Finding the Cold Spot — Stages That Should Be Warm

Section Overview

Not every thermal fault is a part too hot — sometimes the tell is a stage that should be warm and is stone cold, doing none of the work it should, and reading that is the other half of the thermal picture (heat-as-a-diagnostic-signal). Working stages give off heat by doing their job. Self-heating is the warmth a working component produces by dissipating power as it operates, so a regulator, a driver, or a busy processor runs above ambient — and that expected warmth is part of the board's normal signature. A part cold where warmth is expected is a fault. A cold spot is a component or region reading colder than its signature says it should, so a stage at ambient where it should self-heat is doing no work, the mirror image of a hot spot (finding-the-overheating-component). A cold stage is often a casualty, not the culprit. A dead stage is a block drawing little current and producing no self-heating, and its cause is often upstream — a supply rail that never arrived or an enable that was never asserted — so the cold points upstream as much as at itself (power-sequencing-and-enable-logic). The cold is confirmed against the electrical picture. A cold stage is checked for its supply and enable — good supply and enable means a failed part, a missing rail or enable means a casualty of an upstream fault — so the meter resolves the cold into a cause (heat-as-a-diagnostic-signal). Read where a board is cold as well as where it is hot, and both halves of its thermal map lead to the fault.

Why This Matters

A thermal camera is usually pointed at what is too hot, but a stage that should be warm and is cold is just as much a fault, so reading cold completes the diagnostic picture (heat-as-a-diagnostic-signal). This matters because absence of heat is a signal: a working stage self-heats, so a cold one is doing no work, which a hot-part-only search would walk straight past. This matters because cold points upstream: a dead stage often has no fault of its own — its rail never came or its enable never asserted — so the cold leads to the real fault above it (power-sequencing-and-enable-logic). It matters because cold tells failed from unpowered: a cold stage with its supply and enable good is a failed part, while a cold stage with a missing rail is a casualty, so the cold plus the meter separates the two. It matters because a whole cold region has one cause: an area of a board gone cold points at a shared rail or enable that failed, so the region is read like several dead stages with one upstream cause (understanding-power-rails-and-distribution). And it matters because cold confirms a suspicion: a stage suspected dead for other reasons is corroborated by its coldness, so absence of heat backs up a diagnosis rather than standing alone. Read the cold as well as the hot, and the stage that never came alive gives itself away.

Required Prerequisites

• Heat as a Diagnostic Signal — Section 6.1 established reading heat and the normal signature; this section reads the absence of expected heat, the cold spot, as the mirror image of the hot one.
• Power Sequencing and Enable Logic — Section 5.4 explained why a rail or stage may never come up; a cold stage is often exactly that, so the cold points back to the missing rail or un-asserted enable.

Recommended Consumables

• A thermal camera with a tuned span — to find the region cooler than its signature expects (heat-as-a-diagnostic-signal)
• A schematic of the board's stages — to know which blocks should draw power and self-heat
• A notebook for the pattern — to log which stages are warm, which are cold, and where
• A known-good board or its signature — to compare which stages should run warm (power-sequencing-and-enable-logic)
• Fine meter probes — to check the cold stage's supply and enable

Recommended Practice Hardware

• A board with a dead stage — to find the cold spot where warmth should be (heat-as-a-diagnostic-signal)
• A board with a missing rail — to see a whole region gone cold from one upstream cause (power-sequencing-and-enable-logic)
• A thermal camera and a known-good board — to compare the warm-and-cold pattern against normal
• A board with a failed part on a good rail — to tell a failed stage from an unpowered one
• A bench supply with current limit — to power a suspect board safely while reading its heat
• A schematic and a multimeter — to check the supply and enable of a cold stage

Real-World Applications

Reading cold is how a technician finds the stage that quietly never came alive. A repairer with a dead audio output feels the amplifier stone cold where it should be warm and traces a missing supply rather than a blown chip (power-sequencing-and-enable-logic). A technician with a whole dark region sees an area of the board gone cold on the camera and finds one rail that never came up feeding all of it (understanding-power-rails-and-distribution). Someone unsure whether a chip failed or was starved reads its supply and enable, finds them absent, and follows the cold upstream instead of replacing the chip. A repairer confirming a suspected dead stage backs the diagnosis with its coldness, sure it is not working before opening it up. And a technician on a board that half-works maps which stages self-heat and which stay cold, splitting the working half from the dead (heat-as-a-diagnostic-signal). The failures this prevents: replacing a cold chip that was only starved of power, missing a dead stage that a hot-part search ignored, and treating a whole cold region as many faults instead of one missing rail.

