Recreating the Conditions to Find the Fault

Why this matters

A fault you cannot make happen is a fault you cannot test, and a fault you cannot test is one you are guessing at. The most expensive diagnostic mistake is replacing a part that measured fine because the failure was not present when you measured it. Many faults are conditional: they appear only when the equipment is hot, under load, at a certain time of day, after running a while, or when several things line up at once. If you only ever check the system in its calm, idle state, the very condition that breaks it never gets reproduced and the part always passes.

Learning to recreate the conditions is what separates a real diagnosis from a parts swap that happens to work or happens to fail again next week.

Step 1: Pin down the conditions from the story

Before you touch anything, mine the customer's account for the trigger. The words "only," "after," "when," and "sometimes" are gold. "It only does it in the afternoon" points at heat or load. "After it runs a while" points at a thermal or duty-cycle fault. "When the [other appliance] kicks on" points at a shared circuit or a voltage sag. "Every morning" points at a cold-start condition.

Write the suspected condition as a sentence you can test: "Fault appears once the unit has run for roughly twenty minutes and reaches operating temperature." A vague "it acts up sometimes" is not testable; a specific condition is.

Step 2: Reproduce the trigger deliberately

Now force the condition rather than waiting for it. The whole point is to stop being a passive observer and become the one who makes the fault happen on demand:

• Thermal faults: run the equipment until it reaches the temperature the customer described, or apply controlled heat to a suspect component while watching for the symptom. Many connections and electronic parts fail only when hot and expand back into spec when cool.
• Load faults: put the system under the actual load that triggers it, not a bench idle. A motor that starts fine unloaded may stall or draw excessive current under real demand.
• Vibration or movement faults: gently flex a harness, tap a board, or move a component while monitoring. An intermittent open often reveals itself the moment you disturb the bad joint.
• Timing faults: if it only happens at start or after a long run, recreate that exact cycle, do not shortcut it.

Step 3: Instrument before you trigger

Have your meter, gauge, or logger already connected and reading before you induce the condition, because the fault may last a fraction of a second. Watching a display at the moment of failure tells you which value moved. Where the fault is too brief for the eye, use a tool that captures: a meter with min/max or record mode, a data logger, or even the customer's phone video of the symptom as it happened. A captured reading at the instant of failure is worth more than an hour of stable readings while the fault hides.

Step 4: Confirm the fault is the same one

When the symptom reappears, verify it is actually the customer's complaint and not a different problem you stumbled into. Match what you see against the reported behavior. If the customer said "it shuts off and the lights flicker" and you produced a shutdown with no flicker, you may have found a second fault, not the one you were hunting. Reproducing the wrong failure and fixing it is how a callback gets born.

Step 5: Prove the repair under the same conditions

A fix verified only at idle is not verified. After the repair, recreate the exact triggering condition again and confirm the fault is gone. If it took heat and load to make it fail, it takes heat and load to prove it is fixed. This is the single most-skipped step and the most common source of "but it worked when the tech left." The condition that broke it is the condition that clears it.

When you genuinely cannot reproduce it

Sometimes the trigger will not appear in your window on site. Do not paper over that by replacing a part you never saw fail. Be honest about it: leave monitoring in place if you can, document the suspected condition and what you ruled out, and set the expectation that the next occurrence is the diagnostic opportunity. A documented "could not reproduce, here is the plan to catch it" is a professional outcome. A confident part swap on an unreproduced fault is a coin flip you are charging the customer to lose.

References

• Manufacturer service documentation, troubleshooting flowcharts (standard guidance on duty-cycle and load-dependent fault verification).
• Electronics Technicians Association International (ETA), diagnostic and root-cause methodology.
• See related: Intermittent Fault: Cannot Reproduce Decision Tree; It Worked but You Don't Know Why Decision Tree; Load-Dependent Fault Decision Tree.