188.8.131.52 HALT Testing
The term HALT was coined in 1988 by Gregg K. Hobbs (Ref. ). HALT
(also, sometimes referred to as STRIFE (Stress plus Life) testing) is a
development test, an enhanced form of step stress testing. It is typically
used to identify design weaknesses and manufacturing process problems and to
increase the margin of strength of the design rather than to predict
quantitative life or reliability of the product.
HALT testing begins with step stress testing in generic stresses
such as temperature, rate of change of temperature, vibration, voltage, power
cycling and humidity. In addition, product unique stresses such as clock
frequency, DC voltage variation and even component value variation may be the
The type of the vibration stimuli
used for HALT (and HASS) testing is unique. It is not based upon the
universally accepted accelerated (power) spectral density concept. Thus it
does not utilize classical, single-axis, sinusoidal vibration or a random
vibration spectrum, generated by acceleration-controlled electro-dynamic
shakers. Instead an unconventional multi-axial pneumatic (six degree of
freedom) impact exciter is typically used. This type of equipment generates a
highly unique broadband accelerated shock response spectrum (SRS). This is
effectively a repeated shock environment rather than a vibration environment
and is, in its self, much more severe than a classical vibration spectrum.
Because of the choice of this shock stimuli spectrum, the resulting data
cannot be easily correlated with either: (a) the normal environment or with
(b) classical vibration testing using classical vibration modeling approaches.
Thus quantitative prediction of life or reliability is not usually possible
with HALT and HASS.
Using HALT the step stress process
continues until stress levels well above those expected in the normal
operational environments are exceeded. Throughout the process continuous
evaluation is performed to determine how to make the unit able to withstand
the increasing stress. Generally temporary fixes are implemented just so that
the test can continue. When a group of fixes is identified, a permanent block
change is then implemented.
After one stimuli has been elevated
to a level felt to be sufficient, another stimuli is selected for step stress
testing. This progression continues until all stimuli have been applied
separately. Then combined stresses are run to exploit the synergism between
the stresses, that is, the combined effect may generate larger stresses than
either stress alone would create. After design fixes for the identified
problems have been implemented, a second series of step stresses are run to
verify the fixes, assure that the fixes themselves have not introduced new
problems and to look for additional problems which may have been missed due to
the limited sample size. This aspect of HALT must be taken into account in
selecting the appropriate stress levels since a slight increase in stress can
greatly reduce the number of cycles to failure.
For all of these stimuli, the upper
and lower operating limits and the destruct limits should be found or at least
understood. Understood means that although the limits are not actually found,
they are verified to be well beyond the limits which may be used in any future
HASS test and even farther beyond the normal field environments. For example,
a product may be able to withstand an hour of random vibration at 20
Grms without failure. Although the destruct limit may not
have been found, it is certainly high enough for most commercial equipment
intended for non-military environments where the screen environment may be 10
Grms random vibration for 5 minutes and the worst field
environment is a truck ride while in an isolation container. This example of
the capability far exceeding the field environment is quite common when HALT
is properly applied.
There are several reasons for
ascertaining both the operating limits and the destruct limits. Knowledge of
the operating limits is necessary in order to assess if suitable design
margins exist and how large the margins are likely to be as a function of
population. It is also necessary to formulate failure detection tests. These
can be run during any future HASS test since the detection tests run during
stimulation are necessary for high detectability of precipitated defects.
Knowledge of the destruct limits is required in order to determine the design
margins in nonoperating environments and to assure that any future HASS
environments are well below destruct levels.