Failure rate (FIT) calculator
Turn life test results into a failure rate in FIT and a mean time to failure at use conditions, with chi-squared confidence and acceleration from several test legs.
ReliabilityFree to useHow it worksSourcesWhere it helps
Fills the inputs below with typical figures. Replace any of them with your own.
- Failure rate, upper bound at 60% confidence
- 51.1 FIT
- 0 failures in 17,936,083 equivalent device hours
- Point estimate
- 0 FIT
- No failures: the point estimate is zero, the bound is the figure to quote
- Lower bound, same confidence
- 0 FIT
- MTTF at the upper bound
- 19,574,664 h
- 2,235 years
- MTTF, point estimate
- no failures to count
- Chance of a failure over the mission
- 0.447%
- Upper bound over 87,600 powered hours; point estimate 0.000%
- Average acceleration factor
- 77.65
- 231,000 device hours on test
- Equivalent hours for 100 FIT
- 9,162,907
- 0 failures allowed; your data already meet it. About 119 units on the first leg
Show the data as a table
| Confidence level (%) | FIT upper bound |
|---|---|
| 50 | 38.6 |
| 60 | 51.1 |
| 70 | 67.1 |
| 75 | 77.3 |
| 80 | 89.7 |
| 85 | 106 |
| 90 | 128 |
| 95 | 167 |
| 97.5 | 206 |
| 99 | 257 |
- 0.5 eV (solid line, circles)
- 0.7 eV (your value) (dashed line, squares)
- 0.9 eV (dotted line, triangles)
Show the data as a table
| Use junction temperature (C) | 0.5 eV | 0.7 eV (your value) | 0.9 eV |
|---|---|---|---|
| 25 | 29.9 | 4.23 | 0.599 |
| 35 | 56.2 | 10.2 | 1.87 |
| 45 | 102 | 23.5 | 5.42 |
| 55 | 177 | 51.1 | 14.7 |
| 65 | 299 | 106 | 37.8 |
| 75 | 489 | 212 | 91.7 |
| 85 | 779 | 406 | 212 |
| 95 | 1,210 | 752 | 468 |
| 105 | 1,840 | 1,350 | 991 |
Sign in to save your inputs and come back to them later.
| Leg | Units | Hours | Stress Tj (C) | Temperature factor | Voltage factor | Acceleration factor | Equivalent hours | Failures |
|---|---|---|---|---|---|---|---|---|
| HTOL, 3 lots of 77 | 231 | 1,000 | 125 | 77.65 | 1 | 77.65 | 17,936,083 | 0 |
| Confidence level (%) | FIT upper bound | MTTF (years) |
|---|---|---|
| 50 | 38.6 | 2,954 |
| 60 | 51.1 | 2,235 |
| 70 | 67.1 | 1,701 |
| 75 | 77.3 | 1,477 |
| 80 | 89.7 | 1,272 |
| 85 | 106 | 1,079 |
| 90 | 128 | 889.2 |
| 95 | 167 | 683.5 |
| 97.5 | 206 | 555 |
| 99 | 257 | 444.6 |
| Use Tj (C) | FIT at 0.5 eV | FIT at 0.7 eV | FIT at 0.9 eV |
|---|---|---|---|
| 25 | 29.9 | 4.23 | 0.599 |
| 35 | 56.2 | 10.2 | 1.87 |
| 45 | 102 | 23.5 | 5.42 |
| 55 | 177 | 51.1 | 14.7 |
| 65 | 299 | 106 | 37.8 |
| 75 | 489 | 212 | 91.7 |
| 85 | 779 | 406 | 212 |
| 95 | 1,210 | 752 | 468 |
| 105 | 1,840 | 1,350 | 991 |
How it works
A failure rate in FIT is the number of failures per billion device hours. This tool turns the results of one or more accelerated life test legs into a FIT at use conditions, with an upper bound at the confidence you choose, the mean time to failure, the chance of a failure over the mission, and the test hours a target FIT would need.
- Junction temperature. Failure mechanisms follow the temperature of the silicon, not of the oven or the room. If you know only the ambient or case temperature, add the self-heating of the device. Enter either, with the self-heating set to zero when the figure is already a junction temperature.
Tj = T + self-heating - Acceleration factor of each leg. Temperature accelerates a thermally activated mechanism by the Arrhenius factor, with temperatures in kelvin and Boltzmann's constant k = 8.617 x 10^−5 eV/K. If you choose a voltage model, its factor multiplies the temperature factor: exponential (a constant gamma per volt) or a power of the voltage ratio.
