The contribution of environmental test to the success of power electronic devices has a long history. Beginning with early efforts by the military in the 1950s and 1960s, the benefits of environmental testing were recognized as a way to reduce “infantile mortality” and prevent field failures. The subsequent adoption of environmental testing techniques by the commercial electronics industry has been gaining momentum ever since. Increased product robustness and reliability, improved production processes and reduced warranty expenses are some of the benefits that have been realized.
While simple burn-in processes became well known, many power electronics companies adopted thermal cycling and environmental stress screening (ESS) to fully realize the benefits of environmental testing. As these testing techniques have matured, there has been pressure to reduce the overall cost, both in dollars and schedule time. Shorter product development cycles have dictated faster testing. One recent development to address these concerns is accelerated stress testing (AST), commonly referred to as highly accelerated life testing/highly accelerated stress screening (HALT/HASS).
Most companies possess a variety of departments with some involvement in environmental testing. R&D and engineering groups often perform design validation (DV) testing to evaluate product design reliability. Reliability assurance personnel often create test criteria for building reliability into product designs. Design and reliability personnel also are involved in defining and performing test suites to ensure final product meets defined requirements, such as military, governmental or industry standards.
Production personnel often are charged with performing environmental testing for process validation (PV), which verifies manufacturing processes and procedures are in control. An example of this is a production evaluation test that checks the validity of a solder process with a thermal cycle test. Stress screening, in all of its variations, is a production process, where environmental stresses are used to precipitate latent defects into failures prior to shipment.
Stresses and Their Effects
The three most typical environmental stresses are temperature, humidity and vibration. Other environmental stresses, such as altitude and corrosion, are available, but in most cases, have not seen much application in power electronics testing.
Temperature, the most common environmental stress, can cause a variety of effects. Changes in electrical constants due to temperature swings can cause unforeseen problems. Mechanical issues akin to expansion and contraction can become evident when coefficient of thermal expansion (CTE) mismatch troubles arises. Interactions of dissimilar materials, changes in flexibility and PCB delamination can all occur when temperature stresses are used to uncover problems.
Humidity as an environmental stress can be useful under certain circumstances. Physical changes such as swelling and embrittlement can occur for some materials. Oxidation (corrosion) and leakage paths (between conductors) also are attributed to extreme humidity conditions.
The effects of vibration stresses are exhibited in a loss of mechanical strength due to fatigue, cracking, displacement or the impairment of mechanical functions.
A variety of stresses are available, and the physical and electrical characteristics of the items being tested can vary greatly. It is important to realize that for any specific device to be tested, there will be a relationship between failure populations and the stresses utilized (Fig. 1). The environmental stresses appropriate for testing any specific product depend on the physical design of the product, the types of components and what is to be accomplished.
Although not actually an environmental stress, controlled application of current and voltage to the power devices under test can be an additional stress. A simple example of this is to decrease and/or increase the test voltage supplied to the device under test from a nominal value—a process commonly referred to as margining.
Environmental Testing Types and Processes
There are two fundamental forms of e nvironmental testing. The form of environmental testing where actual device usage conditions are duplicated is referred to as simulation. Simulation is used to create an environment that mimics the conditions a power supply would see in normal use. In the other form of environmental testing, called stimulation, stresses are used to discover weaknesses and limits of a product. This often involves stressing a product to the point of failure.
An example of stimulation would be to gradually increase the temperature seen by a power supply in an attempt to find the highest temperature in which it will operate and where damage will occur. Simulation is often referred to as “test to pass.” The differences, advantages and applications of these two forms of environmental test are frequently the cause of confusion and debate.
The oldest and perhaps most common type of environmental testing is burn-in. In this traditional form of testing, an elevated temperature is used to aid in the precipitation of premature failures. This type of environmental test is widely used today for power electronics. Typically, large batches of products are tested together in special burn-in “ovens” or rooms. Elevated temperature conditions are achieved using the heat from products being tested or from a heating element. Although it depends upon the specifics of the devices being tested, a common burn-in temperature for power products is +125°C.
Devices typically are powered throughout the burn-in process. Power to the device under test may be cycled occasionally, which actually causes some degree of thermal cycling as the devices heat and cool. The ease in which burn-in is implemented must be weighed against the relatively low screening effectiveness and long test times.
Temperature cycling is another type of environmental testing. It consists of changing the temperature between predetermined extremes for multiple cycles. While temperature cycling is considered more effective than burn-in, there are a few complications to overcome. Airflow is important, especially for higher power devices. Good airflow maximizes heat transfer and ensures the actual product temperature follows the chamber air temperature. It also prevents the formation of hot/cold spots within the workspace of the chamber.
