Assembly testing: pull, torque and function

Three test families, three questions

An assembly test plan usually mixes three kinds of check. A pull or push test asks whether a joint holds. A torque test asks whether a fastener or a terminal can be tightened to a value and stay there. A function test asks whether the assembly does its job when it is operated.

The three are not interchangeable. A bracket that survives a 200 N pull can still rattle because the torque on its screw was never controlled. A terminal that holds torque can still fail a pull test if the crimp behind it is wrong. A unit that passes both can still fail a function test because a lever binds at the end of its travel.

Build the plan around the failure modes the design review identified, not around the equipment the supplier already owns. If the risk is a joint opening under vibration, the plan needs a pull or a vibration check, not a cosmetic inspection.

Pull and push tests: what the number means

A pull test applies a tensile load to a joint and records the load at which it separates or the load it holds for a stated time. For a crimped terminal, a push test may be used to check retention in the opposite direction. For an insert in a moulded boss, a pull test is the usual check that the insert will not come out.

The number only means something with the method attached. State the direction of the load, the rate of application, the fixture that holds the sample, the free length between the grips and the failure mode that counts as a pass. A pull test at 50 mm per minute and one at 5 mm per minute can give different results on a plastic boss.

Record where the failure occurred, not only the load. A pull test that tears the plastic around an insert is telling you about wall thickness or melt temperature. One that pulls the insert out clean is telling you about insertion depth or knurling. The load is the same; the corrective action is different.

For a moulded boss, the test also depends on the temperature. A pull at 23 degrees Celsius and a pull at the maximum service temperature can differ by a wide margin, because the polymer softens. If the joint sees heat in service, test at that temperature or state why not.

Torque: tightening, breakaway and re-tightening

Torque testing has three distinct numbers. The assembly torque is the value the operator applies. The breakaway torque is the torque needed to loosen a joint that has been tightened and, usually, aged. The re-tightening torque is what the joint takes on a second pass, and it shows whether the thread or the plastic has been damaged.

For a terminal screw, the assembly torque and the breakaway torque are both part of the specification. For a screw into a moulded boss, the useful test is often a strip test: tighten to the specified value, then increase until the thread strips, and record the margin. A boss that strips at 1.2 times the assembly torque has almost no margin for a powered driver.

State whether the torque is applied dry or with a lubricant, and whether a washer is present. A dry thread and a waxed thread at the same torque produce different clamp loads, and the joint sees the clamp load, not the wrench reading.

Function: run it as it will be used

A function test should reproduce the duty the assembly sees in service, at the conditions that matter. For a switch assembly that means operating the mechanism through its full travel, checking that it latches and releases, and confirming continuity or isolation at the right points. For a housing with a lid, it means opening and closing the lid for a stated number of cycles and checking the latch force at the end.

Add the environmental condition only where it changes the result. A latch that works at 20 degrees Celsius may bind at minus 20 if two materials contract at different rates. A seal that passes dry may leak after a humidity soak. The test standard and the condition belong in the specification together.

Keep the function test close to the assembly line. A test that requires a laboratory and a week of queue time cannot catch a drift in the same shift, and drift is what the test is for.

Sampling and acceptance

Sampling has to match the risk. A destructive pull test cannot run on every part, so it runs on a sample per batch or per shift. A non destructive function test can run on every unit if the cycle time allows. A torque check sits between the two: it can be non destructive, but the gauge and the operator time have a cost.

Write the acceptance rule before the first sample is pulled. A rule that says the mean must exceed a value allows a weak part to hide behind a strong one. For a joint that must not open, the useful rule is a minimum on every sample in the sample set.

Fixtures, gauges and repeatability

Most disagreements about a test result come from the fixture. A pull test that grips the sample at a different point from the last lab changes the lever arm and the answer. A torque driver that is out of calibration turns a good joint into a rejected one, or hides a bad one.

Build the fixture to hold the sample the way the assembly holds it. If a bracket is fixed by two screws in service, the test fixture should use two screws at the specified torque, not a vice. If the part is tested in a jig, the jig becomes part of the measurement system and needs its own check.

Run a repeatability check before accepting the test. Test the same sample several times, or several samples from one cavity, and look at the spread. A spread wider than the tolerance band means the test cannot decide pass or fail, whatever the specification says.

Records and what to do with a failure

A test record should let someone else repeat the test and reach the same conclusion. That means the sample identity, the batch, the fixture, the gauge, the calibration date, the rate, the environment and the result, including the failure mode.

Agree the reaction plan in advance. A single failed pull sample on a batch of 5,000 parts should trigger a defined sequence: quarantine the batch, check the retained samples, review the process parameters, then decide on sorting or scrap. Without that sequence, the first failure becomes a conversation about whether the test was fair.

Feed the results back into the drawing. If every batch passes a pull test with twice the required margin, the test may be checking nothing. If batches sit close to the limit, the design needs margin, not more inspection.

Keep one retained sample per batch where the test is destructive, labelled with the batch and the result. When a customer asks why a shipment behaved differently, the retained sample answers the question faster than a review of the paperwork.