Highly Accelerated Stress Screens (HASS) is a production test process that has become a widely accepted method for detecting process or component changes in a product that could reduce product reliability. It is sometimes used in conjunction with, and often as a replacement for, classical burn-in. HASS uses vibration and thermal stresses to quickly force latent defects in a product to fail or to become detectable through marginal performance, thus increasing the out-of-box quality and reducing warranty expenses.

The stress profiles that are used in HASS are based on the results of Highly Accelerated Life Tests (HALT) on the product. During HALT, the operational and destruct limits for the product are determined. The levels for HASS are defined by appropriately guard-banding these results so the screen can be safely applied to production units without over-stressing good products, while still detecting latent defects in marginal products.

Implementing a safe and effective HASS requires a proof of screen (PoS). This process has two parts:

• Screen effectiveness, which demonstrates that the screen will effectively catch marginal products.
• Safety of screen (SOS), which demonstrates that the screen will not damage good products.

Screen effectiveness is demonstrated by subjecting products to the screen that are expected to fail. These would be products with seeded defects or with known design flaws, or products that have been returned from the field as defective but have passed functional tests in-house.

SOS is performed by subjecting a sample of known good products to the proposed screen 20 to 50 times. Then, the products are subjected to a full design verification test (DVT). The ability of a good product to survive, for example, 20 cycles through the proposed screen has been used as evidence that the screen will remove less than 1/20 of the life from the product.

In both HALT and HASS, repetitive shock (RS) vibration systems are used to create the vibration stresses. RS vibration will, by design, fatigue a product much more quickly than vibration from an ED system. The vibration is pseudo-random and broadband, with spectrums that have energy into the 10,000 Hz range. An RS system excites the product in three axes simultaneously, as well as the rotations around each of those axes, resulting in 6 degrees of freedom (DoF) fatigue. The high peak accelerations when an actuator strikes the table cause components to respond by ringing at their own resonant frequencies, even though those frequencies may not be present in the power spectral density (PSD) graph as measured at the table surface. The broad spectrum and high acceleration shocks that characterize a repetitive shock system are very unlike the ED vibration that is typically used in DVT or is specified for use in ESS.

Because the product stress induced by an RS system is so different, there can be a lingering concern that HASS may remove too much life from the product, in spite of the results of the safety portion of the PoS. The classical models used to calculate fatigue acceleration with vibration are not directly applicable to RS systems, so the actual product life consumed in HASS, measured in weeks or months, are unknown. However, these classical methods of calculating stress acceleration can be used to demonstrate the effectiveness of PoS as a tool to estimate the fraction of product life consumed in HASS.

Accumulated fatigue and the S/N curve

When a product experiences a cycle of stress, fatigue is induced. As more cycles of stress are applied, this fatigue accumulates. Wöhler’s equation describes the number of cycles of a stress, at a particular level, that are necessary to accumulate sufficient fatigue in a material to cause failure.

It is:

N = cS-β

Where:

• N = number of stress cycles to failure
• S = stress amplitude
• β = the device fatigue parameter
• c = a proportionality constant unique to the stresses being used

The classic S/N curve is the result of graphing Wöhler’s equation (See Figure 1). The blue line represents the number of cycles, at a given stress, that will result in failure.

Figure 1: S/N curve. Source: ESPEC

Wöhler’s equation describes the cycles to failure at a single stress level. The Palmgren-Miner linear damage accumulation theory (AKA Miner’s Rule) states that the fatigue due to stress accumulates linearly, regardless of the stress level used to induce that fatigue. Using Miner’s Rule allows the calculation of the total fatigue experienced by a product that is subjected to different levels of stress if the number of cycles of stress at each level is known. The fatigue at each stress level is calculated, and then the total fatigue is calculated as the sum of the individual components. The fatigue for a given number of stress cycles is expressed as the ratio of the number of cycles used to the number of cycles at that level necessary to induce failure. Thus, a fatigue level of 1 is fatigue to failure. From Wöhler’s equation, this can be written as:

Fatigue fraction at stress (Si) and number of cycles (Ni) = Ni/N = Ni/ cS-β = (1/c)NiS β

These are the tools that will be used to calculate the effect of HASS on product life.

Creating an S/N curve for HALT

Because HALT drives the product to failure using specific, measurable stresses, the time to failure in HALT vibration can be used as a point on the S/N curve for the product. An S/N curve can be drawn based on this information, with a few reasonable assumptions. Note that this S/N curve will describe time to failure if the only stress considered is RS vibration.

First, the stress seen by the product during HALT must be determined. If the stresses in HALT were constant, this would be straightforward. However, during HALT, the product is subjected to steadily increasing stress levels, with a prescribed duration at each stress level, until failures are detected. (See Figure 2). As Wöhler’s equation shows, fatigue is exponentially related to stress, so this step stress method greatly accelerates fatigue and significantly reduces the total test time to failure. Miner’s Rule allows the fatigue at these various levels to be summed together to determine the total fatigue seen by the product during HALT before failure. The following analysis uses Miner’s Rule to calculate the total fatigue accumulated at failure, based on the following assumptions:

• The number of cycles of stress used in Wöhler’s equation can be replaced by the time that the product is subjected to random vibration. This means that the number of excursions of stress per minute experienced by the product remain constant as the stress level is increased. The random nature of RS vibration makes this a reasonable assumption.
• The HALT vibration stresses are applied in the most commonly described fashion. The vibration is started at a low level, typically an input setpoint of 5 gRMS. The vibration is held at this level for a dwell time of 10 minutes, then the product is fully functionally tested, which takes approximately 5 minutes, resulting in a total dwell time of 15 minutes. Next, the vibration setpoint is increased by 5 gRMS and the dwell and test process is repeated. This continues until a failure occurs (See Figure 2).

