Navmat P-9492
Navmat P-9492
Navmat P-9492
NAVY MANUFACTURING
SCREENING PROGRAM
MAY 1979
1
PREFACE
2
TABLE OF CONTENTS
PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
1.0 INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2.0 TEMPERATURE CYCLES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 Background. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 Scope of Survey.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.2 Number of Cycles and Equipment Complexity. . . . . . . . . . . . . 5
2.2.3 Duration of Temperature Cycles. . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.4 Equipment Operating Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.5 Temperature Range and Rate of Change. . . . . . . . . . . . . . . . . 7
2.2.6 Design Compatibility with Temperature Cycling. . . . . . . . . . . . 7
2.2.7 Types of Defects Simulated. . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.8 Repairs and Failure-Free Cycles. . . . . . . . . . . . . . . . . . . . . . . 8
2.2.9 Board Level Temperature Cycling. . . . .. . . . . . . . . . . . . . . . . . 8
3.0 RANDOM VIBRATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1 Background. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2.1 Verification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2.2 Application. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.3 Synthetic Random Vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.0 RECOMMENDATIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.1 Temperature Cycling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.2 Random Vibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.0 REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3
1.0 INTRODUCTION
Most Navy programs acquiring electronic devices and systems traditionally have depended
on the final acceptance test to catch manufacturing defects. They have relied on this screen as a
sufficient incentive to the manufacturer for the inclusion of additional pre-acceptance test screens of
many different forms in the production operation. Some contracts have called out specific pre-
acceptance tests (e.g., burn-in) for the primary or ancillary purpose of defect detection. For a
variety of reasons, both technical and contractual, the vast majority of electronic devices and
systems delivered to the Navy continue to contain manufacturing defects in parts and workmanship
which could have and should have been discovered and eliminated in the factory.
This publication provides guidance concerning the use of temperature cycling and random
vibration as manufacturing screens for defects in both parts and workmanship. The requirements for
such screens are called out in Navy instructions and reflected in contract requirements. Section 2.0
on temperature cycling is derived from a Martin Marietta report for the National Aeronautics and
Space Administration on industry experience in assuring long-life hardware. Section 3.0 on random
vibration has been prepared by the Grumman Aerospace Corporation under the direction of
the Naval Electronic Systems Command. It summarizes the experience of Grumman and
others supporting the NASA manned space program. Grumman recently has devised a
technique to simulate random vibration at low cost without a sacrifice in effectiveness as a
manufacturing defect screen. This technique is included as an appendix to this publication. Section
4.0 contains the minimum recommended thermal cycling and random vibration manufacturing
screens to be used in the production of Navy equipment.
4
2.0 TEMPERATURE CYCLING
2.1 BACKGROUND
5
2.2 FINDINGS
2.2.1 SCOPE OF SURVEY
The utility of temperature cycling as a workmanship screen is recognized by many aerospace
companies. The approach to this cycling, however, varies widely from company to company. Table 2
shows the degree of variation in the temperature cycling approaches employed by the 26 aerospace
companies and agencies surveyed. The table shows a lack of standardization in employing more
than one temperature cycle. It also shows a temperature range between –65 °F and +131 °F (-54 °C
to +55°C) to be most commonly used.
6
2.2.2 NUMBER OF CYCLES AND EQUIPMENT COMPLEXITY
Test and failure rate data provide some interesting insights into the most effective
number of temperature cycles to use for workmanship screening. Figure 1 provides a
comparative illustration of the number of failures as a function of the number of temperature cycles.
Six to ten thermal cycles are required to eliminate most latent workmanship defects.
Investigation of the equipment to which this data applied revealed a useful correlation
between equipment complexity and the effective number of temperature cycles required. Six cycles
appear adequate for black boxes of about 2000 parts, while 10 cycles are recommended for
equipment containing 4000 or more parts, as shown in Figure 2.
Hughes Aircraft Company has developed mathematical models to predict how many cycles
are required to achieve a specified reliability. The number depends on the previous amount of
screening the quality of parts used, and the exact thermal conditions and profile for the parts being
screened. A significant finding is that many more than 10 cycles are sometimes indicated by these
models. Similarly, when unscreened parts are used and temperature cycling of assemblies is
employed as the main production screen, more than 10 cycles may be required. Programs of 16 to
25 cycles are not unknown.
7
8
Accordingly, the dwell times at high and low temperatures need only be long enough
for internal temperatures to stabilize, a time which is dependent on the equipment design
and which will vary from one program to another. The AGREE cycle in Figure 3 is NOT to be used.
The equipment should be closely monitored during the operating portions of the
cycle. It is desirable to turn off the equipment during chamber cool-down, otherwise self-
generated heat will prevent the internal parts from reaching the desired low temperature.
