Is Your Reliability Testing Program Keeping Pace with Manufacturing and Design Advancements? WHITE PAPER QUALMARK: Accelerating Product Reliability WWW.QUALMARK.COM 303.254.8800 The information contained in this paper is Qualmark proprietary information and is copyrighted. No portion may be copied, modified, distributed or published without the express written permission of Qualmark Corporation.
Is Your Reliability Testing Program Keeping Pace with Manufacturing and Design Advancements? by Neill Doertenbach, Senior Applications Engineer, Qualmark The ever increasing pressure to more quickly deliver higher reliability products at lower costs is prevalent throughout major industries, such as avionics, communications, networking equipment, computer systems, medical, and defense and has led to new design and manufacturing techniques. Now, more than ever, it is imperative to revisit reliability testing techniques and measure their effectiveness against these manufacturing and design advances. A comparison of accelerated stress tests conducted at Qualmark labs 10 years ago, as measured against more recent results, yielded significant differences in the ratios and types of failures found. The study uncovers shifts in operating and destruct limits, reviews causes of the differences between the failure data from the two time periods, and reveals the stresses needed to find failure modes that, otherwise, would go undetected in today s modern electronics. INTRODUCTION The test techniques of HALT and HASS started to gain commercial acceptance in the early 1990 s. They have been used successfully to quickly expose design and process defects in a wide range of products, from purely mechanical systems to electromechanical and electronic products. Over the years that these methods have been used, the technologies employed in product design have changed dramatically. The use of through hole printed circuit boards (PCBs), the norm in the early days of HALT and HASS testing, is now relegated only to power circuits that demand the high current capabilities of that assembly method. Surface mount PCBs, using smaller and smaller geometries of components, dominate electronic product designs. This paper will document the continued effectiveness of HALT stresses for precipitating failures in products using these newer technologies. Since the introduction of HALT and HASS testing, technologies employed in product design have changed dramatically. In 1996, Qualmark published a study on failure results found at Qualmark s Santa Clara, California Accelerated Reliability Test Center (ARTC ). This paper analyzed HALT data from 33 different companies representing a variety of industries. [1] The examples in that study were obtained between May 22, 1995 and March 31, 1996 and were from companies that had completed HALTs at the Santa Clara lab. The study was comprised of data on 47 products from 33 companies across 19 different industries. The majority of products were electrical, but several of the products had mechanical components as well. This Qualmark paper provided a comprehensive snapshot of the effectiveness of HALT on the technologies in use in the mid 1990 s. The key conclusions of that paper included the finding that random six degree of freedom vibration was more effective than temperature cycling for precipitating product failures, and that combined thermal and vibration stresses identified failure modes that could not be precipitated using either stress alone. The purpose of this paper is to compare and contrast new data from tests conducted at Qualmark s Accelerated Reliability Test Centers in Denver, Colorado and Boston, Massachusetts labs in 2007 and 2008, with the data from the 1996 study. The new data is categorized in a similar fashion to allow a straight forward comparison. 2
The data set used for this paper came from a total of 47 tests conduced by 30 companies. This is very similar to the data set used in the 1996 study of 47 tests across 33 companies. The breakdown of the companies by industry is shown in Table 1, and is compared to the results from the 1996 paper. Halt is used by a wide range of industries. The study identified 19 different types in 1996; 15 in 2008. There are some wide differences between these two data sets. Power electronics, which included power supplies, power conversion and battery backup systems, accounted for over a fifth of the testing done in the 2008 study, while no power electronics systems at all were included in the 1996 study. There were similar striking differences in the number of avionics companies (7 in 2008, 1 in 1996) and microwave equipment (0 in 2008, 4 in 1996). There is no clear reason for these differences, although a likely cause could be the type of industries that were prevalent in the geographical area of each of the labs during the timeframe being evaluated. One thing is clear -- the breadth of industry types using the labs still spans a wide range. In 1996, 19 different industry types were identified, while in 2008, 15 were identified. The industries shared by both studies included avionics, communications, networking equipment, computer systems and defense electronics. Another interesting metric to examine is the ratio of the number of tests conducted to the number of companies testing. In 1996, this was 47 tests/33 companies, or 1.4 tests/company, and in 2008 the number was 47 tests/30 companies = 1.6 tests per company. This shows that, just as in the mid 1990 s, some companies are coming back to the lab for repeat tests with new or improved products over the course of the two years covered in the study. This indicates