Diagnosing Regulator and Converter Faults

Section Overview

A rail can be present, enabled, and sequenced correctly and still be wrong, because the part that makes it — a linear regulator or a switching converter — has its own ways to fail, and this section diagnoses those parts directly (load-and-regulation-testing). A regulator holds its output against a fixed internal target. A reference voltage is the stable internal voltage a regulator compares its output against, so the regulated output is that reference multiplied by the feedback divider ratio, and a reference that drifts drags the output with it. That comparison runs in a control loop. A feedback loop senses the output, divides it down, compares it to the reference, and drives the pass element or the switching to correct any difference, so a regulated output is only as good as its feedback — and an open or wrong divider can rail the output high or hold it at the wrong value. A switching converter adds a node that tells its whole story. A switching node is where the switch, the inductor, and the catch diode meet and the voltage swings rapidly between two rails as the converter switches — for the common buck converter, between the input and ground — so scoping it shows at a glance whether the converter is switching at all, and at what duty cycle (measuring-rail-voltage-ripple-and-noise). Linear and switching parts fail in knowable ways. A regulator can be dead, hold the wrong output, oscillate, or shut itself down on over-current or over-temperature, and each failure has a signature the reference, the feedback, and the switching node reveal (load-and-regulation-testing). Read the part that makes the rail, and a rail that is present but wrong gives up its cause.

Why This Matters

When a rail is wrong rather than absent, the sequencing and enable logic is not the problem — the regulator or converter itself is — and diagnosing that part directly is what separates a wrong rail from a replaced-guesswork part (load-and-regulation-testing). This matters because the output has a knowable source: a regulated output is its reference times a divider ratio, so a wrong-but-steady output points at the reference or the divider, not the whole regulator, and you check those before condemning the part. This matters because the feedback pin is a window: a regulating loop holds the feedback pin at the reference, so reading that pin tells whether the loop is closed, and a feedback pin off the reference locates the fault in the sensing path (measuring-rail-voltage-ripple-and-noise). It matters because a switcher is read at one node: scoping the switching node shows whether a converter is switching, dead, or shut down in a single reading, turning a dead rail into a located converter fault. It matters because failures have signatures: dead, wrong, oscillating, and shut-down are distinct patterns with distinct causes, so recognising which one you have points straight at the reference, feedback, switching, or protection. And it matters because guessing is expensive: replacing a converter chip that was only shut down by a downstream short, or a regulator that only lacked its reference, wastes the part and leaves the fault — reading the cause first does not (load-and-regulation-testing). Read the regulator to its cause, and a rail that is present but wrong stops being a mystery.

Required Prerequisites

• Measuring Rail Voltage, Ripple, and Noise — Section 5.2 taught reading a rail's voltage and its ripple and noise on a scope; this section reads those same qualities back to the regulator or converter that produced them.
• Load and Regulation Testing — Section 5.3 loaded a rail and watched how it held voltage, meeting dropout, foldback, and current limit; this section diagnoses the regulator behind those behaviours.

Recommended Consumables

• A schematic of the regulator and its feedback — to see the reference, the feedback divider, and the pass or switching element (load-and-regulation-testing)
• An oscilloscope with a fast timebase — to read the switching node and the output ripple
• Fine scope and meter probes — to reach the feedback pin and small regulator pins
• A notebook for the readings — to log the output, feedback, reference, and switching-node state
• A datasheet or known-good board — to compare the reference, duty cycle, and pin voltages against (measuring-rail-voltage-ripple-and-noise)

Recommended Practice Hardware

• A board with a linear regulator or LDO — to read a reference-and-feedback regulator directly (load-and-regulation-testing)
• A board with a switching converter — to scope a live switching node and its duty cycle
• A regulator with a wrong or dead output — to trace the fault to reference, feedback, or pass element
• A converter that hiccups or shuts down — to see a protection-driven failure (measuring-rail-voltage-ripple-and-noise)
• A known-good identical board — to compare pin voltages, duty cycle, and ripple against
• A thermal camera or careful touch — to find a regulator running hot or in thermal shutdown