AF = exp[(Ea / k) x (1 / Tuse − 1 / Tstress)] x AFvoltage; AFvoltage = exp[gamma x (Vstress − Vuse)] or (Vstress / Vuse)^n - Equivalent device hours. Each leg contributes its units times its hours times its acceleration factor, and the legs are added. This is the exposure at use conditions that the test stands for.
equivalent hours = sum of (units x hours x AF) - Failure rate with confidence. With a constant failure rate and f failures in the pooled test, the true rate is below the upper bound with the confidence chosen. The bound comes from the chi-squared distribution with 2f + 2 degrees of freedom. With no failures it is the single-sided bound that the common 60% and 90% industry figures quote. The point estimate is simply failures divided by equivalent hours.
FIT upper = chi-squared(CL; 2f + 2) / (2 x equivalent hours) x 10^9; FIT point = f / equivalent hours x 10^9 - Mean time to failure and mission risk. For a constant rate the mean time to failure is the reciprocal of the rate. The chance a part fails during the mission comes from the exponential distribution, over the hours the part is powered, which is the mission length times the duty cycle.
MTTF = 10^9 / FIT hours; P(fail) = 1 − exp(−FIT x 10^−9 x powered hours) - Hours a target needs. To show a target FIT with r failures allowed, the equivalent hours must reach the value below. The units needed on your first leg follow by dividing by that leg's hours and acceleration factor.
equivalent hours needed = chi-squared(CL; 2r + 2) / (2 x target FIT x 10^−9)
Assumptions
- The failure rate is constant over the period covered (the flat part of the bathtub curve); early life and wear-out are not modelled here.
- One failure mechanism, with one activation energy, limits all the legs that are pooled.
- The units tested are a random sample of the population, from enough lots to represent it.
- FIT is counted over powered hours. Hours when the part is switched off are not counted.
- Default activation energies are editorial starting points inside the published ranges. Use the value your own data or your customer's method requires.
Limits and warnings
- The upper bound is only as good as the activation energy. A change of 0.1 eV can move the FIT by a factor of two or more at a large temperature step; the temperature chart shows this.
- Several mechanisms with different activation energies need the sum-of-failure-rates method: run each mechanism's data apart and add the FITs.
- Early life failures (infant mortality) are measured by a short, voltage-stressed screen and quoted in ppm. They do not extrapolate with the steady-state FIT method.
- Acceleration factors beyond about 100 rely on the chosen model holding over a long range of temperature; treat them with care.
- Self-heating is entered, not calculated. For power devices it can be large and vary between the stress and the field.
Sources
The methods and default figures rest on these public sources. The wording here is our own; read the source for its full text.
- JESD85, Methods for Calculating Failure Rates in Units of FITs (revised as JESD85A) (opens in a new tab), JEDEC Solid State Technology Association, July 2001; JESD85A July 2021. The method: constant failure rate, Arrhenius acceleration, chi-squared upper bound. The standard itself is free to download after registration; the method here is written in our own words.
- JEP122, Failure Mechanisms and Models for Semiconductor Devices (JEP122H) (opens in a new tab), JEDEC Solid State Technology Association, September 2016. Where activation energies and acceleration models for each mechanism are listed.
- Calculating FIT for a Mission Profile (SPRABY3), Allan Webber (opens in a new tab), Texas Instruments, March 2015. A worked HTOL example (231 units, 125 C to 55 C, 0.7 eV, no failures, 60% and 90% confidence) and how sample size changes the FIT.
- Semiconductor Device Reliability Failure Models (Technology Transfer 00053955A-XFR), Richard Blish and Noel Durrant (opens in a new tab), International SEMATECH, 31 May 2000. Activation energy ranges by mechanism and a worked FIT example from a 24-hour high temperature test.
Where it helps
The lifecycle stages this tool serves: 13 stages in 12 lifecycles.
More reliability tools
- Stress test acceleration factor calculator
- Qualification test planner
- Defect level (DPPM) estimator
- Error rate confidence calculator
- Lot sampling plan calculator
- Stack and multi-die yield calculator
- Radiation dose margin calculator
- Single-event upset rate estimator
- Component derating checker
- Mission profile builder
- Warpage and thermal mismatch estimator
Method notev1Formula, ranges, sources and limits
Turns accelerated life test results into a failure rate in FIT at use conditions with a chi-squared upper confidence bound, the mean time to failure, and the chance of a failure over a mission.
