Diagnosing Intermittent Faults

Written by Anthony Calhoun, ASE Master Tech A1-A8

Why Intermittent Faults Are the Hardest Diagnostic Challenge in the Shop

Every experienced technician has a vehicle in their history that still bothers them. The one that came in three times, never acted up on the lift, went home fixed on paper, and came back two weeks later with the customer furious. Intermittent faults are the great equalizer in the automotive trade. They do not care how many years you have been wrenching. They do not care how good your scan tool is. If the fault is not present when you test, you cannot diagnose it. That is the core of the problem, and pretending otherwise gets shops in trouble fast.

The fundamental challenge is this: most diagnostic processes depend on measuring something that is broken right now. You read a voltage. You check a resistance. You watch a waveform. When a fault is intermittent, the circuit or component is working correctly at the moment you are testing it. Your meter says it is fine. Your scan tool says it is fine. The customer is sitting in the waiting room wondering why you cannot find something that almost left them stranded on the highway twice last week.

Some shops refuse intermittent work entirely. They tell customers to bring it back when it is acting up consistently. On the surface that seems like a reasonable policy. In practice it is a mistake for several reasons. First, it sends a paying customer to your competitor, who may actually figure it out. Second, intermittent faults almost always get worse over time. The vehicle will be back, and now the customer has less patience and less trust. Third, there is real money in intermittent diagnostics if you build the right process around it. The shops that have learned to work these cars systematically are the ones customers brag about to their friends.

The frustration on the customer side is real too. They have been driving a vehicle that behaves unpredictably. Maybe it stalled in traffic. Maybe the transmission slipped going onto the freeway. Maybe the check engine light comes on for two days and then goes off. They have been to two other shops who told them they could not find anything. By the time they reach you, they are skeptical, tired, and expecting to be disappointed again. How you handle the first conversation sets the tone for everything that follows.

Getting the Right Information from the Customer

The customer interview is the most underrated step in intermittent diagnostics. Most technicians skip it entirely or let the service advisor handle it with a generic complaint line on the repair order. That is throwing away the best data you have. The customer has been living with this fault. They have noticed things, even if they do not know how to describe them technically. Your job is to pull that information out in a structured way.

Start with conditions. Ask specifically: does the fault happen when the engine is cold, after it warms up, or only once it has been running a long time? Hot soak restarts are a classic trigger for heat-related failures. A symptom that only appears after the engine has been off for thirty minutes and then restarted tells you something very specific about thermal behavior. Cold morning startups versus a fully warmed highway cruise are completely different operating conditions from a sensor and actuator standpoint.

Ask about weather and environment. Does it happen more in wet weather? Rain can expose moisture intrusion in connectors, antenna grounds, and body harness routing. Does it happen in heat but not in cool weather? That points toward a heat-sensitive component or a connector that expands and loses contact when hot. Does it happen after washing the car? That narrows your search significantly.

Ask about speed and load. Does it happen at idle, at highway speed, under hard acceleration, or coasting downhill? A fault that only appears above 60 mph on the highway but never in city driving tells you it is likely load or vibration dependent. A fault that appears only at idle but clears when you blip the throttle points toward idle control, vacuum leaks, or a sensor that reads incorrectly at low airflow.

Ask about fuel level. This one surprises a lot of techs, but fuel level matters for evaporative emission faults, fuel pump NPSH issues, and fuel level sensor accuracy. A customer who says it always happens when the tank is below a quarter tank is giving you a direct lead.

Ask how often it happens and how long it lasts. Once a week for two seconds is a very different fault profile than three times a day for ten minutes. Ask if anything makes it better or worse — like turning the heat on, hitting a bump, going over railroad tracks, or turning the steering wheel. Every one of those details points you toward a subsystem.

Write everything down on the repair order in the customer's words, then translate it into technical conditions. That document becomes your test plan.

Freeze Frame Data — Your Best Starting Point

When a fault code sets, most modern vehicles capture a snapshot of operating conditions at the moment the fault occurred. This is freeze frame data, and it is the closest thing you have to being there when the fault happened. Do not skip it. Do not glance at it and move on. Read it carefully and think about what it is telling you.

Freeze frame typically captures engine RPM, calculated load, coolant temperature, short-term and long-term fuel trims, manifold absolute pressure or mass airflow, vehicle speed, and throttle position. Some manufacturers capture additional PIDs depending on the system. The combination of those values tells you the operating state of the engine at the exact moment the PCM decided something was wrong.

Look at coolant temperature first. Was the engine fully warm, cold, or somewhere in between? A freeze frame showing 40 degrees Fahrenheit coolant temp tells you the fault happened at a cold start. A freeze frame showing 195 degrees tells you the engine was fully warmed up. That single data point cuts your suspect list significantly.