Real-World Applications

Reading the regulator is how a present-but-wrong rail is finally explained. A technician with a rail sitting at the wrong voltage reads the feedback pin, finds it off the reference, and traces an open feedback divider rather than replacing the regulator (measuring-rail-voltage-ripple-and-noise). A repairer facing a dead switching rail scopes the switching node, sees it stuck and not switching, and knows the converter is dead or shut down rather than merely loaded. Someone chasing a noisy, unstable rail sees the output oscillating and the switching node hunting, and suspects a marginal feedback loop or a failed output capacitor, not a random fault (load-and-regulation-testing). A technician whose converter runs hot then drops out recognises a thermal or over-current shutdown and looks for what is overloading it before blaming the chip. And a repairer with an LDO that will not hold its output finds the input barely above the output and diagnoses dropout, not a failed regulator (load-and-regulation-testing). The failures this prevents: replacing a good regulator that only lacked its reference or divider, condemning a converter that a downstream short had merely shut down, and missing dropout or a protection trip behind a rail that is wrong rather than failed.

Common Challenges

• A regulator can be the victim of its load. A part pulled down by a downstream short looks failed — telling an innocent regulator from a truly failed one is the core challenge (measuring-rail-voltage-ripple-and-noise).
• A switching converter hides its state at the rail. A dead, hunting, or hiccupping converter can look similar at the output — its true state shows clearly only at the switching node.
• Protection mimics failure. A part shut down on over-current or heat looks dead but is working as designed — recognising a protection trip rather than a fault is easy to miss (load-and-regulation-testing).

Safety Notes

Risk Level: Medium. Reading regulators and converters is done on a powered board with a meter and scope, and switching nodes carry fast, high-swing edges, so this is live work and this section is Medium risk.

Professional Tips Before Starting

• Read the feedback pin. A regulating loop holds it at the reference — a feedback pin off the reference locates a wrong output (measuring-rail-voltage-ripple-and-noise).
• Scope the switching node. It tells you in a glance whether a converter switches — stuck means dead or shut down, switching means look at duty and load.
• Suspect protection. A converter can shut itself down on a downstream fault — check for over-current or thermal shutdown before condemning it (load-and-regulation-testing).

Diagnosing Regulators and Converters — From Symptom to Cause

Recap and Frame

Section 5.3 loaded a rail and watched how it held voltage; this section goes inside the part that holds it, and the frame is that every regulated rail is set by a reference and a feedback loop, so a rail that is wrong rather than absent is diagnosed by reading that loop back to its fault (load-and-regulation-testing). A rail is made by a part with a knowable structure. Every rail comes from a linear regulator or a switching converter, and both hold their output by sensing it and correcting toward a fixed internal target, so understanding that structure is what makes the faults readable (load-and-regulation-testing). The output has a source you can point to. The regulated output is its reference multiplied by a divider ratio, so a wrong output has a small set of causes — a drifting reference, a wrong or open divider, or a loop that cannot hold — and each is found by reading a specific point. A switcher shows its health at one node. Scoping where the switch and inductor meet tells you whether a converter is switching, dead, or shut down before you measure anything else (measuring-rail-voltage-ripple-and-noise). The failures are few and distinct. Dead, wrong, oscillating, and shut-down are the recurring patterns, each with its own cause and its own reading, so naming the pattern narrows the search. And protection hides as failure. A part that has shut itself down on over-current or heat looks dead but is not faulty, so recognising a protection trip stops a needless replacement. Hold the frame — a rail is made by a reference and a feedback loop, and each failure has a signature — and a present-but-wrong rail becomes diagnosable.