Common Challenges

• Low-power parts run barely warm when healthy. A modern efficient stage may self-heat only slightly, so cold and working can look alike — judging against its own signature, not an assumed rise, tells them apart (heat-as-a-diagnostic-signal).
• Idle-cold and dead-cold can look identical. A healthy stage idle at the moment reads as cold as a dead one, and the bench may not let you drive it into the activity that would warm it — so a cold reading alone cannot always settle whether a stage is merely idle or truly dead (power-sequencing-and-enable-logic).
• Thermal mass hides a recent change. A part with heft takes time to cool or warm, so its temperature lags what it is doing now — allowing for that lag avoids misreading a stage mid-transition.

Safety Notes

Risk Level: Medium. Finding a cold spot still means a powered board — the cold part is read among live and possibly hot neighbours — so this is live work and this section is Medium risk.

Professional Tips Before Starting

• Know what should be warm. A cold spot only means something against expected warmth — learn which stages self-heat before hunting the cold (heat-as-a-diagnostic-signal).
• Read cold while it should work. An idle stage is cold and fine — power and exercise the board so a working stage would be warm.
• Follow cold upstream. A dead stage is often starved, not failed — check its supply and enable before replacing it (power-sequencing-and-enable-logic).

Finding and Reading the Cold Spot

Recap and Frame

Sections 6.1 to 6.3 read heat and hunted the hottest part; this section reads the absence of heat, and the frame is that a stage which should self-heat and does not is doing no work (heat-as-a-diagnostic-signal). Working stages self-heat. A regulator, a driver, a busy processor dissipates power as it works and so runs above ambient, so its warmth is expected and part of the board's normal signature (finding-the-overheating-component). Cold where warmth is expected is the fault. A part at ambient where its signature says it should be warm is a cold spot, the mirror of a hot spot, marking a stage doing none of its work. A cold stage often points upstream. A dead stage frequently has no fault of its own — its supply never arrived or its enable never asserted — so the cold leads above it to the real fault (power-sequencing-and-enable-logic). Failed and unpowered look alike cold. A cold stage can be a failed part or a starved one, so reading its supply and enable is what tells the two apart. A whole cold region shares a cause. An area gone cold points at one rail or enable feeding all of it, so a region is read as several dead stages with a single upstream cause (understanding-power-rails-and-distribution). Hold the frame — working stages self-heat, cold where warmth is expected is a fault, and the cold points upstream — and the stage that never came alive is found.

Know What Should Be Warm

Reading a cold spot depends entirely on knowing what should be warm, because cold means nothing without an expectation of heat to measure it against (heat-as-a-diagnostic-signal). Learn the self-heating stages. Regulators, drivers, power transistors, bridge rectifiers, and busy processors dissipate power in normal use and run warm, so these are the stages whose coldness is a strong fault signal. Know how warm each should be. A stage's normal warmth is part of the board's signature, so knowing roughly how warm a working driver or regulator should get separates a cold-and-dead one from a merely cool efficient one (finding-the-overheating-component). Mind the quiescent draw. Even an idle stage draws some quiescent current and self-heats a little, so a stage that should sit slightly warm at idle and instead reads dead cold is a signal. Read the signature on a known-good board. Powering an identical working board shows exactly which stages self-heat and how much, giving the baseline the cold is judged against, the surest reference (power-sequencing-and-enable-logic). Account for low-power parts. A modern efficient stage may self-heat only slightly even when working, so a barely-warm part is judged against its own low signature, not assumed dead for being cool. Note the whole warm-and-cold map. The signature is the pattern of which stages are warm and which cool, so a fault can be a stage cold when its like neighbour of the same kind is warm. The self-heating stages known, their normal warmth, the quiescent draw, the baseline, low-power parts, and the whole map understood — and what should be warm is established. Know which stages self-heat and how much, and a cold one stands out as doing nothing.