Look at fuel trims. Short-term and long-term fuel trims at the time of fault tell you whether the engine was running rich or lean when the code set. If long-term fuel trim was already at plus 20 percent and the fault code is a lean condition, the PCM was already working hard to compensate before the code set. That tells you this has been a developing problem, not a sudden failure.

Look at engine load and RPM together. High load at highway speed versus low load at idle eliminates half your suspect components immediately. A misfire code with a freeze frame showing 2,500 RPM and high load points toward a fuel delivery or ignition performance issue under demand. The same misfire code at 700 RPM idle with low load points toward a vacuum or idle circuit problem.

The limitation of freeze frame is that it only captures one moment. If the fault has set multiple times, some scan tools store multiple freeze frames. Pull all of them and compare. If the conditions are similar across multiple events, you have a pattern. If they vary widely, the fault may have multiple triggers or the root cause is more fundamental.

Mode 6 Data — Catching Trends Before They Become Failures

Mode 6 is one of the most underused diagnostic tools available on OBD-II equipped vehicles. It gives you access to the raw results of on-board monitor tests — the actual measured values compared against the minimum and maximum thresholds the PCM uses to pass or fail a monitor. A monitor that has not failed yet but is trending toward the failure threshold is a fault in progress, and Mode 6 shows you that trend before it becomes a hard code.

When you pull Mode 6 data, you are looking at test identifiers, or TIDs, that correspond to specific monitor tests. The scan tool displays the measured value alongside the minimum and maximum acceptable limits. A catalyst efficiency test that shows the measured value at 90 percent of the failure threshold is telling you the catalyst is degrading even if it has not set a code yet. An oxygen sensor response time test that is nearly at the slow limit tells you the sensor is aging and on the way out.

For intermittent diagnostics, Mode 6 is valuable because it gives you a window into how a system is performing when the fault is not present but conditions are marginal. A misfire monitor with a high misfire count on a specific cylinder that has not crossed the MIL threshold yet confirms cylinder-specific misfire even without a stored code. You can use that data to focus your testing before a code ever sets.

Not every scan tool displays Mode 6 data clearly. Some factory scan tools and higher-end aftermarket tools translate the TIDs into readable descriptions. Generic scan tools may show raw hex values that require cross-referencing to the OBD-II specification or manufacturer documentation. The investment in learning to read Mode 6 pays off regularly on difficult intermittent cases.

Wiggle Testing — Systematic Circuit Manipulation

Wiggle testing is exactly what it sounds like, but doing it wrong wastes time and doing it right finds faults that nothing else will. The goal is to physically manipulate wiring harnesses, connectors, and components while monitoring the circuit in real time, so that when the fault appears you know exactly what you just moved.

Set up your monitoring before you start touching anything. Connect your scan tool and go to live data for the affected circuit. If you have a lab scope, connect it to the signal wire you are testing. Set the scope to a time base that lets you see signal dropouts clearly. On a digital signal that should be switching, you want to see any interruption in the pattern. On an analog signal, you are watching for spikes, drops, or noise that should not be there.

Start at one end and work systematically toward the other. Begin at the component connector. Wiggle it gently in all directions while watching the monitor. Rock it front to back, side to side. Pull and push along the axis of the connector body. If the signal changes, you have your location. If nothing happens, move six inches down the harness toward the module or PCM and repeat. Work the entire length of the harness section by section.

When you find a section that causes a reaction, get more specific. Is it the connector itself, the terminals inside it, or the wire just before or after the connector? Pull apart the connector and inspect terminal tension. Terminals lose their spring tension over time, especially in high-heat environments. A terminal that looks perfect visually may have lost enough tension to make intermittent contact. Use a terminal tension gauge or a mating terminal to check how much resistance you feel when inserting. Compare to a known good connector of the same type.

Document every connector you tested and the result. If a shop owns this car for a week of intermittent testing, the next technician needs to know exactly what was already ruled out.

Environmental Testing — Forcing the Fault

If wiggle testing does not reproduce the fault, environmental testing is the next step. The goal is to recreate the conditions under which the fault occurs — specifically temperature extremes, moisture, and vibration — in a controlled way while monitoring the circuit.

Heat testing uses a heat gun applied carefully to individual components and connector areas while monitoring live data. Heat guns produce high temperatures quickly, so keep moving and do not apply sustained heat to plastic components that will melt. You are looking for a fault that appears as the component reaches operating temperature. This reproduces the classic hot-soak failure pattern where a component works fine cold and fails after twenty minutes at highway speed. Cracked solder joints inside modules, heat-degraded semiconductor junctions, and connectors with terminal plating that breaks down under heat all respond to this test.

