SAA for LEO).输入总敏感存储容量与轨道；单比特翻转率为典型飞行统计值（银河宇宙线 + 俘获质子，LEO 含南大西洋异常区）。
  2. Upsets are a fact of life — the design answer is EDAC + scrubbing, not avoidance; the tool shows expected upsets per scrub interval (keep it ≪ 2 per protected word region).翻转不可避免——设计对策是 EDAC + 定期刷新而非回避；工具给出每个刷新周期的期望翻转数（应远小于每个纠错字 2 个）。
  3. Solar particle events raise rates 10–1000× for hours-days; latch-up (SEL) is a separate, destructive concern needing current limiting.太阳质子事件可使速率升高 10–1000 倍并持续数小时至数天；单粒子锁定（SEL）是另一类破坏性效应，需限流保护。
  4. Order-of-magnitude only — real rates need device test data (CREME96/OMERE with measured cross-sections).仅为量级估计——真实速率需器件试验数据（CREME96/OMERE + 实测截面）。

SEU rates and mitigation

Typical per-bit rates for commercial memory: LEO low-inclination ~1×10⁻⁸, polar/SSO ~3×10⁻⁸, GEO ~1×10⁻⁷, MEO ~2×10⁻⁷ upsets/bit·day. System rate = bits × rate × sensitivity.

With single-error-correct EDAC, failures need two upsets in one code word between scrubs — the scrub interval sets that probability. Uncorrected memories accumulate errors linearly; large mass memories see many upsets daily and are designed around it.

All computation runs locally in your browser.

SEU 速率与对策

商用存储器典型单比特速率：低倾角 LEO 约 1×10⁻⁸、极轨/SSO 约 3×10⁻⁸、GEO 约 1×10⁻⁷、MEO 约 2×10⁻⁷ 翻转/比特·天。系统速率 = 比特数 × 速率 × 敏感度。

采用单纠错 EDAC 后，只有同一码字在两次刷新之间发生两次翻转才会失效——刷新周期决定该概率。不加纠错的存储器误码线性累积；大容量存储每天翻转很多次，设计上必须以此为前提。

全部计算在你的浏览器本地完成。