The Regulator's Job and How It Fails

Before the specific parts, the shared idea: a regulator senses its output, compares it to a fixed internal target, and corrects any difference, and every failure is a break somewhere in that chain (load-and-regulation-testing). Know the shared structure. A linear regulator burns off the difference across a pass element to hold its output, and a switching converter chops its input and filters the result, but both close a loop that compares the output to a reference and drive their element to correct it, so the loop is the common thread. Name the failure modes. A regulator can be dead with no output, hold a steady but wrong output, oscillate instead of settling, or shut itself down on over-current or over-temperature, and these four patterns cover most of what you meet. See that each has a place to look. Dead points at input, enable, or a killed part; wrong points at the reference or divider; oscillating points at the loop or output capacitor; shut down points at protection and its trigger — so the pattern chooses the target. Read the symptom on the rail first. The rail measured under section 5.2's methods — its voltage, ripple, and noise — already narrows which pattern you have before you open the regulator up (measuring-rail-voltage-ripple-and-noise). Keep the load in mind. A regulator can be innocent and only reacting to its load — a short pulling it down, an overload tripping it — so the part and its load are read together, not the part alone. The shared structure, the four failure modes, where each is read, the symptom on the rail, and the load's role understood — and the regulator's job and its failures are framed. Grasp the loop, and every regulator fault is a break you can place in it.

Reading the Reference and Feedback — Where the Output Is Set

The output voltage is not arbitrary — it is set by a reference and a feedback divider — so a wrong output is diagnosed by reading those two, not by condemning the whole part (load-and-regulation-testing). Understand the reference. The reference voltage is the stable internal target the regulator compares its output against, so the regulated output equals that reference times the feedback divider ratio, and a reference that has drifted or collapsed drags the output with it. Understand the feedback loop. The feedback loop divides the output down, compares the divided value to the reference, and drives the pass element or the switching to correct any difference, so when the loop regulates, the feedback pin sits right at the reference value. Read the feedback pin against the reference. Measuring the feedback pin and comparing it to the known reference tells whether the loop is closed — a feedback pin at the reference with a wrong output means the divider ratio is wrong, and a feedback pin off the reference means the loop is not regulating (measuring-rail-voltage-ripple-and-noise). Suspect the divider for a wrong-but-steady output. An open upper divider resistor makes the regulator drive the output high to try to reach the reference, and a wrong or shorted resistor sets the output to the wrong value, so a steady wrong output points hard at the divider. Check the reference itself. Where a regulator brings its reference out on a pin it is read directly, and where it does not — most parts — the feedback pin read against the datasheet's reference value stands in for it, so confirming the reference is present and correct separates a failed reference from a good one with a bad divider. Tie it to regulation. A loop that cannot hold the feedback pin at the reference under load is the same fault that section 5.3 saw as poor regulation, connecting the pin reading to the load behaviour (load-and-regulation-testing). The reference understood, the feedback loop understood, the pin read against the reference, the divider suspected, the reference checked, and regulation tied in — and where the output is set is readable. Read the feedback against the reference, and a wrong output tells you whether the divider, the reference, or the loop is at fault.

Diagnosing a Linear Regulator or LDO

The linear regulator is the simpler part — a reference, a divider, and a pass element in series — so its faults map cleanly onto those three plus its dropout (load-and-regulation-testing). Check the input first. A linear regulator can only make an output below its input, so confirm the input is present and high enough before suspecting the regulator, since a missing or low input explains a dead or low output with nothing wrong in the part. Watch for dropout. An LDO or linear regulator whose input has fallen too close to its output can no longer hold regulation and its output sags with the input — this is dropout, the dropout voltage behaviour from section 5.3, not a failed part (load-and-regulation-testing). Read output, feedback, and reference. With the input good, read the output, then the feedback pin against the reference — a wrong output with the feedback at the reference is a divider fault, and a feedback off the reference is a loop or reference fault. Suspect the pass element for a dead or shorted output. A linear regulator's pass transistor can fail open, giving no output, or fail short, passing the input straight through — a short reads as the output sitting near the input, a dangerous over-voltage on the rail. Feel for heat and shutdown. A linear regulator dropping a large difference at high current runs hot and can enter thermal shutdown, cycling off and on, so a rail that comes and goes with a hot regulator points at thermal protection and an overload. Mind the load. A linear regulator pulled down may be innocent, with a downstream short loading it, so a dead-low output is checked for a short on the rail before the regulator is blamed (measuring-rail-voltage-ripple-and-noise). The input checked, dropout watched, output-feedback-reference read, the pass element suspected, heat and shutdown felt, and the load minded — and a linear regulator is diagnosed. Read a linear regulator as input, reference, divider, and pass element, and its fault is one of a short list.

