Fault Only Under Specific Conditions

Why this matters

The hardest faults to find are the ones that hide when you show up. It works fine on the bench, fine while you watch, then fails the moment you leave or the moment a specific condition returns. These are not mysteries - they are condition-dependent faults, and the trick is to identify the condition and reproduce it on purpose. A fault you can reproduce is a fault you can find. A fault you cannot reproduce is a guess.

Start here: do not trust "it works now"

The most common error on an intermittent fault is testing under easy conditions, finding nothing, and calling it fixed. It fails under a specific condition; testing outside that condition proves nothing. Your first job is not to test - it is to figure out what condition makes it fail, then recreate that condition while you measure.

Interview hard for the pattern. When exactly does it fail? Hot or cold? Busy or idle? Wet or dry? Morning or afternoon? After running a while, or right at startup? The condition is the diagnosis.

Step 1: is it temperature-dependent

Heat is the most common hidden variable.

• If it fails only when hot (after running a while, or on a hot day), suspect a part that changes with temperature: a connection that expands and breaks contact, a component whose value drifts as it heats, a marginal device that survives cold but not hot. Run the system until it reaches the failure temperature, then test - do not test it cold.
• If it fails only when cold (first thing, or in cold weather), suspect contraction opening a connection, a fluid too thick to move, or a component weak until warmed.

Step 2: is it load-dependent

• If it fails only under heavy load or high demand but runs fine lightly loaded, suspect a marginal supply or a weak component that copes until pushed. The fault is real but only appears when stressed. Reproduce by applying the load that triggers it, then measure.
• If it fails only at a specific point in the cycle, the condition is a stage of operation. Watch the whole cycle and catch it at that stage.

Step 3: is it movement or vibration dependent

• If it comes and goes with vibration, or when something is moved or flexed, suspect a loose or cracked connection, a chafed wire, or a marginal mechanical contact. Reproduce by gently flexing or tapping suspect areas while watching for the symptom to appear. A reading that jumps when you wiggle a connection has just named the fault.

Step 4: is it moisture or weather dependent

• If it fails only when wet, humid, or right after rain, suspect moisture bridging a connection, water intrusion, corrosion that conducts when damp, or a seal that leaks under specific conditions. These hide completely on a dry day. If you cannot wait for weather, recreate the dampness deliberately and safely.

Step 5: is it time or usage dependent

• If it fails on a schedule or after a usage threshold, something is accumulating: heat building, a part filling, a buffer overflowing, a timer expiring. Map the timeline to find what crosses a limit when it fails.

The core technique: reproduce, then measure

Everything above is one method applied to different variables: identify the condition, recreate it on purpose, and have your meter or eyes on the suspect at the moment it fails. The fault that will not show up for you is just a fault whose condition you have not yet recreated.

When you cannot reproduce it on site at all, the fallback is monitoring - leave readings logged, or instruct the customer exactly what to note when it next fails (time, temperature, what was running). A precise failure report from the customer is the next-best thing to catching it yourself, and it points your return visit straight at the condition.

References

• Trade-standard intermittent-fault diagnosis practice
• Manufacturer operating-condition specifications
• See related: Testing a Component In vs Out of Circuit
• See related: It Worked Yesterday: What Changed