AdvancedMedium risk

Intermittent fault that's hard to reproduce

Something fails occasionally — random trips, dropouts, or stoppages — but works fine when you go to look at it. The classic 'can't fault it on the bench' problem.

Safety first

Intermittent faults can cause unexpected operation. Stay alert for sudden movement/energising. Apply isolation whenever you work hands-on.

Isolate, lock out / tag out, and prove dead before working unless a live test is specifically required, authorised, and carried out under proper supervision. Always follow local regulations, your site procedures, and the equipment manufacturer's documentation.

Apprentice mode

Full walk-through — the why, what readings mean, mistakes & safety.

Likely causes

Ranked from most to least likely.

  1. 1

    Loose / high-resistance connection

    Most likely

    A loose terminal that makes and breaks with vibration or temperature is the most common intermittent.

  2. 2

    Temperature-dependent fault

    #2

    A component or joint that fails only when hot (or cold) appears and disappears with conditions.

  3. 3

    Vibration / movement-related

    #3

    A cable flexing on a moving part, or a connector working loose with vibration.

  4. 4

    Environmental (moisture, contamination)

    #4

    Damp or contamination causes leakage/faults only under certain conditions.

  5. 5

    Electrical noise / event-triggered

    Least likely

    A fault triggered by another event (a large load switching, a specific machine cycle).

Reports are saved on this device to reflect what you actually find.

Testing sequence

Work through one test at a time. Expected reading and what each result means.

On the tools? Job Mode shows one test at a time with thumb-sized buttons — your place and readings carry straight over.

Enter Job Mode
Test 1 of 3
1

Log occurrences with timestamps and conditions (load running, weather, machine step) to find a correlation.

Expected reading

A pattern linking the fault to a condition or event.

If it passes

A correlation narrows the cause group — target it.

If it fails

No pattern yet — apply provocation tests (heat/vibration/wiggle).

Mark this test
View all expected readings at once
1. Log occurrences with timestamps and conditions (load running, weather, machine step) to find a correlation.
A pattern linking the fault to a condition or event.
2. Provoke the suspect area: gentle heat/cool, tap/wiggle connections and cables, while monitoring.
The fault reproduces under provocation, revealing its location.
3. Use monitoring/data-logging (min/max, event capture) across the suspected signals to catch the event.
The logger captures the fault event with context.

Fault-finding flowchart

The same logic as a decision tree.

  1. 1
    start

    Intermittent fault

    → step 2
  2. 2
    decision

    Does logging conditions reveal a correlation?

    Yes→ step 3No→ step 4
  3. 3
    result

    Correlation found — target that cause group.

  4. 4
    decision

    Does provocation (heat/vibration/wiggle) reproduce it?

    Yes→ step 5No→ step 6
  5. 5
    result

    Localised the marginal point — repair it.

  6. 6
    result

    Use data-logging/monitoring to capture the event.

Common mistakes apprentices make

  • Replacing parts at random hoping to get lucky.
  • Not logging conditions to find a correlation.
  • Giving up because it 'works fine now'.
  • Overlooking connections in favour of components.

When to stop & escalate

Persistent intermittents that resist provocation and logging may need specialist monitoring or a planned approach with the wider team. Document everything so the next person isn't starting cold.

If you're past your competence, authorisation, or the safe limits of the job — stop and hand it on. There's no fault worth getting hurt over.

Was this guide helpful?

Related faults