Cold spray testing uses refrigerant-based aerosol sprayed directly on a suspected component to drop its temperature rapidly. This is most useful for components that fail when cold and recover when warm, or components that the technician suspects of thermal expansion causing a mechanical fault. Apply cold spray and watch for the fault to appear immediately after cooling.

Water spray testing is your go-to for moisture-related faults that the customer reports happen in rain or after washing. Use a garden sprayer to apply a light mist to harness routing areas, connector locations, and potential moisture intrusion points while the system is monitored. Body grounds are the first place to test. A body ground connection that allows moisture to wick between the terminal and the sheet metal will show increased resistance the moment water is introduced.

Vibration testing uses a rubber mallet to tap on components, brackets, and harness sections while monitoring. This reproduces road vibration that can cause cracked solder joints in modules to open and close with vehicle movement. Tap the PCM, tap the module housings, tap along harness supports. A fault that appears when you tap a specific location is extremely actionable diagnostic information.

Data Logging and Recording

When you cannot reproduce the fault in the shop, you need to capture it while it happens naturally. Data logging is how you do that. The principle is simple: set up recording, put the vehicle back in service under real-world conditions, and review the recording after the fault occurs.

Most professional scan tools support a data logging or record function. Set up a parameter list that covers the relevant circuits for your suspected fault. Do not log every PID available — the sample rate drops as you add more parameters, and you may miss a brief dropout. Focus on the signals most likely to show the fault: the affected sensor output, related reference voltages, and key operating conditions like RPM and load for context.

Set the scan tool to trigger recording on a specific condition if that feature is available. Some tools let you define a trigger value — for example, start recording when oxygen sensor voltage drops below a threshold, or when misfire counter on cylinder three increments. Triggered recording captures the moments before and after the event, which is often more useful than a continuous log where the fault moment is buried in hours of normal operation data.

Lab scopes with record functions work the same way. Set your trigger condition, set the pre-trigger capture depth, and let the scope run. When the fault occurs, the scope captures the waveform around the event. Review the captured waveform and look for the exact moment the signal deviated from normal.

On a multimeter, use the min/max function. Set the meter to the appropriate measurement, activate min/max recording, and tape the meter to the interior of the vehicle. The customer drives normally. When they return, review the min and max values captured. A ground circuit that should read under 0.1 volts that shows a max of 2.3 volts tells you there was a voltage drop event even if the circuit is reading perfectly normal right now.

For cases that require extended monitoring — days or a week of customer-drive data — a dedicated data logger left in the vehicle is the right tool. Affordable CAN bus loggers and OBD-II data loggers are available that record continuously to a memory card. The customer drives normally, the logger captures everything, and you review the data when the car comes back.

Common Intermittent Failure Patterns

After enough intermittent diagnostics, patterns emerge. Certain failure modes show up again and again across different makes and models. Knowing the common patterns lets you focus your testing faster.

Corroded ground connections are the single most common root cause of intermittent electrical faults. Body grounds, engine block grounds, and battery negative cables develop resistance over time from oxidation, moisture intrusion, and vibration loosening. A ground with two ohms of resistance causes voltage offsets that affect every sensor and actuator on that ground circuit. The fault appears intermittently because the resistance changes with temperature and moisture. Always clean and tighten ground connections as part of any intermittent diagnosis, then verify ground integrity with a voltage drop test under load.

Heat-related semiconductor failures occur in sensors, modules, and actuator drivers. Transistors and integrated circuits degrade over time and begin to fail at operating temperature before failing completely at room temperature. The diagnostic signature is a fault that only appears after the vehicle has been running for fifteen to thirty minutes and clears after the engine cools down. Temp-related sensor drift falls into this category too — an IAT sensor that reads correctly at 70 degrees Fahrenheit but reads six degrees high at operating temperature causes a lean condition the driver only notices on a warm day.

Cracked solder joints inside modules are underdiagnosed because they are invisible from the outside. The module appears normal, no connector issues, no wiring problems. But inside, a solder joint on a circuit board has developed a micro-crack from thermal cycling and vibration. When the board heats up and expands, the crack opens. When it cools, it closes. The module works intermittently. The diagnostic signature is a module-level fault that responds to vibration testing with the rubber mallet and heat testing with the heat gun. The fix is module replacement.

CAN bus termination issues cause a range of intermittent communication faults across multiple modules simultaneously. The CAN bus requires 120-ohm termination resistors at each end of the bus. If a termination resistor drifts from its value or a module loses proper ground, CAN communication becomes unreliable. Multiple U-codes for communication faults across different modules, symptoms that seem unrelated, and faults that clear after a power cycle all point toward CAN bus integrity issues. Measure CAN bus resistance with the ignition off and both modules disconnected — you should read approximately 60 ohms across the two termination resistors in parallel.