Figure 2: Step Stress Testing. Source: ESPEC

• The gRMS product response is equal to the input setpoint. Fatigue does not occur due to the input, but rather, due to the response of the product at that input. If this 1:1 correlation is not the case, then the actual product response gRMS values would be used, rather than the setpoint values. Assuming this correlation simplifies the calculations.
• gRMS is an accurate metric for product stress. There has been some discussion on this point. Since the desired result of the calculations will be a fatigue ratio rather than an absolute fatigue value, the results will not be affected by any error in this assumption, as long as the calculation methods are the same in all cases.
• A value of β must be assumed for the product. β is the slope of the S/N curve when viewed on a log-log graph and is a characteristic of the material’s fatigue response to stress. While values of β have been empirically determined for many discrete materials, the value for a particular product or circuit card will be unique and unknown. When the S/N curve is derived empirically for electronic products, values between 4 and 8 have been measured. MIL-HDBK- 344A recommends a value of 6.4 for solder (not lead-free). For a mid-range representative value, a β of 7 will be used in this example.
• Finally, for this example, it is assumed that the product failed at the end of the 15 minute dwell of the 50 gRMS stress. This point of failure is referred to as the destruct level for the product.

To calculate the dwell time at 50 gRMS that will provide fatigue equivalent to the stepped stress fatigue induced before beginning the 50 gRMS dwell, described above, the following equation must be solved for N1 (N = time, in hours):

N1(50g)7/c = summation from g = 5 to g = 45, step = 5, .25*gi7/c N1(50g)7 = .25*( summation from g = 5 to g = 45, step = 5, gi7)

N1=.25*( summation from g = 5 to g = 45, step = 5, gi7)/ (50g)7 N1=.20 hrs = 12 minutes

Equivalent fatigue at 50g to failure = 15 + 12 = 27 minutes

These calculations show that the fatigue accumulated in stepped vibration used in this example is approximately equal to the fatigue accumulated in a continuous 30 minute dwell at 50 gRMS alone. Figure 3 shows the shape of the S/N curve that result from these calculations.

Figure 3: S/N curve and HASS proof of screen. Source: ESPEC

Now, consider how the stresses of HASS would appear on this curve. A typical HASS screen will use a vibration level that is 50% of the destruct level found in HALT. In this example, that results in a HASS vibration level of 25 gRMS. Also, a typical, somewhat aggressive HASS would consist of four hours of testing, during which the vibration would be on 50% of the time. Figure 4 shows the time and stress level for this HASS on the S/N curve.

Figure 4: S/N curve with HASS shown. Source: ESPEC

The safety portion of the PoS is intended to demonstrate that the proposed screen will not damage good products. During the safety portion, the proposed HASS vibration screen is run multiple times. This example uses a 20 times repetition of the screen. Figure 5 shows the PoS safety portion fatigue level on the S/N curve. In order for a proposed screen to be safe, known good units must be able to survive these repetitions without any induced failures, and still be able to pass all design verification testing. The graph shows that this should be the case since the accumulated stress on the product after 20 cycles is below the failure line.

Figure 5: S/N curve with HASS proof of screen shown. Source: ESPEC

β will increase dramatically (causing number of cycles to failure to drop dramatically) if there are defects or irregularities in the object being stressed. Cutting a circular hole in a plate causes the stress to be concentrated around the hole (see Figure 6). A sharp corner or nick will result in much higher levels of β. Figure 7 shows an S/N curve that includes a defective product with a modestly increased β. Clearly, it would fail within a normal HASS cycle.

Figure 6: Stress lines in a plate with a circular hole, under tension. Source: ESPEC

Figure 7: S/N curve showing HASS and defective product. Source: ESPEC

Conclusions and final considerations

The classical reliability prediction tools based on Wöhler’s equation and Miner’s Rule are difficult to use to directly understand the fatigue induced in a HASS screen. However, the empirical data derived from HASS PoS provides a solid basis for using these standard tools to estimate the relationships between the stresses of HALT, HASS, HASS POS and their fractional effect on real life.

Also, it is important to note that the fatigue the product experiences in HASS, and the early wear-out modes this fatigue induces in defective products, represents only a portion of the effectiveness of HASS. Many of the failure modes found in HASS are not the result of fatigue, but are evidence of parametric drift stack up changes due to process or product changes, or other soft failures. The ability of HASS to detect these subtle changes, which can result in intermittent, no trouble found (NTF) failures if allowed to escape into the field, makes it a uniquely powerful screening method.

Most significantly, HASS stresses serve to continuously verify that component variability, engineering changes, local manufacturing process variability, upstream supplier process variability or other unknown shifts in the product makeup have not changed the high level of product reliability that was achieved by using HALT during product design.

About ESPEC

ESPEC is the largest global brand of accelerated reliability testing systems for performing HALT and HASS that improve product quality. ESPEC’s Qualmark Reliability Test Technology and services help companies in automotive, aerospace, medical, electronics and other manufacturing industries to introduce new products quickly, boost product reliability, slash warranty costs and build lasting consumer relationships with quality products.