Practical Technique
Intermittent Fault Capture
Catch intermittent automotive faults with a lab scope: pulse-width and window triggers, holdoff, roll mode, segmented memory, mask testing, long recordings, and safe unattended capture.
The intermittent problem
A scan tool's freeze frame shows conditions after a module noticed a fault. A scope set up correctly captures the fault itself: the dropout, the spike, the missing pulse. The work is in the setup: trigger on the abnormal event, record long enough, and make the fault happen while you're recording.
Step 1 — Trigger on the abnormal event
Pulse-width (glitch) trigger
Triggers only when a pulse is shorter or longer than a limit you set, and ignores every normal pulse. Examples:
- An injector held on too long: trigger on a low pulse longer than the longest normal pulse width.
- A crank sensor dropping a tooth on a 60-2 wheel: trigger on a period between 1.5 and 2.5 tooth periods. The normal gap is 3 periods, so this ignores it.
Window (level) trigger
Triggers when the signal leaves a voltage band you set: a 5 V reference dipping below 4.5 V, or a supply dipping below 10 V. A normal edge trigger can't catch a sag that never crosses its level.
Trigger holdoff
After a trigger, holdoff blocks new triggers for a set time. On signals with many identical edges, set holdoff just under one cycle so each capture anchors to the same event.
Step 2 — Choose the capture mode
| Mode | What it does | Use it for |
|---|---|---|
| Single | Captures one triggered event, then stops | A rare glitch; the capture waits for you |
| Normal | Recaptures on every trigger | Faults that repeat every few minutes |
| Roll | Scrolls continuously, no trigger | Slow changes: supply sag, sensor drift, module wake-ups |
| Segmented memory | Stores many short triggered captures back-to-back, skipping dead time between them | Catching every glitch in a long road test |
| Mask / limit test | Compares each capture against a tolerance envelope and saves or alarms on any that fall outside | Unattended capture of multiple events |
Step 3 — Record long enough
Capture time = memory depth ÷ sample rate.
| Memory | Sample rate | Capture time |
|---|---|---|
| 1 MS | 1 MS/s | 1 s |
| 1 MS | 10 kS/s | 100 s |
| 10 MS | 100 kS/s | 100 s |
Choose the lowest sample rate that still resolves the fault. For slow signals (supply voltage, sensor drift), a datalogging mode at low rates can record for hours.
Pre-trigger matters: set the trigger point so a large share of the capture is before the trigger. The conditions leading up to a fault usually explain it.
Step 4 — Provoke the fault
While the scope is armed:
- Heat: heat gun on suspect connectors, sensors, and modules.
- Cold: freeze spray on the same suspects, for cold-only faults.
- Vibration: tap connectors and grounds with a screwdriver handle.
- Load: switch on high-current loads (headlights, blower, rear defrost) to load the supply and grounds.
- Harness movement: flex the harness where it crosses pivots, heat sources, and brackets.
Step 5 — Unattended capture
Leaving a scope connected while a vehicle is driven carries safety and liability risk. Leads can reach moving parts or hot surfaces, and equipment can distract the driver. If you do it:
- Route every lead away from the exhaust, belts, fans, steering, and pedals, and secure it.
- Keep all equipment clear of the driver's area, fully secured.
- Test the trigger with a deliberate glitch before the vehicle leaves.
- Power the scope (and laptop, for PC-based scopes) from a source that stays on for the whole drive, and disable sleep modes.
- Use single-shot, mask testing, or segmented memory so a captured event isn't overwritten.
- Brief the driver: note when the fault happens, and don't touch the equipment.
Reading the capture
- Start with the pre-trigger data. What changed just before the fault?
- Compare the faulty event with normal events in the same capture. The signal's own normal behaviour is your reference.
- If the signal looks clean but the fault happened, the cause is elsewhere: capture power and ground at the same time next.
- Save the capture with its settings: timebase, range, sample rate, and trigger. A capture without them can't be interpreted later.
Frequently asked questions
What is glitch triggering?
A pulse-width trigger: it fires only when a pulse is shorter or longer than the limit you set, ignoring every normal pulse. It's the practical way to catch a one-in-thousands event.
When should I use roll mode?
For slow events, such as supply voltage sagging or a sensor drifting, where you want a continuous chart rather than a triggered capture. Roll mode suits slow timebases.
How do I capture a fault that only happens after a heat soak?
Arm a pulse-width or window trigger on the affected signal in single-shot mode, bring the engine to temperature, and apply heat or cold to suspect components while the scope waits.
Can I leave a scope recording in a customer's vehicle?
Only with every lead secured away from moving and hot parts, all equipment clear of the driver, and a trigger mode that can't overwrite the capture. Weigh the risk before doing it.