Diagnosing a Switching Converter — The Switching Node

A switching converter is read differently, because its heart is a node that swings as it switches, and scoping that switching node tells you more in one look than any static reading (measuring-rail-voltage-ripple-and-noise). Find and scope the switching node. The switching node is where the switch, the inductor, and the catch diode or synchronous rectifier meet, and a healthy converter shows a clean, roughly square swing there at its switching frequency, so putting a scope on it is the first move. Read whether it switches at all. A switching node stuck high, stuck low, or flat is a converter that is not switching — dead or shut down — while a clean swing means the converter is running and the fault lies elsewhere, so this one reading splits the diagnosis. Read the duty cycle and frequency. A converter switching at the wrong duty cycle for its input and output, or hunting in frequency, points at the feedback loop or the reference driving it wrongly, connecting the node back to the loop. Check the inductor and rectifier path. A switching node that rings hard, shows odd steps, or will not swing cleanly can indicate an open or saturated inductor, a failed catch diode or synchronous rectifier, or a shorted switch, so the power path is read from the node. Read the output ripple with the node. The output ripple measured under section 5.2 together with the switching-node waveform tells whether the converter is filtering properly, so a rippling output with a switching node hunting suggests a control or output-capacitor problem (measuring-rail-voltage-ripple-and-noise). Watch for hiccup and shutdown. A converter that switches in short bursts then stops, repeating, is in hiccup-mode protection, usually from an over-current or a downstream short, so a hiccupping switching node points at an overload, not a dead chip. The node found and scoped, switching-or-not read, duty and frequency read, the inductor path checked, ripple read with the node, and hiccup watched — and a switching converter is diagnosed. Scope the switching node, and a switching converter tells you whether it is dead, wrong, or merely protecting itself.

Reading the Fault: Dead, Wrong, Oscillating, or Shut Down

The four failure patterns pull together here, each with the reading that names it and the cause it points to, so any regulator fault sorts into one (load-and-regulation-testing). Read a dead output. No output with a good input and asserted enable means a killed part, an open pass element, or a converter not switching — read the switching node or the pass element to place it, and check the load for a short dragging it to zero (measuring-rail-voltage-ripple-and-noise). Read a wrong-but-steady output. A steady output at the wrong voltage is a reference or divider fault — read the feedback pin against the reference to tell a wrong divider from a lost reference — and is rarely the pass or switching element itself. Read an oscillating output. An output that rings, hunts, or will not settle is a stability fault — a marginal or wrongly compensated feedback loop, a failed or missing output capacitor, or a converter hunting — seen as excess ripple and a restless switching node (measuring-rail-voltage-ripple-and-noise). Read a shut-down part. A regulator that is off, cycles on and off, or hiccups is in protection — over-current, thermal, or under-voltage lockout (the part refusing to run below a minimum input) — so find what triggers it, usually an overload or a downstream short, before replacing a part that is working as designed. Separate the part from its load. Across all four, a regulator can be the victim of its load, so a fault is confirmed on the part by removing or isolating the load where possible, telling a failed regulator from an overloaded one (load-and-regulation-testing). Trace the confirmed fault to its origin. Once the pattern and the part are known, the fault is traced to its true cause — the shorted load, the open divider, the failed capacitor, the killed switch — so the repair addresses the origin, not the symptom. Dead, wrong, oscillating, and shut-down read and placed, the part separated from its load, and the fault traced to origin — and the regulator fault is fully diagnosed. Name the pattern, read the point it chooses, and a regulator or converter fault resolves to a cause.