Ramp rate, or the speed in which the chamber can change the air temperature, is an important consideration. Faster ramp rates are more effective stresses for a given number of cycles, but the ramps must not be rapid enough to cause undesired damage. A typical temperature profile transitions between -40°C and +125°C at 5°C to 15°C per minute (Fig. 2). At the temperature extremes, dwell periods (where the temperature is stable) are used to allow the product mass to “catch up” to the air temperature.
A ramp from ambient to a higher temperature normally starts the cycling to help “dry” the air before the low temperature is encountered. The tests typically end after the temperature returns to ambient, to prevent condensation on the devices when the chamber interior is opened to the ambient and possibly humid environment. As demands increase to reduce testing time and as understanding of stimulation grows, companies are transitioning from burn-in to thermal cycling for testing power electronics.
Thermal shock is a form of thermal cycling where small numbers of products are exposed to severe temperature changes. This normally is accomplished by moving the products between hot and cold zones of preconditioned air or a fluid. Thermal shock can be used for design validation and pre-production process validation testing.
ESS is a process used in a production environment to identify weaknesses in materials and processes. In an ESS process, every product produced is subjected to stimulus. It is the objective of stress screening to eliminate those products that would otherwise fail early in use. A complete exploration of ESS in handbook form is available at www.thermotron.com/Resources/esshandbook.html.
HALT and HASS are relatively new processes in environmental testing where aggressive temperature change rates and multiple axis vibration are used to reduce the time required for testing. In HALT, a product is tested to the point of failure to uncover product weaknesses, operating limits and destruct limits. The knowledge gained in HALT can be used to pinpoint bad parts, direct product design enhancements and to set limits for subsequent production testing.
While HALT testing normally deals with a limited number of products prior to production, HASS is a production process. Using the limits of operation and damage limits discovered during HALT, all products are stressed in HASS using fast changing temperatures and vibration. In theory, the aggressive stresses used will reduce the time required for stress screening. An unfortunate effect of using these aggressive stresses is that the fixturing and interconnect required become complex and costly.
More details on accelerated stress testing can be obtained at www.thermotron.com/Resources/asthandbook.html.
Several other types of environmental testing occasionally are performed, depending on circumstances. These types include combined high humidity and temperature (85/85), highly accelerated stress testing (HAST) and liquid-to-liquid thermal shock.
Environmental chambers are available for all of these types of environmental testing. A variety of features are available on these chambers to fine tune the application or performance. Before selecting an environmental chamber for a testing requirement, consider the criteria identified in “Environmental Chamber Selection Checklist.”
General Test and Measurement Issues
Similar to the classifications discussed for environmental testing, test and measurement can be segregated into different categories. The category of data acquisition refers to collection of data during test, normally for subsequent analysis. Although data acquisition can be done in conjunction with environmental testing, the application is relatively limited.
Parametric testing involves the detailed analysis of measured parameters during the test. As this can be used to observe parameters that may shift with changes in environmental conditions, it has special value during DV testing.
Functional testing is widely used with environmental testing throughout the various stages of product development and production. To gain the most value from the functional test, it is important the test be performed on a continuous basis. Changes in product temperature can cause intermittent or “soft” failures to occur. A soft failure appears at one condition and then disappears at another. If the product is not continuously monitored, it may be concluded that all is well. While permanent or “hard” failures can be detected by a low performance test system, failures can be of very short duration and can be easily missed. Eliminating, reducing or at least accelerating any multiplexing helps prevent this problem.
Some basic guidelines of product testing are outlined in “General Rules.” However, engineers implementing environmental testing must consider other factors such as the use of multi-head systems, test fixturing, powering products under test, loading and test stimulus. Other factors include monitoring, software and equipment safety.
• Multi-head testing. Systems built for use with environmental testing are typically “multi-head” to permit the testing of several products in an efficient manner. This efficiency often comes at the cost of complexity.
In multi-head systems, it is challenging to satisfy requirements for stimulus, power distribution, electrical loading, continuous monitoring, communication, interconnect and fixturing. Add to this the test system reliability requirements for long duration tests, and it is obvious why many companies have involved outside test system integrators to implement their systems.
• Fixturing. When placing products in an environmental chamber for test and providing fixturing to hold them, there are a variety of issues to consider. A common technique is to build product carriers for the actual physical product interface. The product carriers can be inserted and removed from the environmental chamber and permit product exchange at a bench while a separate set of carriers is running in the chamber.