that companies that do not own a HALT chamber find the test method valuable enough to keep returning to the lab with new products to test. The majority of the products in the new data set, 64 percent, utilized surface mount technology (a similar data point from the 1996 paper is not available, but is assumed to be lower.) Table 1 - Distribution of Companies by Industry Type Industry Type Number of companies, 2008 Power 10 0 Avionics 7 1 Communications 4 3 Industrial controls/systems 4 0 Locks and controls 3 0 Networking Equipment 3 6 Computer Systems 2 4 Gas Ignition Controller 2 0 Medical 2 0 Clock 1 0 Fan 1 0 Residential Electrical 1 0 Satellite TV System 1 0 Military Avionics 1 0 Defense Electronics 1 4 Microwave Equipment 0 4 Video Processing Equipment 0 1 Hand held measuring equipment 0 1 Monitors 0 1 Printers and Plotters 0 1 Speakers 0 1 Semiconductor Manufacturing Equipment 0 1 Fiber Optics 0 2 Remote Measuring Equipment 0 2 Number of companies, 1996 Qualmark proprietary information. Contents may not be reproduced without permission 3
SUMMARY OF HALT DATA Figure 1 Figure 2 2008 Results By Stress Type 1996 Results By Stress Type COMBINEDD 32% VIBRATION 19% COLD STEP 19% HOT STEP 21% RAPID A D THERMAL R 9% COMBINED 20% VIBRATION 45% COLD STEP 14% HOT STEP 17% RAPID THERMAL 4% Qualmark proprietary data. May not be reproduced without permission The HALTs summarized in this paper were all conducted using generally accepted methodology so the details of the methodology will not be reviewed here. It is important to note that the order of the applied stresses, first Cold Step, then Hot Step, then Rapid Thermal, then Vibration and finally Combined Vibration and Rapid Thermal, was the same in both sets of data. The failures precipitated during this testing were broken down based on the stress that induced the failure. Graphs of this data and similar data from the 1996 paper are shown in figures 1 and 2. The data is tabulated in table 2. Table 2 Distribution of Number of Failures By Stress Type Stress 2008 Results 1996 Results Percent` # of Failures Percent # of Failures Cold Step 19% 55 14% 34 Hot Step 21% 62 17% 39 Rapid Thermal 9% 26 4% 10 Vibration 19% 55 45% 107 Combined 32% 96 20% 46 Total 294 236 The differences between these two sets of data are very interesting and fit with intuitive predictions based on the increased use of surface mount technology. In general, surface mount technology is much more robust from a vibration standpoint. With no leads to shear off in vibration or components with higher mass standing above the board and gaining inertia, it would be expected that these products would tolerate vibration much better. This prediction is born out in the data, with a 58% reduction in the fraction of vibration failures versus the 1996 data. This reduction is offset by increases in the fraction of failures in the other stresses. The combined stresses of HALT found a full 32% of failures that would have otherwise been missed. It is interesting to note that the fraction of failures due to rapid thermal stresses more than doubled between the two studies. Again, this fits the intuitive prediction that surface mount assemblies would be sensitive to the rapid expansions and contractions of boards and components resulting from this stress. Another indicator of the robustness of surface mount assemblies versus assemblies using through hole technology is seen in the increase in failures precipitated in the combined environment. As in the 1996 study, the failures discovered in the combined environment were only precipitated in that environment. They were 4
not found when using the individual stresses. The 1996 data showed that 20% of the failures discovered in HALT required the extreme stresses of the combined environment to bring them out. However, in 2008 that percentage has increased to 32%. This means that, when doing stress testing on today s robust products, a third of the possible failure modes will likely be missed if the stress testing does not include a combined thermal and vibration environment. Another point to note is that the total number of failures has increased from the previous study. Remembering that the same number of tests was examined in each study, this means that the average number of failures per HALT has increased. In 1996, the average was 5.0 failures per test, while in 2008 the average was 6.25. It is difficult to guess at a reason for the change, although it does clearly show that HALT is still very effective in precipitating failures, and that the average HALT will result in several failure modes being uncovered. LIMITS COMPARISONS A key output from a HALT is the identification of the Thermal Upper and Lower Operating and Destruct Limits (LOL, LDL, UOL, UDL) and the Vibration Operating and Destruct Limits (VOL, VDL) for a product [2]. Table 3 shows the average, high, low, and mean of these limits for the tests reviewed in 2008 and compares them to the 1996 data. Table 3 Limits Summary And Comparison Thermal Vibration LOL LDL UOL UDL VOL VDL Year 2008 1996 2008 1996 2008 1996 2008 1996 2008 1996 2008 1996 Average -48-55 -66-73 92 93 108 107 41 61 45 65 Most Robust Least Robust -100-100 -100-100 200 200 200 200 65 215 65 215 10 15-20 -20 45 40 65 40 10 5 15 20 Median -50-55 -60-80 90 90 100 110 40 50 47.5 52 This comparison shows another interesting change between the 1996 study and this 2008 study. While the average thermal limits are roughly the same between the two sets of data, the vibration limits are markedly different, with products in the 2008 study failing at average vibration levels that are 33% less than in the 1996 