Electronic assemblies are smaller than ever before, sometimes making it impossible for even microscopic and X-ray post-production inspections to identify defects. Multiple-layer construction further hampers inspection. For these reasons, screening is required to prove that no defects or weaknesses exist.
During screening, power must be applied while outputs and test points are monitored and recorded. Any anomaly, even a nanosecond in duration, during a period of electrical or physical stress can cause rejection.
Of the various physical stresses, random vibration and rapid thermal ramping are the most effective. Many kinds of shakers, more formally called vibration exciters, are used in product evaluation that typically need simultaneous rather than sequential multiaxis shaking. But today’s smaller assemblies have prompted the development of a new technique—broad-spectrum acoustic excitation.
Evolution From Sequential-Axis Shaking
The earliest types of shakers were mechanical and vibrated a device under test (DUT) in only one axis, usually vertical. Users had to reorient the DUT sequentially into other axes. The earliest electrodynamic (ED) shakers were used similarly, even though real-world vibrations are simultaneously multiaxis.
The multiple electrohydraulic (EH) shaker system simulates multiaxis road-surface and driving-maneuver situations. Alternatively, such systems excite platforms on which major automobile assemblies such as seats and instrument panels are vibrated in multiple axes. Other multiple EH shaker applications include seismic simulations to test such things as nuclear power plant controls that must survive earthquakes.
The U.S. military sanctioned a few multiaxis ED shaker systems to drive weapons test platforms. These very costly arrays of eight shakers can be programmed for six degrees of freedom (6DoF): three translations and three rotations.
These systems could be scaled down to vibrate small loads such as electronic assemblies. However, each shaker would require its own power amplifier and complex individual digital control. As a result, the military system could be too costly and difficult for electronic manufacturers to use.
Pneumatic Repetitive-Shock Platforms
Multiaxis vibrating platforms driven by inexpensive pneumatic repetitive shock (RS) hammers have become popular in electronic manufacturing. IBM pioneered the RS shaker as a final-inspection tool to perform single-axis hammering on the frame of a card reader. Much warranty expense and user unhappiness were saved by finding manufacturing defects at the factory rather than in the field. Westinghouse and General Dynamics were among companies that extended the idea to weapons, using multiple RS units attached at various angles to hammer on electronic assemblies and missiles.
While at Hughes, Dick Baker, founder of Screening Systems, attached several RS units to the softly sprung bottom of a thermal chamber. Since the RS units were connected at several angles, the DUT received X, Y, and Z inputs. The phase between thumpers varied to give 6DoF mechanical stimulation. Incidentally, no one pretends that this simulates any usage environment.
Post-production environmental stress screening (ESS) was the first application of RS machines combined with rapid thermal ramping. Designers began using this equipment during design and development to accelerate product aging in a highly accelerated life testing (HALT) process.
Post-production testing for latent defects was accelerated by increasing random-vibration levels and thermal stressing. This became known as highly accelerated stress screening (HASS).
Limitations of RS-Driven Platforms
A major concern on RS-driven platforms is spatial differences. Visualize a triaxial accelerometer every 3² over a 48² × 48² platform. This is almost 300 triaxial accelerometers and 900 data acquisition channels and displays. The acceleration levels at these points will differ dramatically.
Figure 1 (see the August 2001 issue of Evaluation Engineering) shows vertical-axis rms platform acceleration values averaged over a small interval. Similar differences exist in the horizontal axis.
Why is this so? Mainly because there are relatively few thumpers to drive the platform, and they are some distance apart. Immediately over a thumper, repetitive shocks will be more severe than at a location halfway between thumpers. RS platforms are significantly less rigid than are the tables of ED shakers.
A second consideration with the RS platform is extra acceleration peaks. Figure 1 illustrates rms acceleration levels with a standard deviation of 1 s. Momentary peaks during random vibration tests on ED shakers often are limited to 3 s.
RS machines, however, create some severe momentary peak accelerations, even reaching 12 s. This is much more severe than on ED shakers. It is difficult to predict when these extreme peaks will improve screening and when they may cause unwarranted damage.
Now move to the frequency domain. Unlike ED machines whose users can control the applied spectra, RS systems have no spectral control. The DUT receives the stimulation that is built into the RS platform. Departures from a desired spectrum can exceed ±10 dB. These machines excite various high-frequency resonances within the DUT but are relatively weak at low frequencies.
Acoustic Forcing Advantages
Acoustic forcing offers many advantages over the ED and the RS approaches for inducing random-vibration responses in electronic hardware on PCBs. High-intensity noise has been used for several decades to mimic launch and transonic flight conditions aboard rockets and high-performance aircraft.
Flight-hardware vibratory responses to acoustic pressures are far more realistic than can be achieved with shakers. High-velocity gas flow is modulated by servo valves, then passes into a hard-walled or a reverberant chamber where test hardware is suspended and intensities can approach 180 dB.