When designing the product carriers, care must be taken to accommodate thermal expansion and contraction, moisture effects (if humidity testing) and airflow issues. Noncorrosive materials such as stainless steel and G-10 are typically required. The product carriers should be light to minimize thermal mass and reduce operator fatigue, but rugged enough for constant use.
• Powering products. The selection of power sources for the test system can be somewhat complicated. Much of this complication can be attributed to the multi-head test system. The initial selection of bulk or individual power sources typically is based on cost. Keep in mind that even though individual power sources cost more, they have several advantages over bulk supplies, such as embedded measurement capabilities, individual activation and separate sense line connections.
Because the distance between the power sources and the devices under test can be long, voltage drop issues can become a concern. Usually, sense lines are used to reduce this problem, but in bulk power supply situations, the sense lines can only be placed at the furthest common point—usually bus bars. Also remember to consider the elevated chamber temperatures when temperature de-rating the wires entering the chamber workspace.
A common mistake is to not interlock the supply of power with chamber operation. Imagine the predicament you could encounter if chamber cooling fails and the devices under test continue to pump heat into the insulated interior of the chamber.
• Loading. If electrical loads are needed for testing, problems similar to power sourcing can be encountered. Because it is difficult to keep the loads in close proximity to the product, complexities can arise. While normally not advisable, it may be possible to place loads—especially passive loads—on the product carriers in close proximity to the products. For both fixed and programmable loads, dissipating the heat from electrical loads can be a substantial engineering challenge, perhaps requiring the use of high velocity air or water cooling.
• Stimulus. Typically, there are “stimulus” signals that need to be sent to the product during environmental testing. While programmability and accuracy of these signals must be well thought out, it is important to consider synchronization and separate sources for each product. Just as with the application of power and loading, synchronization with the operation of the environmental chamber is useful. Separate parallel sources can prevent the failure of one product from corrupting a signal that is seen by all products under test.
• Monitoring. On the opposite “side” of the product, issues of product measurement and monitoring need careful consideration. Monitoring the product under test is extremely important, because it can not only help determine problem areas in a unit under test, but also can help fine tune the testing process itself. For example, if a five-day burn-in process has been performed for several months and all witnessed failures have occurred during the first two days of the burn-in, it may make sense to increase throughput by shortening the burn-in period. As mentioned previously, continuous monitoring or at least very fast multiplexing is needed to prevent overlooked soft failures.
When considering ways to solve the monitoring requirements, thought must be given to signal conditioning, synchronization, accuracy and resolution issues. A typical scheme for
detecting failures during a functional test is to set high and low limits on the monitored signal and to record a failure (along with time and environmental conditions) when the failure
occurs. When the failure occurs, remove power to the failed product to prevent a possibly catastrophic meltdown. Note that a somewhat decreased accuracy may be acceptable during functional testing, especially as a tradeoff for more monitoring channels.
• Software. It is worth mentioning issues related to test system software. In addition to the issues of user interface, operational features and usability, consider the amount of test data to be collected for a given test. The typical response is to collect a large amount of data and decide later what is important. This may work in some cases, but it can make for time-consuming and protracted test report generation. Lengthy “life testing” tests can make frequent data recording a formidable problem. A possible scheme to reduce recorded data, especially in functional test systems, is to record only failures, not “passing” information.
• Safety. The safety issues must not be ignored in engineering an environmental test system for power electronics. Dangers exist for operators or others inadvertently touching extremely cold or hot products. All environmental chambers should be equipped with a separate thermal alarm that signals unintentional temperature excursions. If nitrogen or carbon dioxide is being used, an oxygen monitor should be placed in the area. Interlocks should be provided to prevent the environmental chamber doors from being opened during a test and to shutdown test power if a door is opened. Remember to automatically disable product power if the chamber is inactive.
Integrating the Test System
Obviously, integrating a test system like this is not a trivial undertaking. Engineering disciplines, including electrical, mechanical and software, need to be brought together for successful implementation. Safety and regulatory standards must be met. Documentation must be extensive and accurate if future users are to expect success in using the system. The complexity of the test system can easily surpass the complexity of the product being tested (see “Test and Measurement Checklist”).
Integrating and implementing test and measurement with environmental testing can help you by increasing product robustness, improving product reliability, improving the production processes, increasing yields, and reducing warranty expenses and exposure. Despite these issues that must be successfully addressed, if history holds course, environmental testing will continue to contribute to the future success of many power electronic products.