study. This seems to indicate that, in conflict with the earlier conclusions, the products in the 2008 study were in general less robust from a vibration standpoint. However, there is one important piece of information that is not being considered. Between 1996 and 2008, Qualmark made several changes to its vibration system specifically to increase the stress it delivered to a product under test, particularly in the lower (<2000 Hz) frequencies. The change seen in average vibration limits reflects the ability of the new vibration system to drive out failure modes at lower vibration levels. When this change is taken into consideration, the change in the percentage of failures precipitated in vibration testing that was discussed earlier becomes even more significant. The 1996 paper also showed the operating and destruct limits for products broken down by product application and product use environment. A comparison of that data versus a similar breakdown on the 2008 data is shown in tables 4 and 5. As in the 1996 study, the new data clearly shows that the limits found in HALT greatly exceed the stresses that will be found in the end use environment. There is no significant variation in the data from 1996 to 2008. Table 4 Limits by Product Application Thermal C Vibration LOL LDL UOL UDL VOL VDL Year 2008 1996 2008 1996 2008 1996 2008 1996 2008 1996 2008 1996 Military* -60-69 -75-78 50 116 80 123 25 121 30 124 Field -41-57 -100-74 100 94 100 115 50 64 50 66 Commercial -55-48 -66-73 78 90 94 90 36 32 48 39 Avionics** -52-73 102 128 59 60 * 2008 military data was from a single sample, an electro-optical system. This accounts for the low vibration tolerance. **Avionics data was not broken out in1996 paper. 5
Table 5 Limits by Product Environment Thermal C Vibration LOL LDL UOL UDL VOL VDL Year 2008 1996 2008 1996 2008 1996 2008 1996 2008 1996 2008 1996 Office -59-62 -73-80 70 92 90 118 43 46 57 52 Office w/ User -40-21 -40-50 110 67 110 76 15 32 20 36 Vehicle -100-69 -100-78 100 116 100 123 50 121 50 124 Field -41-66 -54-81 93 106 107 124 38 66 43 69 Field w/ user -38-49 -59-68 74 81 94 106 45 62 51 62 Avionics -52-60 -73-90 102 110 128 110 59 18 60 29 TYPES OF FAILURES The data that was used for this paper as well as the 1996 study was drawn from reports created by the Qualmark lab engineer. Consequently, the failure information contained in the reports indicated what was known about the failure at the time of the test and did not reflect the results of any troubleshooting done by the customer after the test. This meant that only the most obvious failures, typically mechanical failures, were detailed in the report. Most of the failures found were described as Troubleshooting in progress. In the 1996 paper, 170 of the failures, or 72%, fell into this category. In the 2008 study, that percentage has risen to 79%. This large number of undefined failures renders any comparison of failure modes between the two studies invalid, so it will not be pursued here. There is one data point of interest, however, which emphasizes the shift from through hole to surface mount. In the 1996 study, 53 failures, or 22%, were broken leads. In the 2008 study, the number was only 20 failures, or 7% of the total. SUMMARY The test techniques of HALT and HASS emerged in the early 1990 s, and were proven effective on products using technologies and assembly methods that are quite different from those in use today. It would be reasonable to question if the methodologies are still valid. This study shows that the stresses used in HALT are still quite effective for precipitating failures in electrical and electromechanical products. The changes in technologies have resulted in shifts in the effectiveness of the individual stresses. However, it is just as difficult for a failure mode to go through a HALT undetected in 2008 as it was in 1996.... REFERENCES: [1] Silverman, Mike. Summary Of Halt And Hass Results At An Accelerated Reliability Test Center. Proceedings, IEEE Reliability and Maintainability Symposium 1998 [2] McLean, Harry W.. HALT, HASS & HASA Explained: Accelerated Reliability Techniques. ASQ Quality Press, 2000. [3] Doertenbach, Neill K.. Highly Accelerated Life Testing: Testing With A Different Purpose. Proceedings of the Technical Program, NEPCON West 2000, Vol. 2 2000: 765-773. ABOUT THE AUTHOR: Neill Doertenbach is a Senior Applications Engineer with Qualmark and serves as the lead for our Professional Services Team. He has a wide range of industry experience, including digital hardware design, software, and Quality Assurance. He has written and presented papers globally on HALT and HASS basics, grms measurement and calculation methods and understanding repetitive shock vibration. Neill holds an Electrical Engineering degree from CSU in Fort Collins, Colorado, USA.... The information contained in this paper is Qualmark proprietary information and is copyrighted. No portion may be copied, modified, distributed or published without the express written permission of Qualmark Corporation. 6
Accelerating Product Reliability Qualmark is the largest global supplier of accelerated reliability testing systems for performing HALT (Highly Accelerated Life Tests) and HASS (Highly Accelerated Stress Screens) that improve product quality. Qualmark 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 based on quality products. www.qualmark.com sales@qualmark.com 303.254.8800 The information contained in this paper is Qualmark proprietary information and is copyrighted. No portion may be copied, modified, distributed or published without the express written permission of Qualmark Corporation.