Starts Cold, Not Hot: Intermittent Module Faults

The symptom you are actually chasing

It goes like this. The car starts every morning without hesitation. You drive twenty minutes, stop for coffee, come back out, and it cranks and cranks and will not fire. You wait forty minutes. It starts. You drive home. The next morning it is perfect again.

Or the mirror image: the dash lights up like a Christmas tree on a cold start, the ABS and traction lights come on, and everything clears once the car has been running for ten minutes. Or the instrument cluster goes blank on a hot afternoon and comes back the next day. Or the key fob stops being recognised in August and works fine in October.

All of these are the same class of problem: a temperature-dependent electrical fault. The circuit works at one temperature and does not work at another. And the single most important thing to understand about this failure class is what it does to the diagnostic process — the vehicle arrives at the shop cold, having sat on a flatbed or cooled off in the parking lot, and every test passes. The scan tool is happy. The module answers. Nothing is wrong.

Nothing is wrong right now. That is not the same thing as nothing being wrong.

Why a module behaves differently hot than cold

There is real physics underneath this, and understanding it changes how you test.

A control module is a printed circuit board — usually fiberglass-epoxy laminate — carrying components made of ceramic, silicon, plastic, and metal, joined by solder. Every one of those materials expands at a different rate when it heats up. The board grows in one direction, the component body grows in another, and the solder joint between them absorbs the difference. Every heat-up and cool-down cycle works that joint a little.

Underhood and in-cabin electronics are specified for brutal conditions. Automotive electronic components are commonly qualified across a range of roughly minus 40 to 125 degrees Celsius, a specification range documented across the standards work of SAE International and the electronics reliability literature published through IEEE. Materials research at the National Institute of Standards and Technology on solder-joint reliability has long characterised thermal cycling as a primary driver of fatigue cracking in electronic assemblies — joints fail not from one hot day, but from thousands of cycles of expansion and contraction.

On top of the mechanical story, semiconductors themselves change behaviour with temperature. Leakage current rises. A voltage regulator that is already marginal drops its output below the threshold the microcontroller needs. An electrolytic capacitor with degraded electrolyte shows a different equivalent series resistance hot than cold. A crystal oscillator drifts. Any of these can move a circuit from "just barely working" to "not working" over a twenty-degree change.

So when a customer says "it only does it when it is hot," they are not being vague. They are handing you the most useful diagnostic clue in the entire case.

The four internal failure modes that behave exactly this way

When the fault really is inside the module, it is almost always one of four things.

1. Cracked solder joints and cold joints. The classic. A joint that was slightly under-heated at manufacture, or one that has been thermally cycled for a decade, develops a hairline fracture. Cold, the two faces sit in contact and the circuit conducts. Hot, thermal expansion opens the crack by a few microns and the circuit goes open. This is why the fault appears at temperature and vanishes overnight.

2. Fatigued BGA and reflow connections. Ball-grid-array packages hide their solder balls underneath the chip where nothing can be inspected visually. Microprocessors, memory, and gateway chips are commonly BGA. A fatigued ball behaves exactly like a cracked joint but cannot be found with a magnifier — it needs a thermal test, X-ray, or a rework station.

3. Corroded vias and lifted traces from moisture ingress. Water gets into a module through a failed seal, a leaking cowl, a plugged sunroof drain, or a rodent-opened grommet. Corrosion eats through the plated barrel of a via — the tiny metal tube that carries a signal from one layer of the board to another. A partially corroded via has resistance that changes with humidity and temperature. This one often correlates with rain more than with heat.

4. Marginal power supplies. A voltage regulator or a filter capacitor that has aged out will hold regulation at 20 degrees and fall over at 70. The processor browns out, resets, or simply stops driving an output. Power-supply faults tend to produce whole-module symptoms — the module drops off the bus entirely rather than losing one function.

If you want the deeper version of what is and is not fixable at that level, our guide to board-level ECU repair and what is actually repairable covers the rework side in detail.

Before you blame the module — and this is most cases

Here is the part that costs people money when they skip it.

Most intermittent electrical faults are not module faults. They are batteries, charging systems, grounds, connectors, and harnesses. A module is a sealed box that sits still; a connector pin is a spring-loaded contact that gets vibrated, heated, and corroded every day of its life. Statistically, the moving, exposed, corrodible things fail more.

Vehicles in service are old enough now that this matters more than it used to. The average age of a light vehicle on U.S. roads has climbed to roughly 12 years according to industry tracking summarised by outlets including J.D. Power and Consumer Reports, and the typical vehicle covers something on the order of 13,000 miles a year of vibration and thermal cycling per data reflected in vehicle-history reporting from Carfax. Twelve years of that is a lot of cycles for a crimp, a ground strap, or a battery terminal.

