Acceleration Factors and the Importance of Field Failures

Electronic circuits are designed and manufactured with an expected product life. The electronics industry employs accelerated life tests that take days or months but are supposed to mimic years of use in the field. These accelerated tests are supposed to find the potential issues in a short amount of time. It can be difficult to verify the design of a circuit card that has an expected product life spanning several years when the test schedule for evaluating the circuit card is significantly shorter.

Calculating Acceleration Factors

An acceleration factor (AF) can be calculated between any two conditions of use. It is most common to apply this calculation to accelerated life tests. The acceleration factor is supposed to be a constant that multiplies the time scale of the test condition to the time of the field condition.

The acceleration factor calculation is:

AF = (Time to Failure in the Field)/(Time to Failure in the Test)

The acceleration factor calculation relies on obtaining the time to failure in the field and comparing it to the time to failure in the test. The inputs to the calculation assume that the field and test failures are counted and analyzed. The circuit cards that are being tested and the circuit card in the field need to be monitored for time to failure and a failure mode verification needs to be performed for every failure.

Example: Test-to-Field Acceleration Factor

Let's assume a simple case where a circuit card is in the FIELD and has a 30°C change in temperature. This temperature cycle occurs once per day. The board is tracked over 10 years and yields a Time-to-Failure (TTF) of 5 years (to make the math simple).

This means that 63% of the units in the field have failed in 5 years. The shape factor in this Weibull distribution happens to be 3.

A TEST is performed on the same example circuit card for 1,000 temperature cycles at 1 cycle per hour. The TEST cycle has a 140°C change in temperature from (-40)°C to 100°C. The failures on the TEST board indicate a TTF of 500 hours.

This means that 63% of the units in the test have failed in 500 hours. The shape factor of this Weibull distribution happens to be 7.

The resulting acceleration factor calculation is:

This means that for these two conditions the acceleration factor is 87.6. A harsher field condition would have a smaller acceleration factor. The calculation does not provide meaningful information about other test or field conditions, however.

Failure Mode Verification

Acceleration factor calculations are only valid when the failure mode in the one condition can be demonstrated in the other condition. In the example above, we would have to verify that the test failures are the same as the field failures.

It is impossible to use a life test to predict behavior in the field without verifying that the test causes the same failures.

A common mistake is to predict the probability of failure of a board due to temperature cycling by performing a vibration test. Solder fatigue due to vibration in a test is not going to predict solder fatigue due to temperature cycling in the field.

It may also seem obvious that failures are needed in order to predict a probability of failure. Any "test to pass" activity would yield no failures, no time to failure calculation, no failure analysis, no failure verification, and no acceleration factor calculation. Field failures are also essential for the same reason, but can be more difficult to obtain. Both the test and field circuit cards need to be monitored in such a way that allows the time-to-failure calculation to be performed.

Summary

Acceleration factors are calculated between two conditions. The failure mode in both conditions needs to match in order to calculate the acceleration factor. The acceleration factor is only valid for those two conditions (different field conditions have different AF). A test without failures cannot be used to calculate a time-to-failure.

Copyright Gil Sharon July 17, 2017. All rights reserved.