Repairing a power supply is not a parts-drawer exercise; it is a sequence, and the sequence starts with safety. The order of actions — discharge, isolate, limit, test — is what separates a safe repair from a dangerous one and a quick diagnosis from a parts swap that fails again. This article builds the safety-first workflow: the discharge-and-isolation sequence, the dim-bulb trick, the component test order from fuse to rectifier, the fault isolation between bridge, cap, switch, and rectifier, and the load verification that confirms the repair.
The Safety Sequence Nobody Skips Twice
The first action of a power supply repair is not diagnosis; it is safety, and the sequence exists because a power supply stores enough energy to hurt. A charged output capacitor, a live input stage, and the AC mains behind it are all real hazards, and the order of the actions is the protection.
The sequence is: discharge, isolate, and only then test. The output capacitors hold charge after the supply is unplugged, and touching them directly can deliver a dangerous shock; they must be discharged safely first. The input stage remains connected to the mains through the plug and the internals, so the work must happen with the supply isolated from the line or through a limiter. The repair bench then adds the dim-bulb trick and the dedicated test sequence, which the following sections build. The ATX supply anatomy article and the surge and ESD testing article cover the energy the stages hold; this article turns that into the safety order.
The rule that nobody skips twice is the discharge: a capacitor that looked drained can hold a second charge, and the discipline of discharging to a defined procedure — not just a glance — is the difference between a close call and a bench incident. The safety sequence is the foundation everything else stands on.
Isolation and the Dim-Bulb Trick
With the supply discharged, the next layer is isolation and current limiting, and the dim-bulb trick is the classic tool that makes first power-on safe.
The dim-bulb trick places an incandescent bulb in series with the supply’s input. When the supply is healthy and draws modest current, the bulb glows dimly and passes the power; when the supply has a fault that draws a huge current, the bulb glows bright, drops most of the voltage, and protects the internals from the fault current. The bulb is a current limiter and a fault indicator at once: a dim glow and the supply is drawing normally; a bright bulb and the supply is shorting. The power supply testing article and the SMPS components article document the isolation and the limiter practice; the dim bulb is the bench’s first-line protection.
The isolation layer also includes the isolation transformer where the supply’s live parts must be handled, separating the bench from the mains. The discipline is to never power a repaired supply directly into the mains without the limiter: the dim bulb catches the fault that a direct power-up would let through, and it makes first power-on a controlled event.
Component Test Order: From Fuse to Rectifier
The test order is the workflow’s efficiency engine: a defined sequence from the cheapest, most basic check to the deeper diagnosis, so a fault is found before the expensive parts are suspected. The order runs from the fuse to the rectifier.
The sequence starts with the fuse — the first and cheapest check, and the one that explains a dead supply with no other fault. Then the input bridge, whose rectifier diodes are checked for open or shorted arms; a failed bridge kills the whole supply regardless of the downstream health. Then the capacitors, whose state decides whether the rail holds under load. Then the switch, whose failure stops the conversion. And finally the deeper rectifier and control checks, only when the earlier stages have been cleared. The bridge testing article and the ripple diagnosis guide document the stage tests; this article orders them.
The order’s value is the elimination property: a blown fuse caught first saves the whole diagnosis; a dead bridge caught second prevents the switch and capacitor suspicion. The technician works the order, and each cleared stage narrows the remaining field, which is what makes the repair fast instead of a random swap.
The test order also has a time-saving logic that the bench internalizes with practice: each stage is cheap to check and eliminates a whole class of suspects. The fuse check costs seconds and clears the most common dead-supply cause; the bridge check, a diode-mode sweep of the four arms, clears the input stage; the capacitor check clears the rail; and only then does the diagnosis reach the switch and the control stages, where the checks are slower and the parts are expensive. Working the order backward — suspecting the switch before checking the fuse — is the classic way a repair replaces an expensive part for a fifty-cent fuse. The power supply repair article and the field reliability checklist both emphasize the elimination order, and this workflow’s value is that it makes the order explicit rather than learned by the bad repairs.
The same economics apply to the verification stage. A repaired supply that passes the no-load bench but fails under load is a repair that was never verified, and the cost of that miss is the second field failure — the supply comes back, the customer counts the downtime, and the bench loses the trust. The load verification with the dim bulb in circuit is not a luxury; it is the cheapest insurance the repair has, and it closes the loop that the safety sequence opened.
Isolating Bridge vs Cap vs Switch Faults
When the supply fails, the symptom does not always name the part, and the fault isolation between the bridge, the capacitor, and the switch is the diagnostic skill. The three faults present differently, and reading the difference isolates the cause.
A bridge fault kills the input: the supply is dead at the wall with no output, and the bridge diodes read open or shorted on the bench. A capacitor fault shows as a rail that is weak or noisy under load — the supply runs, but it cannot hold the rail, or the ripple grows past the budget. A switch fault stops the conversion: the supply outputs nothing and the switch shows no switching activity, even though the bridge and the capacitor read healthy. The bridge failure article and the ripple diagnosis guide treat the three signatures, and the rectifier failure modes guide adds the rectifier-specific patterns.
The isolation method is to test by stage and eliminate: confirm the bridge conducts, confirm the capacitor holds its charge and its ESR, confirm the switch switches, and the fault that survives the eliminations is the cause. The symptom-to-stage map turns a confusing failure into the one part worth replacing.
The fault-isolation map is compact enough for the table that a technician reads at the bench:
| Symptom | Stage most likely | Confirm with | First fix to try |
|---|
| Total dead, no output | Fuse (first), then input bridge | Fuse continuity, bridge diodes | Replace fuse, then check bridge |
|---|
| Output low or sagging | Output capacitor or 12 V rectifier | ESR, rail scope, stage heat | Replace cap, check rectifier drop |
|---|
| No switching, no conversion | Switch | Switch gate waveform | Replace the switch |
|---|
| Ripple or noise on rail | Capacitor / filter | Scope on output | Recheck cap and filter |
|---|
| Runs hot, holds rails | Thermal path / fan | Stage temperature | Fix airflow or mounting |
|---|
The table is the isolation method in one view: each symptom names the likely stage, the confirming measurement, and the first fix, which is exactly the elimination order the workflow prescribes. The bridge testing article and the ripple diagnosis guide provide the deeper tests each row points to.
Verifying the Repair Under Load
The repair is not finished when the part is replaced; it is finished when the supply runs under load and holds its rails. The load verification is the last gate, and it catches the repairs that the no-load bench cannot.
The verification runs the supply with the dim bulb still in the circuit, applies a real load, and confirms three things: the rails hold their rated voltages under the load current, the ripple stays within the budget, and the supply runs at a sensible temperature without thermal runaway. The load test also exercises the stages the no-load test never stresses — the switch under real current, the capacitors under real ripple, the rectifier under real heat — and it is the moment a marginal repair reveals itself. The field reliability checklist and the ripple diagnosis guide turn the load verification into a measurable pass-fail.
The verification closes the workflow. The power supply repair runs as a defined sequence — discharge, isolate, dim-bulb, test from fuse to rectifier, isolate the bridge-cap-switch fault, and verify under load — and the sequence is what makes the repair safe, fast, and reliable. A technician who works the order finds the real fault instead of the first suspect, and a repair that passes the load verification is a repair that holds in the field. The general rectifier category supplies the parts the order tests, and the ATX anatomy article is the board-walk companion to this workflow.