Scan for the Cold Spot Against the Signature

With the expectation set, the board is scanned for the region reading colder than its signature allows, the cold spot that marks a stage doing no work (heat-as-a-diagnostic-signal). Sweep for the missing warmth. A thermal camera swept over the powered board shows which stages are warm and which are not, so a stage cold where the signature expects warmth stands out as a dark patch where a bright one should be. Tune the span to reveal small warmth. Narrowing the temperature span around the board's normal temperature makes a slightly-warm working stage bright and a cold one plainly dark, so the difference between barely-warm and dead-cold is clear (finding-the-overheating-component). Compare warm neighbours of the same kind. Two like stages should self-heat alike, so one warm and one cold among a pair or bank points straight at the cold one as the fault. Read the cold against ambient. A stage at or below ambient where its signature expects a rise is dead cold, so its near-zero delta-T is the cold-spot signal, the opposite of an excess rise. Note a whole cold region. A cold area larger than one part points at a shared rail or block that failed, so the extent of the cold is read, not just a single cold part (understanding-power-rails-and-distribution). Confirm cold with a spot reading. Once the dark region is found, a spot reading pins its temperature and confirms it is at ambient rather than merely cooler, turning a dark patch into a number. The board swept, the span tuned, neighbours compared, cold read against ambient, a region noted, and the cold pinned — and the cold spot is found. Scan for missing warmth judged against the signature, and the dead stage shows as a cold spot.

Read Cold as No Power or No Activity

A cold spot means a stage is doing no work, and the reason is almost always that it has no power or no activity, so the cold is read as an absence to trace (power-sequencing-and-enable-logic). Read cold as no self-heating. A stage produces no self-heating because it is dissipating no power, and it dissipates no power because little or no current flows, so cold is fundamentally a sign of absent current (finding-the-overheating-component). Suspect a missing supply first. The commonest cause of a cold stage is that its supply rail never arrived, so a dead-cold block is checked first for whether its power is even present (understanding-power-rails-and-distribution). Suspect an un-asserted enable. A stage held off by an enable that never asserted draws no current and stays cold, so an enable read against its expected state explains a cold stage that has power but no life (power-sequencing-and-enable-logic). Suspect a missing input or clock. A stage with power and enable but no input signal or clock may do no work and stay cold, so the cold can point at a missing drive rather than a missing supply. Follow the cold upstream. Because the cause is usually above the cold stage — a missing rail, a stalled enable, a dead upstream block — the cold is traced upward to where the power or signal should have come from, not treated as a fault in the cold part. Recognise a partial cold. A stage running cooler than normal but not dead may be working at reduced power — a starved rail, a partial load — so a cold-but-not-dead reading points at a degraded rather than absent supply. Cold read as no self-heating, a missing supply, an un-asserted enable, a missing input, followed upstream, and partial cold recognised — and cold is read as an absence to trace. Read a cold stage as current that never flowed, and the trace leads up to why.

Tell a Dead Part from a Dead Rail

A cold stage can be a failed part or a good part starved of power, and telling these apart decides whether to replace the stage or trace upstream (power-sequencing-and-enable-logic). Check the supply at the cold stage. Reading whether the cold stage's supply rail is present and correct is the first split — a cold stage with no supply is starved, while a cold stage with a good supply is a stronger candidate to have failed itself (understanding-power-rails-and-distribution). Check the enable and inputs. A cold stage with power but no asserted enable, or no input signal, is held off rather than failed, so reading the enable and inputs separates a held-off stage from a dead one (power-sequencing-and-enable-logic). Read the current it draws. A failed-open part may draw no current on a good rail, while a shorted or partly working one draws some, so the current with the supply present hints at the failure mode. Trace a missing rail to its own cause. When the cold stage is starved, the missing rail is itself traced — a failed regulator, a stalled sequence, a distribution fault — so the true fault is found above the cold stage (tracing-a-rail-fault-to-its-cause). Confirm a failed part by isolation. Where the supply and enable are good and the stage is still cold and dead, isolating or substituting the part confirms it failed, so a genuine dead part is proven before replacement (finding-the-overheating-component). Mind a dependency chain. A cold stage may depend on another cold stage before it, so a chain of dead stages is followed back to the first that lost power or life, the one that matters. The supply checked, the enable and inputs checked, the current read, a missing rail traced, a failed part confirmed, and a chain followed — and a dead part is told from a dead rail. Read the supply and enable at the cold stage, and it tells you whether the part failed or was starved.