Common Mistakes

• Condemning the regulator for a wrong output. A steady wrong output is usually the reference or divider — read the feedback pin against the reference first (measuring-rail-voltage-ripple-and-noise).
• Not scoping the switching node. A dead switching rail says little at the output — scope the switching node to see if it switches at all.
• Missing dropout on a linear regulator. An input too close to the output cannot regulate — check the input margin before blaming the part (load-and-regulation-testing).
• Replacing a shut-down converter. A converter hiccupping on a downstream short is protecting, not failed — find the overload before replacing it.
• Reading the part without its load. A regulator pulled down may be innocent — isolate the load to tell a failed part from an overloaded one (load-and-regulation-testing).

Troubleshooting Guidance

Regulator and converter faults come down to reference, feedback, switching, or protection. If a rail is at the wrong voltage but steady: read the feedback pin against the reference — at the reference means a divider fault, off it means a loop or reference fault (measuring-rail-voltage-ripple-and-noise). If a switching rail is dead: scope the switching node — stuck means dead or shut down, a clean swing means look at the output stage and load. If a linear rail is low and sagging: check the input margin for dropout before suspecting the regulator (load-and-regulation-testing). If the output oscillates or is noisy: suspect a marginal feedback loop or a failed output capacitor, and read the ripple with the switching node. If a converter switches in bursts then stops: it is in hiccup protection — find the over-current or downstream short triggering it. If a regulator runs hot then cycles: it is in thermal shutdown from an overload — find what it is driving, not the chip alone (load-and-regulation-testing). If the output sits near the input: a linear pass element may be shorted, passing the input through — a dangerous over-voltage — so remove power and confirm. The throughline: name the failure pattern, read the point it chooses — reference, feedback, switching node, or protection — and separate the part from its load.

Verification & Testing Methods

Confirm you diagnosed the regulator to its cause:

• [ ] I read the output and then the feedback pin against the reference voltage, telling a wrong divider from a lost reference or an open loop (measuring-rail-voltage-ripple-and-noise).
• [ ] I confirmed a wrong-but-steady output traced through the feedback loop and its divider rather than the pass or switching element.
• [ ] I scoped the switching node of a switching converter and read whether it was switching, dead, or hiccupping, and at what duty cycle.
• [ ] I checked a linear regulator's input margin for dropout and felt for a thermal shutdown before condemning the part (load-and-regulation-testing).
• [ ] I separated the regulator from its load to tell a failed part from an overloaded one, and traced the confirmed fault to its origin.

Then try the practice exercises below — regulator and converter diagnosis on powered boards; scenarios differ from the quiz.

Practice Exercises

1. Read reference and feedback (5 minutes, hands-on). On a board with a wrong-voltage rail, read the output and the feedback pin, compare the feedback to the reference, and decide whether the divider, the reference, or the loop is at fault (measuring-rail-voltage-ripple-and-noise).
2. Scope a switching node (5 minutes, hands-on). Find a switching converter's switching node, scope it, and read whether it is switching cleanly, its rough duty cycle, and whether it is stuck or hiccupping.
3. Diagnose a linear regulator (5 minutes, hands-on). For a linear regulator or LDO with a low output, check the input margin for dropout, then read output, feedback, and heat to place the fault (load-and-regulation-testing).
4. Name the failure pattern (5 minutes, reasoning). For a described faulty rail, decide whether it is dead, wrong-but-steady, oscillating, or shut down, and state which point — reference, feedback, switching node, or protection — you would read to confirm it.

