Break it on purpose

John Baron
Director of Operations
Sypris Test & Measurement
Burlington, Mass.

Adam Bahret
Reliability Engineer/Supervisor
Varian Semiconductor
Equipment Associates
Goucester, Mass.

It's often crucial to test any product before it moves from design to production. More specifically, a test program called Highly Accelerated Life Testing, or Halt, can be a way to give designers valuable information that is unavailable by any other means.

Halt subjects products in the design phase to a series of escalating stresses until there are failures. Although Halt destroys the prototype, the process benefits many final designs for all types of equipment and components. Engineers at Sypris Test & Measurement Inc., for example, run Halt on a variety of telecommunication, medical, semiconductor, military, and avionics equipment.

Sypris engineers recently teamed with designers at Varian Semiconductor Equipment Associates Inc. to perform Halt on Varian's system controllers. The controllers operate an integral pump critical to ion implantation in semiconductor processing. Halt uncovered problems designers did not anticipate that let Varian designers fix issues early in the design phase.

Halt exposes product-failure modes and identifies weak links by subjecting equipment to a series of thermal and vibration stresses well beyond normal equipment design and operation limits. Its objective is to force failure modes or weaknesses to the forefront through virtual-time compression methods. Units under analysis never "pass" this test because the goal forces failures on an accelerated basis.

Failure modes are not necessarily hard failures or where the item stops operating completely. They can be soft failures where the item returns to proper operation once stresses are removed. And the item may not completely stop operating. Failure can take the form of degraded performance or an intermittent anomaly. All of these things point to an inherent weakness in the design.

A Halt analysis typically lasts three to five days depending on how quickly a particular unit stabilizes between tests and the operational requirements to verify equipment functions during each step. The tests takes place in special chambers designed to produce controlled vibrations and rapid changes in temperature across broad thermal ranges.

Somewhat contrary to its name, Halt does not replicate wear on products under normal use. Stresses in the Halt environment only expose product failure modes or weaknesses. There is no common or easy correlation between the stress loads and failures in Halt and those seen by products during normal operation.

Test engineers are gunning for two types of failures throughout the process: hard or nonrecoverable failures and soft, or recoverable, failures. The extreme testing does not validate a component's capability to operate in extreme temperatures or under severe vibrations. Most components aren't built to withstand the extremes used during Halt. Instead, Halt helps designers determine the failures and weaknesses that happen first during exposure to extreme environments. These failure modes point out inherent weaknesses in a product during its intended lifespan.

Halt also helps determine actual tolerance margins beyond the designated operating and design range. If a component breaks down at just 5 or 10°C away from its operating range, product capabilities may be too marginal to be considered a robust design. Conversely, a component refusing to break down until it is 50 or 60°C from its operating range assures confidence that a robust operating margin exists. A failure at that point shows the component is capable of operating at extremes well beyond its normal operating range.

Halt's success depends on the involvement of product designers, often working side by side with Halt reliability engineers. The designers bring an intimate grasp of product functions and weaknesses. They determine which stresses matter most in the testing. The designers analyze the results to determine the root cause and take necessary corrective actions in the design to minimize future failures.

As previously stated, Halt is a destructive test to find key failure points during product design. After products reach production, a Highly Accelerated Stress Screen, or Hass, helps maintain quality. Hass is a nondestructive screening process based on Halt findings.

After the temperature and vibration hard limits are found with Halt, engineers can extrapolate a Hass profile of less destructive limits that quickly and aggressively screens products for defects. Hass can reveal deficiencies in both quality and integrity of the product and the production process. The process ensures the units hit the expected levels of durability and reliability. With operating margins determined by Halt, Hass allows higher stress levels than conventional Environmental Stress Screens (ESS). The more aggressive Hass levels are helpful tools in screening to maintain a high level of quality not possible by classical ESS methods. Hass and Halt combine to dramatically shorten the time needed for product test.

