Effects of Stresses in Highly Accelerated Limit Test (HALT)

Transcription

1 Technical Report Effects of Stresses in Highly Accelerated Limit Test (HALT) Takuya Hirata, Raphael Pihet and Yuichi Aoki, Development Headquarters, ESPEC CORP. Abstract The effects of vibration stress and combined stresses in Highly Accelerated Limit Test (HALT) using printed circuit boards have been quantitatively investigated. During the vibration step test at 20 C, the lead of the aluminum electrolytic capacitor has been broken due to the bending stress. On the other hand, during the combined stresses test (temperature cycle and vibration step), cracks and delamination have also been found at the solder joint of the capacitor in relation with the capacitor mass and solder quantity. Moreover, it has been possible to detect those cracks and delamination during the vibration step stress at 80 C. Therefore, different failure modes have been detected depending on the test conditions in HALT. This technical report is a modified and expanded version of information presented at the 47th JUSE Reliability & Material Safety Symposium held in July Introduction In recent years, the usage environment of electronic devices has become more and more severe and complex, enhancing the importance of reliability improvement. Moreover, today s shorter product development cycles make it important to build higher quality into products at the early stage of development. Therefore, the Highly Accelerated Limit Test (HALT) is attracting interest as a test method that helps developers respond to these market changes. HALT is an accelerated test method that applies severe stresses to products with the purpose of improving quality by rapidly exposing product weaknesses. 1, 2 Using HALT during the early stage of product development can reduce the amount of rework needed. While some cases applying HALT have been published, these are very few compared to reliability tests. Moreover, there are very few cases where the effectiveness of the weakness detection in HALT has been quantitatively evaluated. Therefore, this technical report presents our quantitative evaluation of the relationship between stresses and product failures. The stresses applied were vibration with 6 degrees of freedom (a characteristic of HALT), and combined stresses consisting of vibration with rapid temperature change. 2. HALT Overview HALT is an accelerated test method developed in the US in the 1980s and adopted in 2013 in the IEC standard (Methods for product accelerated testing). While typical accelerated life testing is classified as a quantitative accelerated test method, the IEC standard classifies HALT as a qualitative accelerated test method. 3, 4 HALT is similar to the quality engineering concept, and differs from traditional accelerated life testing. Test Navi Report No. 26 (101st Issue)

2 The main stresses applied in HALT are rapid temperature change and repetitive shock vibration with 6 degrees of freedom. Table 2.1 Types of failures detected during HALT In HALT, the sequential application of various types of stresses can identify latent weaknesses in the product. A basic HALT sequence consists of the following five tests: Low temperature step test, high temperature step test, rapid change temperature cycle test, vibration step test, combined test (rapid temperature change with vibration steps). Table 2.1 shows the types of failures detected during HALT. 5 We quantitatively evaluated the effectiveness of HALT to detect these types of failures. This technical report presents examples of the failure detection using a PCB with mounted aluminum electrolytic capacitors. It covers failures Table 3.1 Capacitor types and mounting conditions in (1) leads (broken leads) and (2) PCB mountings (solder joints). 3. Test Methods 3.1 Specimen To assess the impact of HALT stresses on a PCB, we created a test board consisting of aluminum electrolytic capacitors mounted under different conditions. Table 3.1 shows the capacitor PCB conditions. We selected 3 mounting factors thought to affect failures: Capacitor size (weight), capacitor lead length, and solder quantity. We combined these factors to create an experimental design with 9 mounting conditions. We used a uniform component height of 20 mm, and different component diameters. Figure 3.1 shows the layout of the created capacitor PCB. The PCB used an FR-4 board (size: 200 mm capacitors; thickness: 1.6 mm). The PCB was divided into 4 areas (CN1 to CN4) with different capacitor mounting positions and orientations in each (see Figure 3.1). Condition No. Area name Figure 3.1 PCB Layout Capacitor (mounting orientation) Test Navi Report No. 26 (101st Issue)