These core steps — the regulator's structure, reading reference and feedback, diagnosing a linear regulator, reading a switching converter at its node, and naming the failure pattern — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

• A regulated output equals its reference voltage times a feedback divider ratio, so a wrong-but-steady output points at the reference or the divider, not the whole regulator (load-and-regulation-testing).
• A feedback loop senses the output, compares it to the reference, and corrects any difference, so a regulating loop holds the feedback pin at the reference — and a feedback pin off the reference locates a wrong output in the sensing path (measuring-rail-voltage-ripple-and-noise).
• A switching converter is read at its switching node, where a clean swing means it is switching and a stuck node means it is dead or shut down, so one reading splits the diagnosis.
• A linear regulator's faults map onto input, dropout, reference, divider, and pass element, so checking the input margin and the feedback before the part avoids replacing a good regulator (load-and-regulation-testing).
• Regulator faults sort into dead, wrong, oscillating, or shut down, each with its own reading and cause — and a regulator can be the innocent victim of its load, so the part is always separated from its load before it is condemned.

Skills Learned

• You can now explain how a regulator holds its output with a reference and a feedback loop.
• You can now read the reference and feedback pins to locate a wrong output.
• You can now diagnose a linear regulator or LDO that is dead, wrong, or in dropout.
• You can now read a switching converter at its switching node to see if it is switching.
• You can now recognise a regulator that is oscillating or shut down on protection.

Glossary Additions

• reference voltage — the stable internal voltage a regulator compares its output against to set and hold that output. A regulator's feedback loop divides the output down and compares it to the reference, adjusting until the divided output equals the reference, so the regulated output is the reference multiplied by the feedback divider's gain (the output is stepped down to the reference, so the output is that many times larger than the reference). This makes the reference the anchor of the whole regulated value: if the reference drifts or collapses, the output drifts with it, so a regulator whose output is wrong but steady is checked at its reference and feedback divider before the pass element is blamed. Many regulators do not bring the reference out on a pin, but the feedback pin sits at the reference voltage whenever the loop is regulating, so reading the feedback pin against the known reference tells whether the loop is closed and where a wrong output originates.
• feedback loop — the control path by which a regulator senses its own output and adjusts to hold it at the target: the output is divided down, compared to the reference voltage, and the error drives the pass element or the switching to correct any difference, closing the loop around the output. The feedback loop is what makes a regulated output regulated, so many regulator faults are feedback faults — an open feedback divider makes the regulator drive the output high toward the reference, a shorted or wrong divider sets the output to the wrong value, and a marginal or wrongly compensated loop oscillates instead of settling. Reading the feedback pin against the reference — it should sit at the reference when regulating — tells whether the loop is closed and locates a wrong or unstable output in the sensing path rather than the pass element.
• switching node — the node in a switching converter where the switch, the inductor, and the catch diode or synchronous rectifier meet, and where the voltage swings rapidly between two rails as the converter switches — for the common buck (step-down) converter, between the input rail and ground, while a boost swings between ground and its higher output. The switching node is the single most diagnostic point on a switching converter: a healthy converter shows a clean, roughly square waveform there at the switching frequency, so scoping the switching node immediately tells whether the converter is switching at all, at what frequency and duty cycle, and whether its edges are clean. A switching node stuck high, stuck low, or flat means the converter is dead or shut down, while a node that switches but at the wrong duty cycle, rings badly, or hiccups points at the feedback, the inductor path, or the load — so the switching node turns a dead or wrong rail into a located converter fault.

Suggested Next Sections

Must read next:

• Tracing a Rail Fault to Its Cause — Section 5.6 brings the whole chapter together into one method: from a symptom on a rail, through distribution, sequencing, regulation, and the regulator itself, to the single component or node at the root — closing the power-rail analysis chapter.

Recommended:

• Measuring Rail Voltage, Ripple, and Noise — reading a rail's voltage, ripple, and noise, the symptoms that point back to the regulator or converter.
• Load and Regulation Testing — loading a rail and reading dropout, foldback, and current limit, the behaviours a faulty regulator shows under load.