Vibration Testing: Shake, Rattle, and Roll—No Shock!
Vibration testing is one of the key tests that is required in qualification testing. The basic purpose of this test is as follows:
The purpose of vibration testing is:
To determine the effects of vibration within the predominant vibration frequency range and magnitudes that may be encountered during the life of the connector.
To determine if electrical discontinuities at the level specified exist.
To determine the magnitude of axial movement between mating connectors due to vibratory considerations, if applicable.
To establish the mechanical integrity of the connector system exposed to external mechanical stresses.
To evaluate the impact on electrical stability of the contact systems when micromotion between contacting surfaces may be induced by mechanical means (fretting corrosion).
The above sounds simple enough. The perception is that all one has to do is to mount the connectors to the vibration, select a severity level from the test procedures, and perform the test. This is a bad perception.
There are several factors that have to be considered prior to initiating such tests.
- Vibration—should sine or random vibration be used?
- What is the right severity level?
- What is the right duration?
- What variables should be monitored during vibration or performed after the test?
- How should the units be fixtured?
The test procedures are essentially default levels. Sine vibration, by its nature, does not simulate the real world. It simply sweeps through a given range in a sinusoidal manner. When the procedure was first established, 50 or so years ago, vibration levels were not well understood. It was known that there are generally at least two prime resonances that occur—one at high frequency (>1000 Hz), and one at low frequency (<1000 Hz), but the specifics were generally unknown. It was thought that by simply sweeping through a large enough frequency range, the prime resonances would be applied. However, due to the sweep-times specified, the prime resonances will only be applied for a very small fraction of the time. It is also known, that if the resonance frequency of the DUT (device under test) is not the same as the operating frequency, not much will happen (the contact will not move). If a resonance does occur, it could very easily be in a different part of the connector.
Random vibration, on the other hand, is closer to simulating real world conditions. The only problem is selecting the right severity level. A given average amplitude (expressed as GRMs value) or PSD, is selected and represents the average across a frequency band. There is also generally a crest factor of three associated with it. Therefore, if a 6.0 GRMS is selected, it will have a maximum amplitude of 18 Gs, and a low amplitude of 2 Gs for short periods of time, but averaged at six across the band occurring in a random manner throughout the band.
Ten to 15 years ago, sine vibration was the test of choice about 80 percent of the time. Today that percentage is about 60 percent. Many user agencies are changing over with the predominant frequency range being 20-500 Hz. This has occurred as a result of the following:
- Dissatisfaction that sine vibration does not replicate field problems.
- Failures due to vibratory conditions will happen very quickly, so that the common durations used with sine vibration can be significantly reduced.
- The lower frequencies with higher amplitudes are where the predominant failures occur, be that relating to contacts or the mechanical integrity of the DUT.
This shift is continuing as more user agencies give thought to this problem. The military, automotive, and aircraft industries have pioneered this process. For several years, the military has provided the agency with a document (MIL-DTL-810) which has different profiles depicting various applications, such as road vehicle, open field, etc. Some aircraft companies have three different profiles, one for each axis which has been created by the analysis of operating equipment.
One must realize that there are significant differences between system-level vibration and component severity levels. Some think that vibration severity for components should be at the system level. However, the component can react differently than the system due to the amplification which is generated. Contingent on how and/or where a component is located within a system, the vibration/amplitude can magnify up to 10 to 20 times. The basic issue is that this tends to vary from system to system, mounting techniques, etc.
The above eludes to an additional problem—fixturing the DUT for the test to be performed. Very few test specifications specify the fixturing to be performed and leave it up to the testing lab to create the fixturing involved. This can result in under- or over-testing. A part should be fixtured to simulate how its mounted to its equipment relative to mounting features, whether both or one-half is to be restrained, etc.
Vibration testing is performed in the X, Y, and Z axis, as defined in EIA 364, Test Procedure 28. Each axis has its own dynamic. The two most damaging axis are the “Z” axis (in a direction to unmate the connectors), and the “Y” axis (perpendicular to the longitudinal length of the connector). The “Z” axis can result in fretting by moving the contacts in the mating/unmating direction, assuming the energy dynamics are great enough. The “Y” direction can result in a rocking motion, which also can result in fretting (applies mainly to daughtercard/mothercard board applications).