3 3.2 HALT chamber The testing was done in a Typhoon 2.5 HALT chamber made by US manufacturer Qualmark. Table 3.2 shows the main specifications of the HALT chamber used. Figure 3.2 is a schematic diagram of the HALT chamber. The test chamber contained a floormounted vibration table and ceiling-mounted air temperature control chamber. HALT uses air pressure actuators installed below the vibration table to apply repetitive shocks that subject the specimen to complex vibrations having 6 degrees of freedom. The specimen is also subjected to sudden temperature changes using liquid nitrogen as a coolant. Blower Heater LN 2 nozzle Table 3.2 Main specifications of Qualmark HALT chamber used for testing Specimen Vibration table Actuator Figure 3.2 HALT chamber schematic 3.3 Test conditions The HALT testing consisted of vibration step testing and combined testing (with temperature cycles and vibration steps). The test conditions are described in detail in the sections below. To apply vibrations to the specimen, we secured it to the HALT chamber s vibration table at 9 points with fasteners. To detect failures during testing, we monitored the voltage of each capacitor. We applied voltages of 5 VAC, and recorded the voltage changes during the test. A drop in the measured AC voltage of at least 1% from the initial value was evaluated as a component failure. 4. Lead Breaks in Mounted Components 4.1 Vibration step test results We performed a vibration step test on the capacitor PCB. Keeping the temperature at a constant 20 C, we increased the vibration acceleration to 70 Grms in 5 Grms increments every 5 minutes. Figure 4.1 shows the appearance of the PCB with mounted electrolytic capacitors after the HALT vibration stress test. After subjecting it to a vibration step test of about 70 minutes, we found wire breaks in 8 components of 3 different types. These wire breaks were all in capacitor leads, and all had the same failure mode. Figure 4.2 is a photo showing a capacitor failure location. Figure 4.3 is a SEM image of a broken lead cross-section. All the breaks occurred at the base of lead on the capacitor side. Test Navi Report No. 26 (101st Issue)

4 Figure 4.1 PCB after testing 5 Figure 4.2 Capacitor failure location (lead) 5 Observations of break surfaces suggested that the breaks were mainly the result of brittle fractures. We used the test results to determine the HALT vibration stress impact relative to various mounting conditions. As shown in Figure 4.4, vibration stress impact increased as lead length increased. When we examined the vibration stress impact in relation to capacitor weight, we found that the stress impact was highest for a weight of 3.8 grams (capacitor diameter of 12.5 mm). This result suggested that factors other than size and weight played a role. One of these other factors may have been the difference in capacitor lead thicknesses the 1.7 and 3.8 gram capacitors had a lead diameter of 0.6 mm, while the 6.2 gram capacitors had a lead diameter of 0.8 mm. The effect of lead thickness is therefore examined in the next section. We did not find any effect from solder quantity in this test, which we infer was because the failures were located at the base of the capacitor leads and not in solder joints. Solder joint failures are discussed in Section 5. Figure 4.3 Break surface of capacitor lead Bending stress calculation results Figure 4.4 Impact of mounting conditions for the vibration step test To examine the lead break test results, we calculated the bending stresses applied to the capacitor leads using a simplified model, and used the results as relative indices.5 This model uses component weight, lead thickness and lead length as parameters. As shown by the formulas below, bending stress σb is calculated from bending moment M and section modulus Z: σb = M / Z (1) M = (m g / 2) L (2) Z = (π d 3 ) / 32 (3) where m is the component weight, g is the gravitational constant, L is the lead length and d is the lead thickness. Test Navi Report No. 26 (101st Issue)

5 Figure 4.5 shows the bending stress calculation results for the leads. For a constant lead length (L) of 3.0 mm, the highest bending stress results from a component weight (m) of 3.8 grams (component diameter of 12.5 mm). The reason is that the lead thickness (d) is relatively low in relation to the component weight (m). The results also show that for a constant component weight (m) of 3.8 grams, the bending stress increases in proportion to the lead length (L) and makes the lead more prone to breaking. The similarity between the calculation results and HALT test results demonstrated HALT s effectiveness in exposing vibration stress weaknesses. 5. Solder Joint Cracking 5.1 Combined test results The combined test performed on the capacitor PCB consisted of the HALT vibrations and sudden temperature changes shown in Figure 5.1. We ran a total of 5 cycles with temperatures of -40 C and 80 C set for 10 minutes each. We increased the vibration acceleration to 70 Grms in 5 incremental steps of 14 Grms each. The combined test results revealed two failure modes. One failure mode was lead breakage. As with the vibration test, the breaks were at the capacitor bases. The other failure mode was solder joint failure, which exhibited solder cracking, or peeling at the lead-solder interface (Figure 5.2). We used the solder joint failure data to determine the impact of the combined test on solder joints (Figure 5.3). The impact of the solder joint failure mode was different from the impact of lead breakage. Solder joint failure occurred mainly in the heaviest (6.2 gram) capacitors (component diameter of 16 mm), which could indicate that weight load had an effect. Failures also tended to increase as solder quantity decreased, suggesting that solder joint surface area had an effect. Bending stress (mn/mm 2 ) L = 3.0 mm Component mass m (g) m = 3.8 grams Lead length L (mm) Figure 4.5 Calculation results of the lead bending stress Temperature ( C), acceleration (Grms) Time (min) Figure 5.1 Combined test profile Temperature 温度 Vibration 振動 Figure 5.2 Solder joint failure locations after the combined test Number of failures Small Low 少 High Large 多 Component mass (g) Lead length (mm) Solder amount Figure 5.3 Impact of mounting conditions for the combined test (solder joints) Test Navi Report No. 26 (101st Issue)