Work the cheap causes first, in this order:

• Battery and charging. A battery that load-tests weak, or a charging system that sags under electrical load, produces bizarre multi-module symptoms that look exactly like a failed control unit. Heat also hurts batteries — high underhood temperatures accelerate internal degradation, which is precisely why so many "only when hot" complaints turn out to be a battery on its last summer.
• Grounds. A corroded or loose ground point raises the reference voltage for everything attached to it. Do voltage-drop testing, not resistance testing, and do it with the circuit loaded. Our write-up on charging system and ground faults that cause repeat module failure exists because we see modules destroyed twice by the same untouched ground.
• Connectors. Fretting corrosion, spread terminals, backed-out pins, water in a connector body. Pull the connector, inspect the pin faces under light, and check terminal tension with a matching test pin — never by feel.
• Harness. Chafe points where a loom crosses a bracket, rodent damage, and repairs done with crimp caps and tape. If something has been chewed, rodent damage versus module failure walks through telling them apart.

If a network communication code is present, resist the urge to buy the module the code names. Lost-communication codes usually indict the network, not the node — the reasoning is laid out in our guide to U0100 and U0140 lost-communication codes.

Symptom, likely cause, and what to do next

Symptom pattern	Most likely cause	Next diagnostic step
Cranks but no start only after a heat soak, fine cold	Marginal ignition/immobilizer circuit, weak battery, or a heat-affected joint	Recreate the heat soak, then measure cranking voltage and check for security-system indication before condemning anything
Multiple unrelated warning lights on cold start, clear when warm	Low battery state of charge or a poor ground reference	Load-test the battery, then voltage-drop test every ground on the affected circuits
One module drops off the bus at temperature, others fine	Cracked joint or failing power supply inside that module	Data-log during the fault, then freeze-spray that module to see if it comes back
Dash or cluster goes dark intermittently, returns after sitting	Fatigued solder on the cluster board or its connector	Wiggle test the cluster connector under power before removing anything
Key fob or push-button start unreliable in hot weather	Receiver module, antenna, or a marginal keyless control unit	Test with a second known-good key, then check the receiver, not the fob
Fault follows rain and humidity rather than heat	Moisture ingress and corroded vias in a module or connector	Find and fix the water path first, then evaluate the board
Fault follows bumps and rough roads, not temperature	Connector, harness chafe, or a cracked ground	Wiggle test with the circuit loaded and monitor for dropout

The diagnostic sequence you can actually perform

You do not need a lab to make real progress. You need discipline and a notebook.

Step 1 — Document the pattern. Every time it happens, write down: outside temperature, how long the engine had run, whether it had been sitting hot, whether it was raining, and exactly what failed. Three or four entries usually reveal whether you are chasing heat, vibration, or humidity. This log is the most valuable thing you can bring to a technician.

Step 2 — Battery and charging, properly. Load-test the battery. Measure charging voltage at idle with accessories on; you should see roughly 14 volts, not 12.4. Check the battery terminals and the main ground strap for corrosion. Do not skip this because the battery is new — new batteries fail too.

Step 3 — Voltage-drop test the grounds. With the circuit energised, measure between the module ground pin and battery negative. More than about a tenth of a volt is a problem worth chasing. Resistance readings on an unloaded circuit lie; voltage drop under load does not.

Step 4 — Scan for history codes and freeze-frame data. Even if nothing is present now, stored and history codes often carry the coolant temperature or ambient temperature recorded when the fault set. If every stored event has a high engine temperature stamped on it, you have just proven the thermal hypothesis. If the vehicle is a no-start with a security indication, our immobilizer green key light no-start guide covers how to read those specific patterns before assuming the engine controller is at fault.

Step 5 — Data-log during the fault. This is the single highest-value step. Leave a scan tool or logger connected, drive the car until it fails, and capture the moment. Watching a module stop responding in real time is worth more than a hundred cold tests. Note which parameters go away and which survive.

Step 6 — Wiggle test under load. With the system powered and, ideally, a live data display running, gently move connectors and harness sections one at a time. If a wiggle drops the circuit, you have found the fault and it is not the module.

Step 7 — Localise with temperature. Covered in detail next.

Freeze spray and heat gun: localising a thermal fault

This is the technique that separates guessing from proving. It works because you can create the failing temperature on demand, one small area at a time.

Heat first, to make it fail. With the module powered and its behaviour being monitored, warm one section of the board with a heat gun on a low setting, held well back, moving constantly. Use an infrared thermometer and keep the surface below roughly 80 degrees Celsius — you are reproducing an in-service temperature, not reflowing anything. Mask adjacent areas with card so you know which region you actually heated. When the circuit misbehaves as one specific area comes up to temperature, that area is your suspect.

