Early Reliability Testing
Almost every industry is competing on Reliability and therefore, there is a need to develop more reliable products faster. However, reliability testing and improvement often are performed when development is nearly complete, when the product is nearly frozen, when time is short, and when improvements are more difficult and costly. Instead, Early Reliability Testing is a development tactic that can provide higher reliability and quality, with less cost and time for development, and less development risk.
The big drawback to early testing are: low test coverage, low number of samples, and samples with immature manufacturing processes. However, all of these issues can beaddressed and overcome, and when this happens, the benefits are early life testing are substantial.
When to perform a Reliability Test?
Figure 1 below shows a typical product development life cycle. Given this, when should we perform reliability testing?
Figure 1: Typical Product Development Life Cycle
Many would say that you must do in P1 in order to affect the design. Others would argue that you cannot perform reliability testing until P2 when the design is more mature and functional tests are more complete. Which is correct?
Both and Neither, really! Best solution is really to do both to get the benefits of early testing (P1) but also to test later in the design (P2) when test coverage is higher and samples are cleaner and then use Verification Testing to validate the design (P3).
And if we are working with a modular design, we should be testing the subassemblies as they become ready (P1) rather than waiting for system test at the end – may be too late. See Figure 2 below for a depiction of when testing should take place.
Figure 2: Reliability Testing within a Product Development Life Cycle
Strong Correlation Between Successive Units
When developing a product, we typically have a succession of units: a prior product, a preliminary prototype, an intermediate prototype, a final product. The test unit can be a complete product, complete prototype, a sub-assembly, or a significant component.
Typically, successive units are strongly correlated in their wear, defects, failures, and their corresponding mechanisms and root causes. If we start the testing earlier, this will enable smarter testing later with fewer compromises driven by a tight schedule. This also enables technology development on early units, and technology reuse on later units. Thus, reliability work for several successive units involves little more investment than reliability work for solely the final unit. And because of this strong correlation, even if we only test a small sample for our early testing, defects found here are usually indicative of what we would find with a larger population. Even if we have a sample of one, we can still get useful information.
These correlations enable smarter tactics: In parallel with early development, some upstream units should be used early to develop technology. This may include test apparatus, methods, and analysis, and possibly acceleration techniques and monitoring methods. Early reliability testing often will provide early understanding of defect, wear and failure causes and mechanisms that otherwise would degrade the final product.
Also, earlier testing provides much more time for reliability work. This permits longer test runs, milder acceleration, easier extrapolation, and minimizes schedule-driven compromises.
Most reliability effort is invested "up front", such as designing, constructing and programming test apparatus, fixturing, and test software. Also, for the first unit, reliability testing and learning may incur special effort. This may include development of monitoring instruments and acceleration techniques.
Thereafter, these "up front" investments provide physical and intellectual technology for successive work. Successive test runs are typically quite automated. They can use acceleration methods and monitoring instruments that are already understood and available. Therefore reliability testing on a succession of units may incur only moderately more investment than reliability testing on just the final unit. This moderate extra investment is rewarded by a better development cycle and better final product.
Overcoming Immature Manufacturing Practices in Early Development
In early development, the manufacturing process used to make the prototype is immature. However, this should not dissuade us from using these early prototypes for reliability testing. Any good reliability test program requires good failure analysis capability. It is this failure analysis that will be able to determine the root cause of each failure. Sometimes, failures will be due to immature processes which are less interesting because it is assumed that manufacturing process refinement will not come until later in the development cycle. However, these types of failures are usually very easy to detect - sometimes with the naked eye or an inexpensive microscope. Therefore, rather than paralyzing a reliability test program due to fear of these types of failures, start testing early and just weed out the relevant from non-relevant failures. And occasionally, even the process defects are "little gold mines" of information because they may be indicative of what would have occurred even with a mature process.
Overcoming Low Test Coverage
Early in product development, the in-house test coverage is usually quite low. Sometimes this means that we just need to look for more gross issues and cannot detect some of the finer issues, and even if this is the case, an early test that finds this type of gross issue may still wind up saving thousands of dollars if the issue was one that would have required board spins or mechanical chassis changes later in the design.
And sometimes, the work around is to use commercially available test equipment rather than the custom test scripts that won't be available until later. With this approach, we may not be able to achieve 100% test coverage, but it is often times good enough to make the test worthwhile.
Expertise when you need it:
Ops A La Carte LLC can provide consultants with the expertise and manpower to perform Early Reliability Testing. This can be done in parallel with your product development program, withOUT depleting the time of your in-house development team. As described above, this enables a development cycle that is shorter, with less total development cost, and lower risk for reliability. Nevertheless this can achieve a better final product, with superior reliability.
PROGNOSTICS IN RELIABILITY
Name three uses for prognostics in reliability testing and explain the methods on how to use to detect the on-set of failure.
