
ESG POWER
ACCELERATED LIFE TESTING
GUIDELINES
TABLE OF CONTENTS
1.0 Purpose 3
2.0 Planning 3
3.0 Designing 3
4.0 Test Units 4
5.0 Data Analysis-Probability Plot 4
6.0 Data Analysis-Relationship Plot 4
7.0 Example [1] 5
1.0 Purpose
Exposing test items to a systematic increase in an environmental stress factor precipitating system failure for the purposes of estimating normal use life. HALT provides a mechanism of testing that allows system failure mechanisms to manifest themselves in short periods of time.
2.0Planning
The Supplier will provide the following test specifics prior to the start of HALT:
1.1Parametric values that will be measured.
1.2Thermal stress levels; constrained to precipitate field failure modes only.
1.3Define product configuration; test units should be either pilot production or early production models.
1.4Accelerated thermal stresses will not exceed maximum Supplier rated design limits.
3.0Designing
Thermal stress levels will overlap the normal system operating ranges as shown below:
4.0
Test Units
A total of 20 units will be allocated to HALT; nine units at the lowest level and seven and four at the higher levels. Three units will be considered a minimum for the higher levels of stress. Given that commercial grade electronic components are generally utilized in all DELL designs the recommended thermal stress levels should be 125 C, 150 C and 175 C. The maximum operating temperature will be established as 85 C.
5.0Data Analysis-Probability Plot
The operating performance of the test units will be recorded in terms of time-to-failure. The time to failure for most electronic and electromechanical devices is typically characterized as a Weibull or Lognormal probability density function (pdf). The pdf defines the distribution of failures with respect to operating time. The supplier will generate the probability plot as follows:
5.1Rank the failure times from first to last for each level of test stress. Non-failed units will be last in the ranking.
5.2For each failure time, defined as rank (i), calculate the plotting position as
where n= the total number of units under test at that level.
5.3Plot P versus the failure time for each failure at each stress level on applicable probability graph paper, e.g. Logarithmic or Weibull.
5.4Visually plot lines through each set of points for a given stress level, for example, for the three stress levels there will be three separate lines. The lines should be plotted such that all three are parallel.
5.5If the lines can not be plotted in parallel fashion then, an investigation of the failure modes is in order and a retest condition probably exists.
6.0Data Analysis-Relationship Plot
The relationship plot is developed from the median life values from each of the three parallel lines for each separate stress level and temperature. There are a total of three stress levels. The relationship plot is constructed as follows:
6.1On a log-log scaled graph representative of the assumed relationship, e.g. (Arrhenius), plot the 50% points determined from the probability plot in section 5.0, for each of the three stress levels.
6.2Plot a single line through the three 50% points projecting beyond the upper and lower points.
6.3Plot a horizontal line at the normal stress temperature limit.
6.4Note the intersection of the plotted line through the normal stress temperature horizontal line. At this point, 50% of the units will fail while operating under normal use conditions.
6.5Plot the time determined in step 6.4 above on the probability plot developed in section 5.0. Draw a line through the time point that is parallel to the other three lines. This line represents the distribution of failures in a normal stress level environment.
7.0Example [1]
Given a test unit that demonstrates an Arrhenius performance/stress relationship. The expectation is to determine the unit’s reliability in Mean Time Between Failure (MTBF) at 90 C (maximum operating temperature). 20 unit’s are tested. The unit’s are allocated to three stress test levels and the resulting failure times are as follows:
| 9 units @ 150 deg C | 7 units @ 180 deg C | 4 units @ 230 deg C | ||||||
| Time to Failure (Hrs) | Rank | P | Time to Failure (Hrs) | Rank | P | Time to Failure (Hrs) | Rank | P |
| 567 | 1 | 5.5 | 417 | 1 | 7.1 | 230 | 1 | 12.5 |
| 688 | 2 | 16.6 | 498 | 2 | 21.4 | 290 | 2 | 37.5 |
| 750 | 3 | 27.7 | 568 | 3 | 35.7 | 350 | 3 | 62.5 |
| 840 | 4 | 38.8 | 620 | 4 | 50.0 | 410 | 4 | 87.5 |
| 910 | 5 | 50.0 | 700 | 5 | .3 | |||
| 999 | 6 | 61.1 | 770 | 6 | 78.6 | |||
| - | 7 | - | 863 | 7 | 92.9 | |||
| - | 9 | -* | ||||||
[1] RBPR-3, “Reliability Analysis Center, Blueprints for Product Reliability”, Rome NY, Oct 1996, Revision 1.0, pp 66-70.
