1.2 Qualification, Acceptance, Development, Screening, Margin

After this section you can: say which of the five purposes a test serves, what a pass proves, and whether the article on the shaker is allowed to be damaged.

There are five reasons to put an article on a shaker, and they are not interchangeable. The profile can look identical on the controller screen and mean five different things.

Qualification proves the design survives the sequence with margin. It is run on one or a few articles built to the production drawings, tested harder and longer than service. The qualification (qual) article is often never fielded; it has been used up proving the point. A pass says the design is good. It says nothing about the unit built next month.

Acceptance proves this unit was built right. It is run on every unit, or a sample, at or near service levels, with no margin and no damage intended. Sometimes there is only one unit. The acceptance test is counted as part of that unit's life: it is there to find gross assembly errors and issues, not to take life from the unit. In that sense it screens each unit for workmanship, but it is not the screening purpose described below, whose levels are set to make a hidden defect fail. The purpose is workmanship: the cracked solder joint, the missed torque, the wrong part. A pass says this unit is fit to ship. It says nothing about the design; that was qualification's job. If a unit is damaged, everything stops, and all parties meet to decide the next steps and whether the hardware is still viable.

Development exists to learn something. Levels are chosen to find the weak point, not to pass. The article may be a breadboard (the circuits built up loose, before packaging), a partial build, or a qual unit run past its requirement on purpose. There is no pass or fail; the deliverable is knowledge, and the report reads differently: it says what was learned, at what level and on what build, not whether a requirement was met.

Screening, whether it is called environmental stress screening (ESS) or highly accelerated stress screening (HASS), is there to make hidden workmanship defects show up in production units, so they fail on the bench and not in the field. The levels are chosen to fail the bad unit and not the good one. That is a harder target than either qual or acceptance, and screening programs go wrong in both directions: set too low, the bad unit passes; set too high, the good unit is damaged or ships with life used up.

Margin finds out how much is left. The levels are stepped above qualification, on the qual unit after it has passed or on a unit set aside for the purpose, until something fails or until a defined margin has been demonstrated. The requirement here is the service environment as the program writes it down, often the measured or predicted level enveloped at the 99th percentile with 50 percent confidence, a common program convention; it already sits above the levels the article is expected to see in service. Qualification adds its margin on top of that level, and margin testing steps on from qualification. The deliverable is a number in dB, with the level it is counted from stated. On a random profile the decibels are on the acceleration spectral density, the ASD of 1.3: 3 dB doubles the ASD and raises the overall rms level (root mean square, the one number for how much vibration there is in the whole signal) by a factor of about 1.41, and 2.4 defines the rms and works the arithmetic. A test that ends in a failure gives two numbers, and the report carries both: the last level the article survived, which is the level demonstrated, and the level at which it failed. A test that reaches a stated margin without a failure stops there, and that level is the result. That number is what lets a program accept a heavier environment later without requalifying, extend a service life, argue down a requirement that was too conservative, or know which part goes first when something finally does. The article is expected to be damaged or destroyed, and that is the clearest line between margin and qualification: qual stops at the requirement plus its built-in margin; margin testing keeps going. The lab's part is the step size, the dwell at each step, what gets inspected between steps, and when to stop. A margin test with no stop criterion runs until the article breaks.

Development and qualification blur, and that is normal. A qual unit is rarely all production hardware. Mass mockups stand in for parts that are late, expensive, or irrelevant to the dynamics, and the test is still called qualification as long as the mockups carry the right mass, center of gravity, and stiffness at the interfaces that matter.

The other direction is more common than the textbooks admit. A new environment, or a gap in the flight data, is investigated by putting a highly instrumented unit on the shaker and driving it with whatever flight data exists, to fill in the response the field test never measured. That test looks like qualification on the controller and is development in purpose, and the levels it produces often become part of the next program's requirement.

A qualification test that keeps going past its pass/fail test level has become a margin test, whether or not anyone renamed it.

The lab's job is to know which parts of the article are real and which are mockups, which channels (the measurement channels the controller records, one per sensor) are there to learn unknown responses and which are there to qualify the article, and to define each channel's purpose in the report, because a reader ten years from now will not be able to tell from the profile what each channel was for. A report rarely carries too much; where its format is fixed, the extra material goes in an appendix.

Each purpose sits somewhere on the sequence from 1.1. Qualification, development, and margin answer for the whole sequence at once. Acceptance and screening answer for the factory step, because the question they ask is about the unit, not the environment. Margin is usually the last thing done to a qual unit, because it uses the unit up.

The trap in the lab is running a qualification profile on flight or production units, the units that will be fielded, because it was the profile on file. The unit passes, and it has used up life it still needed in service. The reverse trap is running acceptance levels and calling the design qualified.

One blend is sanctioned on purpose. Protoflight, common on spacecraft programs, runs a unit that will fly at qualification levels, usually for the shorter acceptance duration, so that one test qualifies the design and accepts the unit. A pass says both, with less margin shown on each than separate tests would show, and it uses up some of the unit's life. The program decides on it and approves it in writing before the test; it is not a profile the lab runs because it was on file.