Formula
Tj = T + self-heating; AF = exp[(Ea / k) x (1 / Tuse − 1 / Tstress)] x AFvoltage; AFvoltage = exp[gamma x (Vstress − Vuse)] or (Vstress / Vuse)^n; equivalent hours = sum of (units x hours x AF); FIT upper = chi-squared(CL; 2f + 2) / (2 x equivalent hours) x 10^9; FIT point = f / equivalent hours x 10^9; MTTF = 10^9 / FIT hours; P(fail) = 1 − exp(−FIT x 10^−9 x powered hours); equivalent hours needed = chi-squared(CL; 2r + 2) / (2 x target FIT x 10^−9)
| Input | Unit | Range it holds for | Comes from |
|---|---|---|---|
| Activation energy | eV | -0.5 to 1.8 | Your figure |
| Use junction temperature | degrees C | -55 to 175 | Your figure |
| Use self-heating | degrees C | 0 to 100 | Your figure |
| Use voltage | V | from 0 | Your figure |
| Voltage acceleration model | choice | none, exponential, power | Your figure |
| Voltage gamma (exponential) | per V | any | Your figure |
| Voltage exponent (power law) | number | any | Your figure |
| Confidence level | choice | 60, 90, 95, other | Your figure |
| Other confidence level | % | 1 to 99.99 | Your figure |
| Mission life | years or hours | from 0 | Your figure |
| Mission life unit | choice | years, hours | Your figure |
| Duty cycle (powered share of the mission) | % | 1 to 100 | Your figure |
| Target FIT | FIT | from 0 | Your figure |
| Failures allowed for the target | count | 0 to 50 | Your figure |
Outputs: FIT, upper bound at the confidence (FIT); Average acceleration factor (ratio); MTTF at the upper bound (years); Chance of failure over the mission, upper bound (ppm).
Assumptions
- The failure rate is constant over the period covered (the flat part of the bathtub curve); early life and wear-out are not modelled here.
- One failure mechanism, with one activation energy, limits all the legs that are pooled.
- The units tested are a random sample of the population, from enough lots to represent it.
- FIT is counted over powered hours. Hours when the part is switched off are not counted.
- Default activation energies are editorial starting points inside the published ranges. Use the value your own data or your customer's method requires.
Uncertainty: The statistics are exact for a constant failure rate. The weak point is the activation energy: a 0.1 eV error moves the result by a factor of about 2 for a 70 C step. The test legs themselves are not part of the register inputs, which use the default HTOL leg of 231 units for 1,000 hours at 125 C.
Known limits
- The upper bound is only as good as the activation energy. A change of 0.1 eV can move the FIT by a factor of two or more at a large temperature step; the temperature chart shows this.
- Several mechanisms with different activation energies need the sum-of-failure-rates method: run each mechanism's data apart and add the FITs.
- Early life failures (infant mortality) are measured by a short, voltage-stressed screen and quoted in ppm. They do not extrapolate with the steady-state FIT method.
- Acceleration factors beyond about 100 rely on the chosen model holding over a long range of temperature; treat them with care.
- Self-heating is entered, not calculated. For power devices it can be large and vary between the stress and the field.
Where the method comes from
Follows JEDEC JESD85, Methods for Calculating Failure Rates in Units of FITs (opens in a new tab). The arithmetic here is the same; the platform claims no improvement on it.
Reference example, checked by test: Worked example in Texas Instruments application report SPRABY3 (March 2015): 231 units, 1,000 h at 125 C, used at 55 C, 0.7 eV, no failures, 60% confidence, about 50.9 FIT; hand-worked in the tool's unit test. FIT, upper bound at the confidence 51.0864; Average acceleration factor 77.6454; MTTF at the upper bound 2,234.55; Chance of failure over the mission, upper bound 4,465.17.
Versions
- v1 (2026-10-25): First version, built from the tool idea.
Terms on this page
- FIT
- The standard failure rate unit, one failure per billion device hours of operation.
- MTTF
- The average time to the first and only failure of a non-repairable item such as a semiconductor die.
- Junction temperature
- The temperature of the active silicon itself, as opposed to the case, board or ambient.
- Silicon
- The group IV element that carries almost all commercial semiconductor volume, valued for its abundance, its stable native oxide and a manufacturing base refined over sixty years.
- Acceleration factor
- The ratio of time to failure at use conditions to time to failure at stress conditions, derived from temperature, voltage, humidity or cycling range models.
- Bathtub curve
- A plot of bit error ratio against sampling phase across the unit interval, shaped like a bathtub.
- Failure mechanism
- The physical, chemical or electrical process that causes a failure, such as electromigration, time-dependent dielectric breakdown, corrosion or solder fatigue.
- Activation energy
- The energy barrier of a failure mechanism, in electron volts, that sets how strongly its rate depends on temperature in the Arrhenius model.
Tell us what you're after and leave your email: we'll connect you with the right person on the Hub.