Chafed wiring creates an intermittent short or open depending on how the harness moves. A wire that chafes against a bracket will short to ground only when the harness flexes in a specific direction. This is why systematic wiggle testing with harness inspection matters. Look for any location where a harness passes near a sharp edge, a moving component, or a heat source.

Connector terminal tension loss happens as connectors age and see thermal cycling. The stamped metal terminals inside a connector are designed to grip the mating terminal with spring tension. That tension relaxes over thousands of heat cycles. The result is a connection that has continuity under no-load conditions but develops resistance under the current load of an actuator. Check terminal tension on any connector in a circuit that shows intermittent faults, especially on actuator and injector circuits where current draw is significant.

The 80/20 Rule for Intermittent Diagnostics

In real shop experience, approximately 80 percent of intermittent faults trace back to wiring and connector issues. The remaining 20 percent are component failures — sensors, actuators, or modules. This ratio should drive where you invest your diagnostic time first.

Before you condemn any component on an intermittent fault, verify the wiring and connector integrity for that circuit completely. That means a voltage drop test on the supply and ground circuits, a check of terminal tension at every connector in the circuit, an inspection of the harness for chafing and damage, and a ground integrity test on any chassis grounds in the circuit path. Only after those items are verified clean should you shift focus to the component itself.

This rule also protects you from comebacks. A sensor that is reading erratically almost always has wiring upstream that is contributing to the problem. Replace the sensor without fixing the wiring and the new sensor fails prematurely or the symptom returns. The customer is back in your bay, unhappy, and you have lost the parts cost and labor time to a warranty repair.

The 20 percent of component failures are real and they do happen. But they are the minority. Do the wiring work first, every time, on every intermittent fault.

When to Keep the Car Versus Send It Back

Not every intermittent fault gets resolved in a standard diagnostic visit. Setting the right expectations with the service advisor and the customer from the first conversation determines whether this ends well for everyone or turns into a conflict.

When the vehicle comes in with an intermittent complaint and no current fault, be upfront immediately. Explain to the advisor — in plain language they can relay to the customer — that diagnosing an intermittent fault requires either reproducing the condition or capturing data during a fault event. Neither of those things may happen during a standard diagnostic appointment. This is not a failure of skill. It is the nature of the problem.

Get authorization for extended diagnostic time upfront. Some shops use a tiered authorization process for intermittent work: an initial block of time for data review and preliminary testing, then a second authorization for extended monitoring or environmental testing if the initial phase does not reproduce the fault. This protects the customer from an open-ended bill and protects the shop from doing unpaid work.

If the fault cannot be reproduced and no data captures an event, communicate that clearly and document everything you did test. List every connector inspected, every circuit tested, every mode 6 PID reviewed. That documentation serves two purposes: it shows the customer that real work was done even without a fix, and it gives the next technician a starting point rather than starting over from zero.

Sometimes the right call is to send the car back with a data logger installed and instructions for the customer to return when the fault occurs again. Give them a specific return trigger — if the check engine light comes on, come in immediately without clearing it. If the symptom happens, pull over safely and note the exact conditions. The more information they capture in real time, the faster the diagnosis moves when the car comes back.

Keep the car when you have a specific test in progress — environmental testing underway, a data logger actively recording, or a known suspect that requires extended monitoring. Do not keep the car just because you have not figured it out yet. That burns goodwill without producing results.

Intermittent diagnostics rewards patience, process, and documentation. The technicians who are best at this work are not necessarily the ones with the most technical knowledge. They are the ones who build a systematic approach, follow it every time, and do not skip steps because a step seems unlikely. The fault is there. The evidence exists. Your job is to find the conditions that make it visible.

Quick Reference: Intermittent Diagnostic Sequence

1. Interview the customer thoroughly — conditions, frequency, duration, patterns
2. Pull all stored codes and read freeze frame data for each code
3. Check Mode 6 data for monitor results trending toward failure thresholds
4. Inspect and test all ground connections in the affected circuit with a voltage drop test
5. Perform systematic wiggle testing with live data or scope monitoring
6. Perform environmental testing — heat, cold spray, water, vibration — as appropriate for the symptom pattern
7. Set up data logging or scope recording and attempt to drive the fault into an event
8. If no event captured in shop, install a data logger for customer drive cycle
9. Document every test performed and every location inspected
10. Communicate status and next steps clearly to the advisor and customer

Intermittent faults are the most demanding work in automotive diagnostics, but they are also the most satisfying to solve. Every piece of information is a clue. Every condition that triggers the fault is pointing you toward the root cause. Build the process, trust the data, and follow the evidence. That is how you find problems that come and go.