COTS Vibration Testing
by Wayne Tustin

Designs that use COTS (commercial-off-the-shelf, a military term) products must take into consideration the harsh environments in which these commercial devices may be used. One of the common sources of COTS system failure is vibration. COTS equipment vibration testing can mean the difference between system success and failure.

Vibration testing identifies design flaws and identifies potential failure before the equipment is fielded. To perform an adequate vibration test on an existing COTS circuit card or "black box" requires at least 6 pre-production units. These units will be subjected to a variety of experiments, including Step-Stress tests, development tests and temperature ramping.

Testing requires familiarity with three types of standard environmental laboratory equipment. Shakers are controllable source of vibration. Test fixtures, which are typically custom designs, physically mount the hardware to the shaker. Accelerometers are attached to the fixture to sense input vibration.

No MIL Standards

Just a few years ago, it was typical for military customers to specify a Test Method (such as 514) within a standard such as MIL-STD-810. Tables and charts would indicate test severities, frequencies and durations appropriate to various locations on aircraft, tanks, etc. And most commercial test labs would have known exactly what to do.

The COTS initiative has changed that. Now the norm is to test a commercial item that was originally intended for desk or similar non-threatening use. It’s hoped the commercial product will survive the warranty period while mounted on some MIL vehicle or other threatening application. Vibration testing can improve the odds that the device will survive. Numerous commercial test labs can provide experienced test personnel, a shaker system and appropriate mounting fixtures.

Step Stress Testing (SST)

Step-Stress Testing (SST) is a type of environmental test. Let's examine the vibration portion. For SST the product is mounted on a fixture which is attached to a shaker. Supply power to the product, and turn it on. After power-up, monitor some of the hardware’s functionality to determine if it fails even momentarily. If the product is functioning, the shaker is started, preprogrammed for 1g RMS random vibration 10-2000 Hz. At this point the hardware is observed to identify any misoperation. Misoperation in the product indicates trouble. If finger pressure at some location restores operation then there is a poor connection that needs to be identified. If a resonating part moves, friction or damping needs to be added. If a part flies off it needs to be secured with epoxy and cord. Experienced lab people have seen similar failures and will be able to suggest "Band Aid" fixes. Shake the product again (on the same unit or substitute another unit) at 1g RMS. If it survives at 1g RMS no further testing is required.

Is Developmental Testing Allowed?

If the "Band-Aid" fixes don't work, further testing, properly called development testing, is needed. Management approval for further testing should be pursued, because another phrase for COTS is NDI (non-developmental item) and there may be at least a temporary edict against development testing.

It is important to understand how the hardware responds to all-frequencies-at-once random vibration. This is best done by one-frequency-at-a-time sinusoidal testing. The shaker is reprogrammed to slowly sweep, for example 5 to 2,000 Hz. 1g RMS intensity is held, measured by an accelerometer on the fixture. Several response accelerometers should be mounted at any critical locations on the hardware. These outputs are recorded. Resonance is the problem at vibration frequencies where it is found (a strobe light is helpful) that response>input or where response accelerations>1g.

Don't Stack Resonances

An important commandment in dynamics is "Thou Shalt Not Stack Thy Resonances!" That is, when a printed circuit card (PCB) is discovered to have certain natural flexing frequencies, it should not be in a box whose natural frequencies are nearby. The box should not be located on a platform where structural vibration approaches PCB or box natural frequencies. In the best case, the platform vibration is already known and can be determined. If resonances are stacked, some design changes will be needed which are outside the scope of this article.

Back to SST

Once the hardware operates correctly at 1g RMS with random vibration, it should be rotated 90o into another (and later the third) axis. Obviously, simultaneous multi-axis shaking will save handling.

At some point, when the hardware can operate satisfactorily at 1g RMS random in any direction, the vibration should be increased. The intensity should be step-increased to 2g, 3g, etc. in each axis. As before, everything that fails needs to be fixed. The product should be made as rugged as possible in the available time.

Combine with Temperature Ramping?

Step-stress testing is even more effective when combined with rapid (20 to 50o C per minute) thermal ramping, stepwise increasing temperature extremes toward 50o. Rapid expansion and contraction of PCBs and components helps to reveal latent or hidden defects. It is better to find defects in the lab than in the field, under warranty. Temperature ramping also ensures operation over a range of field temperatures.

How much is enough?

This question is not easy to answer. The general rule is that more is better. The good news is that changes that dramatically increase dynamic ruggedness are often achieved with zero increase to manufacturing cost.