The critical importance of testing and validation of electrical connectors and other components, regardless of the industry or market, is obvious. Automotive reliability has been improving steadily as the industry addresses the various weaknesses and failure modes specific to the internal combustion engine. However, with new architecture and innovations in the industry such as electrification and autonomous driving, new components and a shift in failure modes have changed the landscape. Simply following the same old test methodologies may no longer suffice. Improved methods of validation are necessary.
The Importance of Reliability and Validation
In the global automotive industry, common standards for reliability and testing help establish a shared language and understanding of component and design reliability. For electronic solutions, engineers use standards included in ISO, SAE, USCAR, and IEC guidelines to validate a specific product’s performance across a range of use environments. Automotive engineers can use adherence to these standards as shorthand for determining a product’s electrical, mechanical and thermal performance.
For the most part, a product’s design must serve as intended through the duration of the test period with zero failures. These validation tests use accelerated models to compress the expected lifespan into a shorter period so that this process can take place within a reasonable time frame. The process is known as “success run” or demonstration testing. This testing is essentially go/no-go in nature: either the product passes the test with zero failures, or it doesn’t. Unfortunately, it does not tell us about the failure mode or provide a quantifiable reliability value.
Product Life Cycles and the “Bathtub Curve”
Establishing the true reliability of a component is far more complex than a go/no-go determination. Reliability needs to be managed throughout the product life cycle. Typical product life cycles exhibit initial failure rates that can be relatively high. These are called early life or infant failures and are usually attributed to poor assembly, defective components or transportation damage. One can apply quality control measures or conduct end-of-line testing to detect these early failures to avoid such defects in the field.
This initial period of elevated failure rates gives way to a period of lower and constant failure rates during the product’s intended life span. During this period, random failures – often caused by external factors such as accidents, overstress, or even natural events like lightning – occur at a reasonably constant rate. Generally speaking, the best way to reduce or eliminate random failures in the product is to have redundancy in the system.
Finally, as the product approaches the end of its life span, the predominant mode is “wear-out,” in which the failure rate increases. Wear-out failures include but are not limited to fatigue, corrosion and wear. Typically, replacement of components addresses this wear out or the inevitable end-of-life failures. The elongated “U” shape of this failure rate is referred to as the “bathtub curve.”
Defining Reliability
Reliability is the probability that a product will perform its intended functions without failure in specified environments for a specified period of time. Reliability is a function of time and the failure rate, which was discussed previously in terms of the product life cycle and the bathtub curve. Additional parameters in calculating reliability include Failures In Time (FIT), or the number of failures per billion hours; Parts Per Million (PPM); and BX-life, or the product life at which some percentage (X) of the product will fail. Reliability will always be in the range of 0 to 1.
An exponential distribution function is the most common way of calculating reliability. However, that only works for the constant failure rates in the random failures region of the bathtub curve. A two-parameter Weibull distribution uses the Weibull slope to modify this function for the initial and wear-out stages of the life cycle. This can provide useful and insightful data to drive decisions on design and material changes.
In addition to reliability, another important – and quantifiable – metric to understand is the confidence of the reliability value. Confidence is the minimum certainty that the claimed failure rate is accurate. This is important in gauging the validity of testing. Confidence is a factor of the reliability target, number of test samples, and number of failures. Using the parametric binomial form of the success formula permits trading off test sample sizes and test durations to create a higher confidence level.
Leveraging Reliability in Current Validation Methods
The challenge with “success run” testing is that it does not provide information about the physics of failure, nor does it provide a quantitative reliability measure. It is possible to leverage the chi-squared function to get an approximation, however, this will not be sufficient for safety systems with low FIT or high Automotive Safety Integrity Level (ASIL) rating requirements. This lack of hard data makes it difficult to make reliability-based decisions during the development process.
To truly define reliability, stress life curves – sometimes referred to as Wohler curves – and the duty cycle need to be understood. Once the stress life curve is established, it’s possible to understand the field stress versus the design or system strength. Leveraging a cumulative damage approach (Minor’s rule) provides a picture of the true reliability and safety margins. Furthermore, the stress life curve provides information for developing an accurate Accelerated Life Test (ALT) to speed development.
Data-Driven Decisions
Why is it so important to have accurate and quantitative data on product reliability during the design stage? Engineers can make data-driven decisions about their product to optimize its reliability, efficiency, cost-effectiveness and manufacturability. The benefits of a full picture of product reliability are significant:
Quantitative performance data providing greater confidence in the product
Understanding of design strength and safety margins
Guidance in choosing the best materials and manufacturing methods
Reducing time to market by understanding the physics of failure and establishing accelerated tests
Modern vehicles are host to an increasingly complex network of electronic wiring and devices. Trends like miniaturization, electrification, shared mobility and “million-mile” lifespans demand a better understanding of product reliability. This improvement in reliability goes hand-in-hand with predictive engineering to both speed up the product development cycle and improve the end product.
The time to ensure these comprehensive processes, tools and practices are used in testing and validation is now. Engineers in the Molex Reliability Lab have been implementing and fine-tuning these methods to construct a clearer picture of the reliability of new miniaturized electrical solutions like the MX-DaSH Connector System. This informs design decisions made by Molex engineers to improve the durability, manufacturability and versatility of new products. Read more about the latest miniaturization innovations for the transportation industry on Molex’s miniaturization page.
Share