Acronym · Reliability & Quality
FIT (Failures In Time)
The standard failure rate unit, one failure per billion device hours of operation. Rates from accelerated tests are converted to FIT at a stated use temperature, voltage and confidence level, and FIT values add across components in a system.
- Also known as
- Failure unit, FIT rate
- Type
- Acronym
- Domain
- Reliability & Quality, Reliability modelling
- Also relevant to
- Automotive, Safety & Security
Notes
A FIT number is meaningless without its assumed junction temperature and confidence level, which is where most vendor comparisons go wrong.
Overview
FIT, short for failures in time, is the unit the semiconductor industry uses for failure rates: one FIT is one expected failure in a billion (10^9) device hours of operation. A part rated at 10 FIT, for example, would be expected to fail about once a year across a fleet of roughly 11,400 units running continuously. FIT figures are derived from accelerated stress tests and quoted at stated use conditions, and for independent parts in series the rates simply add.
How it works
A billion device hours can be made up in many ways: a thousand parts for a million hours each, or a million parts for a thousand hours. A FIT figure assumes a constant failure rate, which applies to the flat middle of the bathtub curve, after early failures and before wear-out. Under that assumption the mean time between failures is the reciprocal of the rate, so 1 FIT corresponds to an MTBF of a billion hours, a statistical figure rather than an expected lifetime for any single part.
No one tests parts for a billion hours. Instead, manufacturers run accelerated tests such as high temperature operating life (HTOL), which stresses parts with heat and voltage, count the few failures, often none, and convert the result to use conditions. The temperature conversion uses the Arrhenius equation with an activation energy. In one TI example, a rate of 16.7 FIT at 55 °C becomes 69.2 FIT at 75 °C with an activation energy of 0.7 eV. Because so few failures are seen, the result is stated as an upper bound at a chosen statistical confidence level, calculated with chi-square methods; TI's calculators follow JEDEC formulas.
Where wear-out matters, the constant-rate assumption breaks down, and lognormal or Weibull models give an average failure rate over a chosen period instead. Handbook methods such as MIL-HDBK-217F offer predicted rates when test data are missing.
Examples
- TI's Arrhenius example, in which 16.7 FIT at 55 °C becomes 69.2 FIT at 75 °C for an activation energy of 0.7 eV.
- An HTOL qualification with no failures converted into an upper-bound FIT rate at a stated confidence level.
- Adding the FIT rates of every component on a board to estimate the failure rate of the whole module.
- Using a Weibull model to estimate the average failure rate of a part whose failures rise with age.
- A handbook prediction from MIL-HDBK-217F used early in a programme before test data exist.
Why it matters
Related terms
Part of the Reliability & Quality glossary, with every term in the topic and the lifecycles, tools and standards that use them.
Where it appears on the Hub
Used on 8 pages of the Hub. In each group, the pages that mention it most come first.
Suggest an improvement
Suggest an improvement to this entry
A clearer definition, an example, a correction or a source an editor can check. Editors review every suggestion before the entry changes.
Last reviewed . Written by The Semiconductor Hub editorial team. The full glossary