HALT IN ACTION
Varian Semiconductor Equipment Associates Inc. produces ion-implantation equipment used in the manufacture of semiconductors. Engineers from Sypris performed tests using Halt protocols on Varian's vacuum system controller used in the ion implantation process.

Sypris engineers set the vacuum system controller inside a Screening Systems Inc. QRS-410T Halt chamber. The unit consists of a power supply and three modules, each with one I/O board, a CPU board, and a Bit-Bus board. Thermocouples fitted to various points on the unit monitored operating temperatures. The open controller chassis had forced-air blown across its internal components to ease thermal transition and stabilization during the test. Outside the chamber were the power source, an RS-232 filter set, a test computer to compile results and run the controller software, and a system pump simulator.

Testing followed the normal course of Halt's five phases: A Low Temperature Step Stress lowered temperatures in 10° steps from 20 to 60°C. A High Temperature Step Stress raises temperatures in 10° steps from 20 to 120°C.

Rapid Thermal Cycling rapidly varies the temperature between hot and cold limits. A Vibration Step Stress applies incremental vibrations up to 15 grms across the six axes of motion. Last, a Combined Environment Testing simultaneously applies both the Vibration Step Stress and Rapid Thermal Cycling tests.

Engineers recorded many events during the extensive test, but most were inconsequential. However, two events at the extremes of thermal and vibration testing revealed several failure modes spotlighting the effectiveness and benefits of Halt.

During the Low Temperature Step stress test, the unit was left to stabilize at each step for a minimum of 10 min. After the allotted time elapsed, engineers put the controller through several power cycles. At the 60°C step, the system started but lost power for a second and then rebooted. Pump 1 showed no data although the pump was rotating. Pumps 2 and 3 worked normally.

The system quit functioning after another power cycle. The engineers raised the temperature to 50°C to verify operation. In two consecutive power cycles, the unit operated properly each time. Varian took no corrective action because the 60°C failures were far outside the normal operating range of the controller.

The second test failure happened during the vibration analysis. Test personnel clamped the system controller to the six-degree-of-freedom vibration table in the chamber. The fixture and test units were positioned to allow for the best energy transfer from the table to the unit. Six control accelerometers attached to the vibration table monitored and controlled vibration levels. The overall composite vibration spectrum was analyzed and calculated using the root-meansquare method (grms).

The grms method gives a single feedback value for vibration that includes a wide-band frequency range acting in all three axes simultaneously. Engineers put accelerometers at various locations on the system controller. They fed into a digital vibration analyzer to measure the unit's response to the input (table) vibration levels. The vibration test started with a low-level vibration input of 3 to 5 grms. Each vibration level lasted for 10 min and then rose in 120% steps to a maximum level of 15 grms.

Test engineers discovered the vibrations shook loose two machine-screw nuts. The nuts held together the three standoff-separated circuit boards in the controller module. The addition of locking compound on the nuts solved the problem.

There was a critical failure at 15 grms when test engineers noticed electrical sparking on a network card. They stopped the test at which point the unit lost power. Visual inspection revealed wire leads from a voltage regulator had broken off a CPU board on a network module. The network card handles the communication between the highlevel implanter computer and the ion-implantation pump, so a communication failure of this magnitude would make the whole unit fail. Another production change secured the 5-V regulator to the CPU board using a componentbonding compound.

Finally, the test engineers subjected the system controller to simultaneous vibration and thermal cycling. Throughout the process, the test units tended to shut down at high temperatures and several leads broke from fatigue, confirming and validating earlier tests.

Thanks to Halt, Varian has a clearer picture of the system controller that is so crucial to its ionimplantation process. Testing gave the company confidence in the performance of its product. In addition, test engineers identified several parts that needed mechanical reinforcement. Over the long run, the findings should save Varian and its customers both time and money. The Halt program produced a baseline performance upon which to gauge future changes to the vacuum system, manufacturing process, or components.

MAKE CONTACT
Sypris Test & Measurement, (800) 775-2550, wetest.com
Varian Semiconductor Equipment Associates, (800) 447-1762, vsea.com