It is common to monitor for 1.0 microsecond interruption during vibration testing. Quite frankly, the connectors would have to be physically damaged, or resistance changes have to be well into the ohmic region, or contact chatter has to exist (actual lifting off the contact surfaces) for this to be possible.
A more meaningful monitoring test has been developed called low nanosecond event detection. This technique is used when high-speed devices are being used. It will detect if a 1.0 volt shift occurs in the event of a resistance spike, for a very short period of time. This technique is described by EIA 364, Test Procedure 87. It is unfortunately being misused due to misunderstanding the intent of the test. It also requires significantly different equipment: special characterization with detector channels committed to monitoring only a few contacts per channel (one or two). For high-density connectors where a high-contact population exists, it would be impossible to monitor for this type of variable (due to equipment limitations) through a series circuit. Thus, monitoring end and middle positions (generally where problems will develop) is reviewed. This test is used when high-speed logic applications are involved.
Low-level circuit resistance (LLCR) is another variable that is an effective measurement, which can and should be performed. If fretting occurs, a significant resistance change can occur. When such a change occurs, it will not usually recover. It also may be of such a magnitude that the discontinuity monitoring will not detect it if it’s less than the triggering resistance of the detector. Thus, one can pass the discontinuity requirement and still have non-functional contacts go undetected.
For critical applications, consideration should be given to monitoring LLCR after each axis, since each axis has its own dynamic. It has been shown that the contacts may remain stable in two axis, and fail in the third. Contingent on the order in which the axis are performed, the problem could be missed if LLCR is performed only after the vibration test has been concluded.
The intent of the above comment is to indicate that specifying a meaningful vibration requires careful thought prior to testing. Basic decisions have to involve the following:
· Sine or random vibration (the latter becoming the most popular).
· Duration—1.0 hr./axis duration is appropriate and recommended for the random technique.
· Monitoring Criteria—a 1.0 microsecond interruption or low nanosecond event detection (1.0, 2.0,
10.0, or 50.0 nanoseconds), but the latter should only be used in high-speed logic applications.
· Severity Level—this is a major problem area because the severity level could change, contingent on:
- System Construction
- Location of the DUT in the system
- Restraint and mounting features
- Robustness of the connector itself
- Frequency bandwidth expected
The above are interacting factors but also should simulate how the component is going to be used, or as close as possible to how it will be used. Manufacturers may wonder, “then what levels should we be testing to?” One should use a consensus approach. But do not be surprised if users require their own unique set of requirements, and may require additional testing. This is particularly true as electronics become a more segmented industry and worldwide applications evolve.
This topic could be covered in several additional pages. While I’ve attempted to highlight a few of the issues involved, the following are some other factors that should be considered.
- Low-level circuit resistance (LLCR) should always be performed after vibration, and in certain applications, after each axis.
- When complex fixtures are used, model analysis should be performed.
- Control accelerometer should be placed on the table, not on the samples. An additional monitoring accelerometer can be placed on the fixture or samples, if desired.
- When shock testing is used in sequence with vibration, it should be performed prior to, and not following, vibration.
I recognize that mechanical shock has not been discussed in detail. So I will now say that from my perspective, the shock tests currently being performed are not very meaningful, except as a mechanical integrity test. As a functional test, it simply doesn’t meet the need, and does require rethinking. It is a transportation test, not an operating test. My intent is to hopefully generate a dialogue as to how to improve vibration testing as a meaningful tool.
So, I will now sign-off, sit back and relax, and wait for the phone calls from San Diego to see if my friends there are still talking to me. Editor’s Note: If I’m not mistaken, I believe Max is referring to the fact that the Patriots beat the Chargers to qualify for the Super Bowl!
By Max Peel, Senior Fellow, Contech Research
- Nanocrystalline Silver Alloy Contact Finishes in Electronic Applications - April 6, 2015
- Nanocrystalline Silver Contact Platings - March 16, 2015
- Dr. Bob on Gold Flash Contact Finishes (and Max Peel) - September 22, 2014