Temperature ranges of –65 °F to +131 °F are the temperatures most commonly used. Most
parts will withstand temperature cycling with power off through a temperature range of –65 °F to
+230 °F. Heat rise with power on under test cooling conditions should be calculated to limit the
chamber temperature to a maximum safe value. The maximum safe range of component
temperature and the fastest time rate of change of hardware temperatures will provide the best
screening. The rate of temperature change of the individual electronic parts depends on the
chambers used, the size and mass of the hardware, and whether the equipment covers are taken
off. In general, the rate of change of internal parts should fall within 1 °F per minute and 40 °F per
minute, with the higher rates providing the best screening.
Temperature cycling with good parts and packaging techniques is not degrading even
with several hundred cycles. However, the packaging design must be compatible with the
temperature cycling program or the acceptance test yield will be reduced (to zero in some
special cases). This compatibility is established by temperature cycling the pre-production
hardware.
1) Electronic components assembled on printed circuit boards impose loads on the solder joint,
and temperature cycling may produce solder joint cracking. Heavy coats of conformal
coating on even a stress relief bend can negate the beneficial effect of the bends.
2) Transistors mounted on plastic spacers and coated with conformal coating will produce
cracked solder joints in a few temperature cycles if the leads are not stress relieved. This
problem arises because the coefficient of thermal expansion for plastics is about 8 to
30 times greater than Kovar transistor leads, or Dumet diode leads.
3) Large multi-pin modules soldered into the printed circuit board may result in solder joint
cracking, particularly if the conformal coating bridges between the module and the board.
4) Cordwood modules potted with a rigid, solid polyurethane or epoxy may produce cracked
joints and even crush weak parts such as glass diodes on the very first application of a
temperature cycle.
5) Filters, motors, and transformers containing fine wire (#40 or #50) may constitute a problem.
To avoid the problem, wire sizes larger than #40 should be used.
6) Single or double sided printed circuit boards without plated-through holes are undesirable.
7) Breakage of glass diodes can be expected if great attention is not given to the
encapsulating material and the process.
9
2.2.7 TYPES OF DEFECTS SIMULATED
An approximation of the types of failures detected in mature hardware by temperature cycling is:
Design Marginalities 5%
Workmanship Errors 33%
Faulty Parts 62%
These figures are based on the experience of eight manufacturers as illustrated in Table 3.
The concept of augmenting the black box temperature cycling with additional cycling at the
printed circuit board level should be considered. Hughes Aircraft Company, on one program, "stores"
their assembled printed circuit boards in a temperature chamber for one week during which time 158
temperature cycles are accrued. The boards are not powered or monitored. This comprises a very
cost-effective approach to reliability.
10
3.0 RANDOM VIBRATION
3.1 BACKGROUND
Historically, acceptance of electronic equipment was originally limited to a form, fit and
function appraisal through visual inspections and a functional "smoke" test conducted under room
environments. In 1957 the Advisory Group on Reliability of Electronic Equipment (AGREE), created
in 1952 by the Department of Defense Research and Development Board to "monitor and stimulate
interest in reliability and recommend measures that would result in more reliable equipment,"
published its recommendations. These included specific requirements for establishing environmental
test profiles to be used during reliability demonstration testing. It was also suggested that these
same conditions be utilized for acceptance testing of electronic hardware. Vibration was established
as one of the environments and was limited to a sinusoidal excitation of ±2g at a fixed non-resonant
frequency between 20 and 60 Hertz. This form of vibration persisted for years and was used, with
few exceptions, in the majority of electronics and avionic equipment acceptance tests conducted.
Evolving from the McDonnell Douglas Mercury and Gemini manned spacecraft programs,
random vibration was utilized to more effectively screen workmanship defects. The unprecedented
success of the Apollo manned space program, attributable in large measure to the intensive test
program (Reference 2), generated some new thinking in industry and the military concerning the
utilization of effective testing (including random vibration) in achieving reliability requirements.
Skeptics still maintained that, while those techniques might work for Apollo whose vehicles were
essentially "one shot" devices, they probably would not be effective for hardware (such as aircraft
avionics) which had to survive thousands of takeoff, flight, and landing hours. The Grumman
Aerospace Corporation decided at this time to investigate the merits of sine and random vibration
testing. Intuitively, it appeared that random vibration, which provides simultaneous excitation of
many modes in contrast to the single frequency sine test, must be more effective in disclosing
manufacturing defects. Dr. John Dreher of Wright-Patterson Air Force Base supported this intuition
in his paper (Reference 3) where he stated:
"While it is true that the associated sine sweeps do excite most of these other
resonances, one must consider the short time period spent in any one resonance bandwidth
and the fact that many of these resonances aren't excited long enough to peak out. In
contrast, the random test excites every resonance for the duration of the test."
It appears, then, that the random test proposed is a less severe but more thorough test.”