Lower-intensity acoustic testing can be achieved at low cost with loudspeakers. This is the basis of the ESS acoustical device (ESSAD).
The ESSAD is very effective in vibrating relatively thin PCBs. A board has several natural frequencies (fn). When exposed to relatively intense sound (125 dB typical) that contains matching forcing frequencies (ff), the board flexes.
Figure 2 (see the August 2001 issue of Evaluation Engineering) shows a typical vibration spectrum on a telecom board. The gold plot indicates the target spectrum, and the blue line shows the response. In flexing, the board applies alternating forces to wiring traces, components, sockets, soldered connections, and disconnects.
Figure 3 (see the August 2001 issue of Evaluation Engineering) shows 21² and 10² speakers driven by a 5-kVA power amplifier. The speakers do not overheat at room temperature with continuous operation. Also, they operate at -70o C to +70o C.
A Case History
To illustrate the use of the ESSAD for acoustic screening, several accelerometers were attached at various locations on a PCB. The test engineer identified fn and peaks in the spectrum. This feedback was used in a closed loop to control speaker output so mechanical resonances are sustained.
A validation procedure assured that good PCBs were not harmed. However, laboratory personnel were alerted to any momentary or permanent malfunction. Then failure analysis was used to identify the root cause of each.
What kinds of defects did the ESSAD precipitate? The list begins with SMT solder joints on discrete components and IC leads, plated through-hole solder joints on components and connectors, and defective or broken components. Other discoveries were loose fasteners and defective solder connections.
Thermal Stressing With the ESSAD
In addition to acoustic forcing, stress screening of PCBs calls for thermal stressing by a high-velocity flow of conditioned air. Electrical heating alternates with refrigeration, either mechanical or liquid nitrogen. Temperatures on the DUTs ramp at greater than 40oC per min.
ESSAD-to-DUT attachments have low thermal inertia, which permit fast thermal ramp rates. This contrasts with the thermal performance of aluminum or magnesium fixtures used to attach DUTs to massive aluminum or magnesium RS platforms and even more to the massive aluminum or magnesium ED shaker armatures.
ESSAD Random-Vibration Advantages
The spectra of both the ED systems and the ESSAD can be quite closely (±1 dB) established and maintained. Many users are familiar with the Willoughby spectrum, 0.04g2/Hz from 80 to 350 Hz, sloping downward at 3 dB to 20 Hz and to 2,000 Hz with root area 6g rms. With ED shakers as well as the ESSAD, energy is used more efficiently if the spectrum is low and concentrated in relatively narrow bands of ff surrounding DUT fn. The ESSAD achieves vibratory responses as high as 20g rms with relatively low off-peak vibration.
Note the capability of both the ED shaker and the ESSAD to excite a DUT with a pure tone (sinusoidal wave). This can be useful if you want to identify fn with greater precision than can be achieved with random vibration or wish to dwell at a resonance to study a response mode. Using the ESSAD, sine responses nominally reach 10g peak and have been observed as high as 100g.
Conclusion
The ESSAD is a valid alternative to ED shakers and RS platforms at the PCB level (Table 1, see the August 2001 issue of Evaluation Engineeering). While the ff used to date is 30 Hz to 2 kHz, this range likely will be extended to 20 kHz. That may be increasingly useful as hardware continues toward smaller physical size and higher fn.
Additional Information
- Tustin, W. and Gray, K., “Don’t Let the Cost of HALT Stop You,” EE-Evaluation Engineering, September 2000, pp. 72-78.
- Grothues, E., Technical Brief: Accelerated Tests—Descriptions and Applications, October 2000, U.S. Navy, [email protected]
- Lafleur, F., Development of an ESS Acoustical Device—ESSAD, IEEE Workshop on Accelerated Stress Testing, Boulder, CO, Oct. 4, 2000.
- Canadian Patent CA2276693, and International Patent PCT/CA/00784. U.S. patent pending.
About the Authors
Wayne Tustin heads the Equipment Reliability Institute, providing education and consultation on the science of vibration and shock testing. He began his vibration experience on the Boeing XB-52. Equipment Reliability Institute, 1520 Santa Rosa Ave., Santa Barbara, CA 93109, 805-564-1260, e-mail: [email protected].
Francois Lafleur is manager of testing research at the Centre de Recherche Industrielle du Quebec. He has a master’s in physical engineering at the Polytechnic School of Montreal and is a Ph.D. postulate in the field of novel ESS techniques. Centre de Recherche Industrielle du Quebec, 8475 Avenue Christophe-Colomb, Montreal, Quebec H2M 2N9, Canada. 514-383-1550, ext. 3519, e-mail: [email protected].
Return to EE Home Page
August 2001