Confirm Cold Against the Electrical Picture

A cold reading is one signal, and confirming it against voltage, enable, and current is what turns a cold spot into a proven cause rather than a guess (heat-as-a-diagnostic-signal). Confirm the supply with a meter. The cold stage's supply is read with a meter to confirm it is present or absent, so the thermal cold and the electrical rail agree on whether the stage has power (understanding-power-rails-and-distribution). Confirm the enable state. The enable is read against its expected asserted level, so a cold stage held off by an un-asserted enable is confirmed electrically, not just inferred from the cold (power-sequencing-and-enable-logic). Read the current to confirm no work. Measuring the current into the cold stage confirms little or none is flowing, so the absence of self-heating is backed by an absence of current, the two pictures agreeing. Exercise the stage to confirm. Where possible, giving the stage its input or load and watching whether it then warms confirms whether it can work at all, so a stage that warms when driven was only idle, while one that stays cold is truly dead. Agree the thermal and electrical stories. A cold stage with a confirmed missing rail or un-asserted enable is a starved casualty, while a cold stage with good supply, enable, and inputs is a failed part, so the two pictures together name the cause (heat-as-a-diagnostic-signal). Trace to the true origin. Once confirmed, the cause is traced to its origin — the regulator that failed, the sequence that stalled — so the repair addresses why the stage was cold, not the cold stage alone (tracing-a-rail-fault-to-its-cause). The supply confirmed, the enable confirmed, the current read, the stage exercised, the pictures agreed, and traced to origin — and the cold is confirmed against the electrical picture. Prove the cold with the meter, and a cold stage is diagnosed as failed or starved on evidence.

Common Mistakes

• Only hunting the hot. A hot-part search walks past a dead stage — read the cold as well, for warmth missing where it belongs (heat-as-a-diagnostic-signal).
• Replacing a starved stage. A cold stage is often unpowered, not failed — check its supply and enable before replacing it (power-sequencing-and-enable-logic).
• Treating a cold region as many faults. A whole cold area shares one upstream cause — find the rail or block feeding all of it (understanding-power-rails-and-distribution).
• Calling a cold-and-idle stage dead. An idle stage is cold yet fine — read it while it should be working before judging it dead.
• Trusting cold alone. A cold reading is one signal — confirm the supply, enable, and current before condemning the stage.

Troubleshooting Guidance

Reading cold comes down to signature, cold spot, no power or activity, and confirm. If you only searched for heat: turn the hunt around and scan for a stage cold where its signature expects warmth (heat-as-a-diagnostic-signal). If a stage is cold: read it as no self-heating and check first whether its supply rail even arrived (understanding-power-rails-and-distribution). If the supply is present but the stage is cold: read the enable and inputs — a held-off or undriven stage stays cold with power (power-sequencing-and-enable-logic). If a whole region is cold: look for the one shared rail or block that failed and feeds all of it. If you cannot tell failed from starved: check the supply and enable — good means a failed part, missing means a starved casualty. If a stage is cold but you doubt it is dead: exercise it with its input or load and see whether it warms. If the cold stage is starved: trace the missing rail upstream to its own cause (tracing-a-rail-fault-to-its-cause). The throughline: know what should be warm, find where it is not, read the cold as an absence of power or activity, and confirm against the meter.

Verification & Testing Methods

Confirm you read the cold spot and traced it:

• [ ] I knew which stages should show self-heating in normal use, and judged the cold against that expected warmth (heat-as-a-diagnostic-signal).
• [ ] I scanned the board and found the cold spot — a stage at or near ambient where its signature expects a rise — pinning it with a spot reading.
• [ ] I read the cold as a dead stage doing no work, and checked first whether its supply rail even arrived (understanding-power-rails-and-distribution).
• [ ] I told a failed part from a starved one by reading the supply and enable, and traced a missing rail upstream to its own cause (power-sequencing-and-enable-logic).
• [ ] I confirmed the cold against voltage, enable, and current, and where possible exercised the stage to see whether it could warm at all.

Then try the practice exercises below — cold-spot practice on powered boards; scenarios differ from the quiz.