3.2 FINDINGS
3.2.1 VERIFICATION
The scarcity of random vibration application data prompted Grumman to embark in 1971
upon a laboratory test evaluation structured to directly compare the effectiveness of sinusoidal and
random vibration (Reference 4). A technical approach was conceived wherein the time-to-failure of
typically occurring defects could be examined under controlled environmental conditions and
selected durations. Typical workmanship defects, representing 80% of manufacturing problems
found in avionic hardware, were selected from Lunar Module (LM) and aircraft test and field failure
data. These defects were simulated in quantities considered sufficient for analysis and were inserted
into a typical avionic "black box". The test plan provided for a total of 100 simulated defects to be
included in any given test matrix of different levels and durations.
Tests were conducted using sine fixed frequency, sine sweep and random vibration
excitations at different levels and for varying periods of time. Figure 4, as an example, depicts the
matrix for sine sweep testing. Similar matrices were developed for sine fixed frequency and random
vibration tests.
11
The results clearly indicate that random vibration, at a 0.04 g^2/Hz level (Figure 5), was significantly
more effective than either of the sinusoidal tests. Figure 6 compares the effectiveness of the three
forms of vibration for two of the most common defect types at levels "typically" used in acceptance
testing. The results of this comparison are obvious. Figure 7 compares the "typical" random level
with a 5g level for each of the sine-type tests. The results show that even at increased levels, the
random vibration is more effective (for a given fault type) than sine fixed frequency or sine sweep. In
the Figure 8 comparison, levels of vibration up to and exceeding qualification were used for the sine
type of test. Although the sine sweep test was close to the "typical" random test for both failure
types, it required durations of approximately one hour at qualification levels (10g) to achieve this
type of effectiveness. Testing production hardware at these levels and durations would certainly
present a potential fatigue problem and would never be utilized in an acceptance test. The "typical"
random vibration spectrum achieved its maximum effectiveness in only 10 minutes of testing.
0.1
Overall Level = 6.0 grms
2
0.04 g /Hz
+ 3 dB / octave
-3 dB / octave
PSD ( g / Hz )
2
0.01
0.001
20 80 350 2000
FREQUENCY (Hz)
12
13
Some concern has been expressed that the application of a 0.04 g^2/Hz random vibration
level would cause fatigue and structural damage if applied to equipment even if that equipment had
proven its structural integrity during qualification tests. During the advanced development program
conducted by Grumman, a correctly manufactured example of each fault type was inserted in the
test article as a control. Even after many hours of exposure at the 0.04 g^2/Hz level, not one of these
correctly manufactured examples failed. Further, equivalency analyses performed by Grumman and
Wright-Patterson Air Force Base indicate that the 0.04 g^2/Hz level is much less severe than
qualification levels currently used. In his paper (Reference 3), Dr. Dreher points out that a fatigue
test level of Wf = 0.10 g^2/Hz is equivalent to a sinusoidal level of only Gf = +2.5g. He further
indicates that it takes a level of Wf = 1.6 g^2/Hz to be equivalent to a +10g sinusoid. It should be
noted that these equivalencies, developed analytically, apply universally to any type of equipment
undergoing vibratory excitation.
Additionally, Grumman has had extensive experience in the use of random vibration as an
acceptance test, workmanship screen and/or troubleshooting aid. During the LM program over 7,000
tests were performed. In all the history of random vibration applied at this level at Grumman, no
known instance of degradation or subsequent field failure attributable to the vibration test has
occurred.
3.2.2 APPLICATION
The results of the study conducted by Grumman have had wide distribution. A paper on this
subject was presented at the Joint Logistics Commanders' System Reliability Workshop in May
1975. The random form of excitation is now being used by many equipment manufacturers for
acceptance purposes. Mr. Wayne Tustin, internationally known vibration lecturer and consultant,
strongly endorses its use as an effective workmanship screen (Reference 5). Grumman's initial
experience with random vibration was first obtained on the LM program. Following the Advanced
Development study efforts described above, the technique was also applied to various avionic
equipment installed in Grumman aircraft. The use of random vibration on the LM was originally
limited to a level of approximately 39 RMS. In 1967, the NASA informed Grumman that their Gemini
experience indicated the level being imposed on LM equipment was not rigorous enough to detect
quality/workmanship defects. The level was increased to 6g RMS (currently used - Figure 5) and
approximately six months after implementation of the higher level, the effectiveness of the program
was assessed. The results indicated that a 3:1 reduction in equipment failures, during spacecraft
testing, had been realized.
Application of random vibration to avionic equipment was first accomplished on an airborne
computer. After approximately four seconds of random vibration exposure, a malfunction occurred
and subsequent analysis revealed several broken solder joints. It is significant that this unit had
previously undergone over 100 hours of the "classical" burn-in which included thermal cycling and
fixed-frequency sinusoidal vibration. Random vibration has since been applied to many avionic items
with a high degree of success. Table 4, typical of the results achieved, shows the delivered MTBF
improvement after random vibration was included as part of the acceptance test for these
equipments.