Send Responses to:
You can email us at email@example.com. The first individual that emails us a correct solution shall receive Free Admission to our upcoming Certified Reliability Engineer (CRE) Preparation Course OR to the Applied Reliability Symposium (see http://www.arsymposium.org for more details on the symposium) on June 20-22 in San Diego. This is a $1295 value for the CRE and an $800 value for the ARS.
Solution to Last Quarter's Problem of the Month on Software Reliability:
If I am performing a reliability prediction and have the following 3 assemblies with their respective failure rates and standard deviations (assume quantity of 1 for each),
a) what is the Total Failure Rate and Standard Deviation for this potion of the circuit?
b) how is it that the fan has a lower failure rate than either the microprocessor and the memory when we know fans to have a higher failure rate in the field?
This problem illustrates one of the new uses for the Telcordia Prediction guide SR-332 Issue 2. It contains standard deviations for all failure rate numbers in an effort to give bounds around the final result. In the past, many people saw this prediction standard as either inaccurate or outdated. Now this new version helps to accomplish both of these problems.
Congratulations to Archana Pawse of Superconductor Technologies for being the first one wih the correct answer. She won a free admission to our Design for Reliability Seminar held last month. The correct solution was:
a) The formula for calculating Lambda is Total lamda = lamda1+lamda2+lambda 3 = 96
The formula for calculating Standard Deviation is the square root of the sum of the squares, or
Total sigma = sqrt (sigma1**2 + sigma2**2+ sigma3**2) = 53.71
b) The fan has a wear dominated failure mode that is really time dependent and increases rapidly over time. The suggestion here that the number is constant is a common error. Further more, for semiconductors the failure rate is often identified at about a year, which is near the bottom of their bathtub curve. For wear dominated components like fans, most often, the bottom of the bathtub occurs about 500 hours and increases from there. Hence while having a low initial failure rate they also have a high total failure. Many manufacturers treat fans as a maintenance item. Therefore, the fan never goes through a full life cycle. The low apparent failure rate is a reflection of the maintenance activities, even if the fan has a high field failure rate which is observed only when people allow them to run to failure.
SigmaQuest enables you to automatically upload & analyze your product, supplier, manufacturing test, RMA/Warranty & repair data using any web-browser.
Collapse time to quality by interrogating product yields & marginal measurements, at multiple sites or product lines, all in real-time.
SigmaQuest works with all test programs including ICT, X-ray, AOI, Flying Probe, NI LabVIEW, TestStand, LabWindows/CVI, Agilent/HP VEE & others.
SigmaQuest, Inc. | www.sigmaquest.com | firstname.lastname@example.org | +1.408.524.3180
Link Silicon of Valley, LLC (LinkSV) is an online networking resource for researching, identifying and contacting the companies and people within the 6,000 plus active & inactive companies in the greater Bay Area. Our company records identify the senior team, board members, financing, key partners and customers. There are many features which allow you to view the information from different angles and "connect all the dots".
LinkSV helps you tap into previously scarce and extraordinarily hard-to-find information on early stage companies and the key people associated with them. This will improve your effectiveness without the hassle and time of trying to do it all on your own. LinkSV is ideally suited to help you quickly identify and leverage your own connections in career search, in identifying new business opportunities, new investors and Board members.
DfR Solutions has world-renowned expertise in applying the science of Reliability Physics to electrical and electronics technologies, and is a leading provider of quality, reliability, and durability (QRD) research and consulting for the electronics industry. The company's integrated use of Physics of Failure (PoF) and Best Practices provides crucial insights and solutions early in product design and development and throughout the product life cycle. DfR Solutions specializes in providing knowledge- and science-based solutions to maximize and accelerate the product integrity assurance activities of their clients in every marketplace for electronic technologies (consumer, industrial, automotive, medical, military, telecom, oil drilling, and throughout the electronic component and material supply chain). for more information visit www.dfrsolutions.com.
The GoldSim simulation framework allows one to probabilistically simulate the reliability and performance of complex engineered systems over time. GoldSim provides the ability to model the interdependence of components through requirements and fault trees, as well as the capability to define multiple independent failure modes for each component. This facilitates both reliability modeling and risk analysis within a variety of industries, including space and defense, manufacturing, mining, telecommunications, electronics and infrastructure. For more information, contact Tim Schmitt at email@example.com or go to www.goldsim.com
Ops A La Carte's newsletter goes out to over 8500 subscribers. If you would like to advertise in next quarter's "Reliability News", email us at firstname.lastname@example.org or call at (408) 472-3889.
Abbreviated listings are shown. For a full listing, click on this Job Openings tab.
Sr. Reliability Engineer
Sr. Reliability Engineer
Senior Reliability Engineer
Quality Manager -TL9000 Program
Sr. Quality Engineer
Senior Reliability Consultant
Ops A La Carte's newsletter goes out to over 8500 subscribers. If you would like to put a job opening in next quarter's "Reliability News", email us at email@example.com or call at (408) 472-3889.