6 We infer that the solder joint failure mode is a combined stress failure resulting from constant temperature and vibrations, or from temperature changes and vibrations. The next section provides the combined stress test results for constant temperature and vibrations. 5.2 Vibration step test results (constant 80 C temperature) We performed a vibration step test on the capacitor PCB in an 80 C environment. All the failures exhibited the same failure mode as in the combined test (solder joint cracking and peeling). As shown in Figure 5.4 (a), cracking occurred inside the solder joint when the solder quantity was low. As shown in Figure 5.4 (b), peeling occurred at the lead interface when the solder quantity was high, possibly indicating poor soldering. Figure 5.5 shows the impact of the vibration step test on solder joints in an 80 C environment. The results show a larger number of failures for higher component weights and for lower solder quantities. The solder joint failure mode detected by the combined test described in the previous section is also affected by mounting conditions (component weight and solder quantity), suggesting that the solder joint failure mode of the combined test is the result of vibration stress in an 80 C environment. The findings above indicate that HALT combined testing can rapidly detect a greater number of failure modes, making it an effective method of detecting product weaknesses. But a method that can test anticipated stresses in isolation like HALT step testing is preferable for explicating failure mechanisms. Figure 5.4 Cross sections of solder joint after vibration step test at 80 C Figure 5.5 Impact of the mounting condition for vibration step test at 80 C (solder joints) Test Navi Report No. 26 (101st Issue)

7 6. Conclusion In this technical report, we have quantitatively investigated the effectiveness of the weakness detection in HALT by focusing on the vibration step test and combined test (vibration step and temperature cycle between -40 C and 80 C) using PCBs with mounted aluminum electrolytic capacitors, and obtained the followings: (1) The vibration step test (at 20 C) has produced the lead break of aluminum electrolytic capacitor. The failure mode has occurred in relation with the lead bending stress. (2) The combined test has also caused cracks and delamination at the solder joint in relation with the capacitor mass and the solder quantity. On the other hand, the vibration step test at 80 C has only given solder joint failures. Therefore, the failure modes of the solder joint have occurred due to the vibration stress at high temperature. As a result, during HALT the detection of different failure modes and its effectiveness have been quantitatively evaluated. Therefore HALT is considered as an effective method to improve the quality of electronic devices. 7. References 1. Gregg K. Hobbs, Accelerated Reliability Engineering HALT and HASS, Wiley, Harry W. McLean, HALT, HASS, and HASA Explained Accelerated Reliability Techniques, American Society for Quality, IEC 62506:2013, Methods for product accelerated testing 4. Fumiaki Harada, Outline of International Standard IEC Accelerated Testing and Key Points, Proceedings of 23rd Spring Symposium on Reliability, pp , Reliability Engineering Association of Japan, Raphael Pihet, Takuya Hirata and Yuichi Aoki, Investigation on the Effects of Vibration Stress in Highly Accelerated Limit Test, Proceedings of 29th Autumnal Reliability Symposium, pp , Reliability Engineering Association of Japan, Takuya Hirata, Raphael Pihet, Hideki Kawai and Yuichi Aoki, Effects of HALT Vibrations in PCBs, 31st JIEP Spring Meeting, pp , Japan Institute of Electronics Packaging, 2017 Test Navi Report No. 26 (101st Issue)