MTBF MTBF
MTBF
Equipment ADDITION OF RAND IMPROVEMENT
W/O RAND VIB (HRS)
VIB (HRS)
“A” 100-165 250 50 – 150%
“B” 125 380 200%
“C” 58 153 160%
14
3.2.3 SYNTHETIC RANDOM VIBRATION
The major deterrent to universal acceptance of random vibration is the impact this
type of test would have on program costs since a random vibration test facility is extremely
expensive. A concept for economically generating random vibration was evolved which capitalizes
on the fact that most major electronic equipment manufacturers maintain basic electrodynamic
sinusoidal vibration test facilities. This technique, which was structured to utilize these existing
facilities, employs a cassette tape deck, in lieu of expensive random programming devices, to excite
the basic shaker system.
The detailed procedures necessary for generating random vibration using an
electrodynamic vibration system and a cassette tape deck as a signal source are included herein as
an Appendix.
15
4.0 RECOMMENDATIONS
The temperature cycling screen should be conducted in accordance with the guidelines shown in
Table 5.
No. of Temperature
Type of Equipment
Cycles
Simple ( 100 electronic parts ) 1
Moderately complex ( 500 electronics parts ) 3
Complex ( 2000 electronic parts ) 6
Very complex ( 4000 electronics parts ) 10
Temperature Range
The suggested range is –65 °F to 131°F, or as a minimum, a temperature range of at least
160 °F is recommended.
Temperature Rate of Change
The rate of change of internal parts should fall within 1 °F and 40°F per minute. The higher
rates provide the best screening.
Temperature Soak Times
The next temperature ramp may be started when the internal parts have stabilized within 5°F
of the specified temperature and the functional checks have been completed.
Equipment Operation
Equipment should be energized and operated during temperature cycling, except the
equipment should be turned off during chamber cool-down to permit internal parts to become
cold.
Equipment Monitoring
While it is desirable to continuously monitor the equipment during the temperature cycling,
cost considerations may dictate otherwise. In such cases, periodic checks plus close
monitoring of the final cycles is appropriate.
Failure Criteria
The last cycle shall be failure free. Each repair should be reviewed for the possibilities of
introducing new defects into the hardware and additional temperature cycles added when
appropriate. If repairs are complex or difficult to make and inspect, or many unscreened
parts are used as replacements, additional cycles should be implemented as appropriate to
the individual case.
Table 5. Guidelines for Temperature Cycling Acceptance Testing of Electronic Hardware
16
4.2 RANDOM VIBRATION
The random vibration screen should require that the equipment under test be hard-mounted to
a shake table capable of reproducing random vibration having the power spectral density
characteristics shown in Figure 5 (repeated herein). A pseudo-random or synthetic random vibration
shaker capable of reproducing this power spectral density function will be an acceptable
substitute for a true random vibration fixture.
The equipment under test should be oriented on the fixture such that the axis of
vibration is perpendicular to the printed circuit boards. Where electronic components in the
equipment under test are oriented in more than one plane, such equipment should be shaken
sequentially in each of three orthogonal axes.
0.1
Overall Level = 6.0 grms
2
0.04 g /Hz
+ 3 dB / octave
-3 dB / octave
PSD ( g / Hz )
2
0.01
0.001
20 80 350 2000
FREQUENCY (Hz)
The duration of random vibration should be at least ten minutes if a single axis is sufficient.
Where vibration in more than one axis is required, the duration of random vibration should be at
least five minutes in each axis.
17
5.0 REFERENCES
1. Burrows, R.W., Long Life Assurance Study For Manned Spacecraft Long Life
Hardware, Report MCR-72-169, Martin-Marietta Corporation, Denver. September 1972.
2. Low, G., "What Made Apollo A Success," Technical Article, Aeronautics and
Astronautics, March 1970.
3. Dreher, J.F., "Aircraft Equipment Random Vibration Test Criteria Based on Vibrations
Induced By Turbulent Air Flow Across Aircraft External Surfaces," Technical Paper, Air Force
Flight Dynamics Laboratory, Wright-Patterson Air Force Base, Dayton.
4. Kube, F., and Hirschberger, G., "An Investigation To Determine Effective Equipment
Environmental Acceptance Test Methods," Advanced Development Report ADR 14-04-73.2,
Grumman Aerospace Corporation, April 1973.
6. R.E. Schafer, S.P. Gray, L.E. James, E.A. McMillan, "Electronic Equipment
Screening and Debugging Techniques," Rome Air Development Center Final Technical
Report RADC-TR-78 55, dated March 1978.
18