Symptom Won't Reproduce on Arrival (Decision Tree)

Why this matters

"No fault found" is one of the most awkward calls in the trades. The customer swears it failed, you arrive, and everything runs perfectly. Walking away with a shrug burns the relationship and almost guarantees a return trip. The skill is to treat "works now" as a clue, not a dead end, and to leave with either a real diagnosis or a plan that captures the fault next time.

Start here: do not declare it fixed

The most expensive mistake is telling the customer "it's working fine, must have been a fluke" and leaving. If it failed once, the fault is real and still there. Your job is to find why it is hiding right now.

Branch 1: did the act of arriving change something?

Often the fault clears because you disturbed the system getting to it.

• Did you open a panel, move a cover, or unplug to inspect? Reseating a loose connection by accident makes the fault vanish. If so, you have actually found a loose connection - now locate the exact joint before you button it up.
• Did power cycle when you arrived or when the customer reset it before you came? A fault that clears on reset points to a control that latches or a thermal device that has cooled.

Ask the customer plainly: "Did you turn it off, reset it, or unplug anything before I got here?" Half the time the answer explains the mystery.

Branch 2: are the failure conditions even present?

A system only fails under the conditions that cause the fault. If the trigger is not present, the fault cannot appear.

• Temperature: if it only fails on the hottest part of the day or first thing on a cold morning, a mid-day mild visit will never reproduce it.
• Load or demand: if it only fails under heavy use, an idle test proves nothing. Put the system under real load before you judge it.
• Time: if it fails only after running for an hour, a five-minute check is too short. Let it run.
• Moisture or weather: if it fails only when wet, you need to recreate the wet condition or wait for it.

Recreate the failure conditions deliberately. Most "won't reproduce" calls are really "wasn't tested under the right conditions."

Branch 3: hunt for the fault's fingerprint

Even with the system running clean, the prior failure usually left a mark. Look for:

• Heat damage: discoloration, melted insulation, a faint scorch smell at a connection.
• Corrosion or moisture tracks on connectors and boards.
• Loose, backed-out, or pitted terminals and contacts.
• Stored fault history, error counts, or indicators the equipment may keep.

A physical fingerprint lets you fix the fault even when it refuses to act up.

Branch 4: still nothing? Set a trap

When you cannot reproduce it and find no fingerprint, do not swap parts blindly. Instrument it and come back:

• Leave a recording meter or logger on the most-suspect signal.
• Mark suspect connections so you can see if anything shifted.
• Give the customer a simple log: date, time, exact conditions, what they saw.

This is honest and it works. You return with recorded evidence and run a normal diagnosis instead of a guess.

What to tell the customer

Be straight: "It is running fine right now, which tells me the fault only shows up under certain conditions. Here is what I checked, here is what I am leaving in place to catch it, and here is what to write down next time." That keeps trust intact and sets up the real fix.

References

• Trade-standard "no fault found" practice: condition recreation, fingerprint inspection, data logging.
• Manufacturer documentation on stored fault codes and diagnostic indicators.
• See related: Intermittent Fault How to Catch It, Reading the Failure Pattern.