Practice Exercises

1. Map warm and cold (5 minutes, hands-on). Power a board and read which stages self-heat and which stay cold, building the warm-and-cold map against the known-good signature (heat-as-a-diagnostic-signal).
2. Find the cold spot (5 minutes, hands-on). On a board with a dead stage, scan with a tuned span for the stage cold where its signature expects warmth, and pin it with a spot reading (finding-the-overheating-component).
3. Failed or starved (5 minutes, hands-on). For a cold stage, read its supply and enable and decide whether it is a failed part on a good rail or a starved casualty of a missing one (power-sequencing-and-enable-logic).
4. Exercise a cold stage (2 minutes, reasoning). For a stage you suspect is only idle, state how you would give it its input or load and confirm whether it then warms.

These core steps — knowing what should be warm, finding the cold spot, reading cold as no power or activity, telling failed from starved, and confirming electrically — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

• A working stage shows self-heating — warmth from the power it dissipates — so a stage at ambient where its signature expects a rise is doing none of the work it should (heat-as-a-diagnostic-signal).
• A cold spot is the mirror of a hot spot: a component or region colder than its signature allows, marking a stage that never came alive rather than one over-dissipating (finding-the-overheating-component).
• A dead stage is often a casualty, not the culprit — its supply rail never arrived or its enable never asserted — so the cold points upstream to the real fault above it (power-sequencing-and-enable-logic).
• A cold stage is told failed from starved by reading its supply and enable: good supply and enable means a failed part, a missing rail or enable means a starved casualty to trace upstream (understanding-power-rails-and-distribution).
• A whole cold region shares one upstream cause, so it is read as several dead stages fed by one failed rail or block, not as many separate faults.

Skills Learned

• You can now identify which stages of a board should self-heat in normal use.
• You can now scan for a cold spot against the board's normal thermal signature.
• You can now read a cold stage as a sign of no power or no activity.
• You can now tell a failed part from a stage that is merely unpowered.
• You can now follow a cold spot upstream to a missing rail or a dead stage.

Glossary Additions

• cold spot — a component or region of a board that reads colder than its thermal signature says it should, revealing a stage doing none of the work it should. Where a hot spot marks a part dissipating too much power, a cold spot marks the opposite — a part or block that should be dissipating some power in normal operation and is not, because it has no supply, is not enabled, or has failed open. Reading a cold spot means knowing what should be warm: a working regulator, driver, or busy processor self-heats, so one of these sitting at ambient when its signature expects a rise is a cold spot pointing at a stage that never came alive, often with its cause upstream in a missing rail or an un-asserted enable rather than in the cold part itself.
• self-heating — the warmth a working component produces by dissipating power as it operates, which is why a healthy stage runs above ambient and a cold one stands out. A regulator, a driver, a power transistor, or a busy logic block all draw current and dissipate some of it as heat in normal use, so their self-heating is expected and forms part of the board's thermal signature. The diagnostic value of self-heating is in its absence: a stage that should self-heat and instead sits at ambient is doing no work, so no self-heating where the signature expects it is a cold-spot fault pointing at a dead or unpowered stage, just as surely as excess heat points at an over-dissipating one.
• dead stage — a functional block of a board — a regulator, an amplifier, a driver, a subsystem — that is not operating, drawing little or no current and producing no self-heating, so it reads as a cold spot. A dead stage is often not itself faulty: its supply rail may never have arrived, its enable may never have been asserted, or an upstream stage it depends on may have failed, so the cold stage points upstream as much as at itself. Telling a dead stage that is a failed part from one that is merely unpowered is the key step, done by reading whether its supply and enable are present — a cold stage with its rail and enable good is a failed part, while a cold stage with a missing rail or un-asserted enable is a casualty of an upstream fault to be traced.

Suggested Next Sections

Must read next:

• Freeze Spray and Localized Heat for Isolation — Section 6.5 makes temperature an active probe rather than a passive reading: using freeze spray to cool one part at a time and a warm-air source to heat it, provoking, finding, and confirming a fault by how it responds to deliberate cold and warmth.

Recommended:

• Heat as a Diagnostic Signal — reading heat and the normal signature that a cold spot is judged against.
• Power Sequencing and Enable Logic — the rails, enables, and power-good signals whose failure leaves a stage cold.