Reliability Testing on OEM Systems

Introduction — OEM Systems with Limited Samples

I was assigned to a new project at HP as a reliability engineer was to develop a rear projection, large screen TV.   As the project was being defined, I was informed that an Asian Original Equipment Manufacturer (OEM) would be developing and manufacturing the HW while HP developed the SW.  HP would own the service center support of field issues.  My role was to represent the customer and evaluate the reliability of the final product.

As I got into the project, I was surprised to find out that the OEM would not share any of their internal data on the HW and any history of similar systems they produced.  HP would be provided 30 prototype TV units to evaluate SW performance.  The reliability evaluation phase would get one (1) unit for testing.  This is a good example of the limitations placed on reliability studies for large, OEM systems where test samples are severely limited.

The approach I took was to:

1. Define the functions in the project using the Functional Architecture (ref. REI book – page 23) and their linkages. Gain consensus between HP and the OEM.
2. Establish reliability goals for the project based on known competitor experiences and specified field repair costs. Share results with service and marketing groups.   Gain agreement within SW team to allow SW upgrades in the field using zip drives.
3. Testing the units available using two approaches (this will be discussed below):
   1. Use HALT testing on the one reliability unit by testing to failure (ref. REI book, pages 28-30)
   2. Use the other Proto units for Home evaluation.

HALT Testing of Large System

Since I did not have access to a chamber large enough to handle a large screen TV (for thermal cycling) and the vibration table was located in the shipping area, I took the strategy of making the TV itself a thermal chamber (putting a heater into the air intake vent of the TV with temperature sensors to monitor the internal levels).  The TV (with heater) was placed on top of the vibration table allowing me to operate the TV while performing vibration and thermal cycling stresses simultaneously. Cameras were placed inside the TV to capture any changes real time.

Varying operation settings helped to maximize electrical operation coverage. Thermal cycling and vibration started at low levels and increased until a failure occurred.   When a failure occurred, the problem area was repaired, and stresses continued to increase.  This allowed us to find all marginal areas on the one unit.

Failures found were loose cables, flexing PCBs, and marginal components.  These results were fed back to the OEM.   Fixes were implemented making the TV design more robust to shipping and long-term operation. 

Using Customer Simulation Before Product Release

Each project team member at HP (including the reliability engineer) was asked to take home one of the TVs during development and use it like they would their own.  The entire family was encouraged to be part of this test.   Feedback sheets were brought back to HP weekly.

The result was a combination of known issues (those the team was working on) and many unexpected issues that were added to the project deliverables.  Issues found were:

• HW issues: failures of power supplies due to lack of ventilation, projection lamp early failure, remote control button failure and screen image irregularities.
• SW issues: compatibility between HW and SW, incorrect sequencing, and limited performance capability.
• Other issues:
  • Documentation: setting-up TV, and selecting options
  • Procedures: updating of SW,
  • Expectations: This was the most enlightening to the project team.  The family members used the TV in unexpected ways and gave their inputs openly to the team members.  Some of the surprising expectations were.
    • Slow time for turning on TV,
    • Buttons on remote control were sluggish and not intuitive,
    • Viewing angle limited so whole family could not see screen,
    • Placement of TV on tables was sometimes unstable.

The use of the HALT and Home TV testing approaches helped to solve many issues before becoming customer experiences; thus, improving the overall TV performance.