Then cold, to make it work again. Circuit chiller or freeze spray on the suspect area, in short bursts, through a straw or nozzle so it hits a small target. If the function returns as that spot cools, the diagnosis is essentially made.

Three warnings. First, do not blast a very hot board with freeze spray — the thermal shock can crack ceramic components and create a second fault. Let it come down first, then chill the target. Second, freeze spray on a live board condenses moisture; dry it before drawing conclusions about a humidity-sensitive fault. Third, keep the heat gun away from plastic connector bodies and electrolytic capacitors; both are damaged well below the temperature at which solder moves.

Done carefully, this method routinely finds a single suspect square centimetre on a board in twenty minutes — a fault that a scan tool cannot see at all, ever, because the module reports itself as fully functional right up until the joint opens.

Why "it fixed itself overnight" is a symptom, not a cure

Customers apologise for this constantly. They call, book an appointment, and by the next morning the car is fine, so they cancel.

They should not. Self-resolution overnight is a positive finding. A fault that clears after a cool-down and returns after a heat soak is close to diagnostic proof of a thermal mechanism. Random faults that come and go with no pattern are much harder; a fault with a clean thermal correlation has already told you half of what you need to know.

The same applies to the "I disconnected the battery and it fixed it" story. That usually means a module latched into a fault state and a power cycle cleared it, not that the underlying weakness went away. It will come back, and it will come back when the conditions repeat — which is to say, on the next hot day.

The ones that scare me are not the cars that are broken when they arrive. Those I can fix. The ones that eat a week are the ones that arrive working, and the only honest thing you can do is stop guessing, reproduce the temperature, and let the board tell you where it hurts. — Independent automotive electronics technician, 18+ years (anonymized)

The security and comfort variants of the same problem

Not every intermittent fault is a no-start.

Immobilizer and key recognition. A security system that intermittently refuses to authorise start can be a heat-affected immobilizer control unit, a marginal antenna ring around the ignition lock, a failing transponder, or a poor connection at the engine controller. The order of investigation matters here because the components have very different costs. The immobilizer green key light no-start walkthrough covers the pattern-reading in detail, and it applies well beyond one brand. Note that any key, immobilizer, or security work — on the bench or in the car — requires proof of ownership, and there are no exceptions to that.

Keyless entry and remote start. Intermittent fob range that tracks with weather is usually the receiver module or its antenna, not the fob. A keyless entry module repair is a $125 bench job when the receiver is genuinely the fault; it is money wasted if the real problem is a broken antenna wire, so test with a second known-good key first.

Body and comfort modules on European platforms. The BMW footwell module is the textbook example of a body module whose failure produces a shifting collection of lighting and comfort symptoms rather than one clean fault; a BMW FRM footwell module repair runs $175 on the bench. On Jaguar Land Rover platforms, keyless and RF authorisation problems often route through the KVM or RFA unit, and our KVM and RFA virginize walkthrough explains what that module actually does before you spend money replacing one.

Instrument clusters. Clusters are the most common single victim of solder fatigue, because they combine a big board, a hot windscreen-facing location, and a heavy connector. An instrument cluster repair with mileage sync is $200. An ABS module repair is $499 when the internal fault is confirmed — and confirming it first is exactly the point of this article.

What a bench evaluation actually does that a scan tool cannot

A scan tool asks a module whether it is working. A cracked solder joint does not know it is cracked.

On the bench, the module is powered through a purpose-built harness, exercised through its real inputs and outputs, and then deliberately heat-soaked while its behaviour is monitored. That is the difference. We can hold a module at temperature for as long as it takes, chill sections of the board, put a scope on individual pins, and inspect joints under a microscope — none of which is possible with the module bolted behind a dash in a car that will not misbehave on command.

A bench evaluation is $150 and produces a straight answer: the fault is reproducible and repairable, the fault is reproducible and not economically repairable, or the module tests clean at temperature and the problem is in the vehicle. That third answer is not a failure of the evaluation. It is worth $150 to stop replacing modules that are fine. If you want more detail on the method, see what a bench evaluation finds when scan tools cannot.

Realistic cost ranges

Numbers, honestly stated, from cheapest to most expensive.

• Battery, terminals, ground repair: typically the least expensive fix on the list, and it resolves a large share of intermittent complaints. Parts pricing guidance from outlets such as Kelley Blue Book puts routine battery replacement well under the cost of any module work.
• Connector or harness repair: usually a labour job. A single pin replacement or a repaired chafe point is far cheaper than any module, which is the whole reason to test before you buy.
• Bench evaluation: $150, and it either finds the fault or clears the module.
• Module repair on the bench: $125 for a keyless entry module, $175 for a BMW FRM, $200 for a cluster repair with mileage sync, $250 for an ABS module — flat rates, not estimates.
• Replacement module from a dealer: frequently several times the repair figure once programming is included, and on VIN-locked units it is not a plug-and-play swap. Coverage of repair-versus-replace economics from outlets including Car and Driver and MotorTrend consistently finds the electronics replacement path the expensive one.

Return shipping is chosen at checkout and starts at $24.95, with faster tiers available. A technician verifies fitment before anything ships back.

When to stop and send it to a bench

Ship the module when you can say most of these:

• The battery and charging system have been tested good, not assumed good.
• Grounds have been voltage-drop tested under load, not ohm-checked.
• Connectors have been inspected and terminal tension checked.
• A wiggle test does not reproduce the fault.
• The fault correlates cleanly with temperature, and you have log entries proving it.
• The fault follows the module — or heating that specific module reproduces it.

At that point, further roadside guessing costs more than the evaluation. If you are unsure which unit to send, our guide on how to know which module failed before you ship it is written for exactly that decision, and how to remove, package, and ship a car module covers doing it without adding shipping damage to the original fault. Modules ship to PO Box 120241, Arlington, TX 76012 by USPS, or to 1009 Oakwood Ln # 120241, Arlington, TX 76012 by UPS or FedEx, and the work is performed at the workshop on West Pioneer Parkway in Arlington.

Frequently asked questions

Why does my car start cold but not when it is hot? The most common causes are a weak battery, a poor ground connection, or a cracked solder joint inside a control module that opens as the board expands with heat. All three behave identically from the driver's seat, so the diagnosis has to be done in cost order: battery and charging first, grounds second, connectors third, and the module last.

Can a scan tool find an intermittent module fault? Usually not on its own. A scan tool asks a module to report its status, and a module with a cracked solder joint reports itself as healthy right up to the moment the joint opens. What a scan tool can do is capture history codes and freeze-frame data with temperature stamps, and data-log the moment the fault occurs if you leave it connected while driving.

Is it worth spending money on diagnosis if the car is working today? Yes, and a fault that clears overnight is actually good news diagnostically. Self-resolution after a cool-down is strong evidence of a thermal mechanism, which narrows the search dramatically. Cancelling the appointment because the symptom went away simply guarantees the same fault returns on the next hot day.

How do I tell a connector problem from a module problem? Wiggle-test the connector and harness with the circuit powered and live data displayed. If moving the connector drops the circuit, the fault is in the connection. If heating the module body reproduces the fault while the harness stays undisturbed, the fault is on the board. Doing both tests before buying parts is what saves the money.

Can a cracked solder joint be repaired, or does the module need replacing? Many can be repaired. A cracked joint on an accessible component is straightforward rework for a technician with the right station, and repairing the original module avoids VIN-locking and programming problems entirely. Fatigued ball-grid-array joints under a processor and heavily corroded multi-layer boards are the cases where replacement is honestly the better call.

Do I need proof of ownership to have a module worked on? For any key, immobilizer, or security-related module, yes — proof of ownership is required with no exceptions, because that work touches the vehicle's anti-theft system. Non-security repairs such as a cluster, ABS unit, or footwell module do not carry that requirement, though a technician still verifies fitment against the vehicle before returning the unit.

Should I just replace the module to rule it out? That is usually the most expensive way to test a theory. Many modern modules are VIN-locked, so a replacement is not a plug-and-play swap and needs programming before it will function at all, and if the real fault was a ground or a connector the new module fails the same way. Testing first is cheaper than guessing twice.

The bottom line

A car that starts cold and fails hot is not being mysterious — it is telling you the fault is temperature-dependent, which is more information than most complaints ever provide. The failure mechanisms inside a module are well understood: cracked solder joints, fatigued BGA connections, corroded vias, and marginal power supplies, all of which behave exactly this way and none of which a scan tool can see.

But the honest order of investigation puts the module last. Batteries, grounds, connectors, and harnesses cause more intermittent faults than boards do, they are cheaper to fix, and skipping them is how people end up with a new module and the same symptom. Test the cheap things properly, log the pattern, and use heat and cold to localise.

When the evidence genuinely points at the board — the fault tracks temperature, the vehicle-side tests are clean, and heating that specific module reproduces it — a bench evaluation at $150 will reproduce the fault under controlled conditions and give you a straight answer instead of another guess. Text us the vehicle details and what you have already tested, and a technician will confirm fitment and the right next step before you ship anything.