SMPS Repair Guide: Where to Start When a Switch Mode Power Supply Fails
A working order for dead, clicking or unstable switch mode power supplies: stay safe, discharge the bulk capacitor, check parts cold, then power up through a dim-bulb limiter.
1. Interactive Symptom-to-Cause Diagnoser
Select the exact symptom and observable bench behaviors to generate a ranked checklist of likely component failures in diagnostic order.
SMPS Repair Guide: Where to Start When a Switch Mode Power Supply Fails
LETHAL VOLTAGE WARNING: READ BEFORE TOUCHING ANY SCREW
The primary side of a switch mode power supply is connected directly to the mains, with no isolation from the grid. Rectified mains gives a DC bus of about 170V on 120V grids and 325V on 230V grids, and up to 400V DC where active Power Factor Correction (PFC) is fitted. Voltages like these can kill. Check your local electrical rules and do not work alone.
- The bulk capacitor can stay charged long after unplugging: In a dead supply the bleeder resistors often fail open, and a high-voltage capacitor can then hold close to full voltage for hours or days. Measure it every time.
- Never discharge a capacitor with a screwdriver: A dead short makes a bright arc that can burn you, throw molten metal and damage the capacitor. Use a power resistor instead.
- Keep one hand away from the chassis during any live check: A current path from one hand to the other passes across the chest. Keeping a hand clear of the board lowers that risk.
- Primary-side heatsinks are often live: The tabs of the primary MOSFET and the bridge rectifier can sit at mains-referenced voltage. Do not assume any heatsink is grounded.
- Do not clip an earthed scope probe to the primary side: The ground clip of a mains-powered oscilloscope is tied to earth through its power cord. On a non-isolated primary, that clip can create a short to the live side and blow up the probe or the supply. Use an isolation transformer or an isolated differential probe, or skip scope work on the primary.
- Use an isolation transformer and a dim-bulb tester for any powered test: If mains must be applied, feed the unit from a 1:1 isolation transformer through a series incandescent lamp. The transformer lowers the shock risk but does not make the primary safe to touch.
Figure 1: A resistor on insulated leads across the capacitor, then a multimeter check before anything is touched.
2. How an SMPS Works in Plain English
A linear supply uses a heavy 50/60 Hz iron transformer to step the mains down, then wastes the excess voltage as heat in a pass transistor. It is quiet but inefficient, often 50% or less. A switch mode supply works differently. It rectifies the mains straight to high-voltage DC, chops that into pulses at roughly 50 kHz to 500 kHz, passes them through a small ferrite transformer, and rectifies and smooths the secondary into clean DC.
A transformer needs far less core at 100 kHz than at 50 Hz, so the heavy iron block shrinks to something the size of a walnut. The primary MOSFET is also either fully on (almost no resistance) or fully off (no current), so little energy is lost as heat. Typical efficiency is 85% to 95%.
Figure 2: Block diagram of a flyback SMPS with an optocoupler carrying the feedback across the isolation barrier.
The optocoupler is there for safety. Output terminals that people can touch must be isolated from the mains, as standards such as IEC 62368-1 require, so no copper can cross from secondary to primary. To regulate the output, an LED inside the optocoupler sends light across the gap to a phototransistor on the primary side, which tells the controller to adjust its duty cycle.
3. The Correct Diagnostic Order: A Systematic Field Method
Most wasted repair time comes from two habits: desoldering parts at random, or fitting a new fuse and switching on. A steadier approach takes six steps.
- Stage 1: Ask what happened. Did it die during a storm, after a drop, under heavy load or in standby? Was there a bang, smoke or a burning smell? A failure after a surge points at the input MOV, fuse and bridge rectifier. A unit that squealed and restarted more and more over a few months points at dried-out secondary electrolytic capacitors.
-
Stage 2: Look closely, with magnification.
Inspect both sides of the board in good light. Look for:
- Electrolytic capacitors with bulged, domed or vented tops.
- Brown crust or white residue around capacitor bases.
- Cracked, burnt or missing resistors, especially the low-ohm current-sense resistor at the MOSFET source.
- Craters or cracks in the controller IC package.
- Ring-shaped cracks in solder joints around the transformer, inductors and TO-220 heatsink pins.
- Stage 3: Discharge, then cold meter checks. Confirm the bulk capacitor reads below 12V DC. Check the mains fuse with continuity. If it is open, do not fit a new one yet: a fuse does not blow without a reason. With the meter on a resistance range, a reading near 0 Ω across the bulk capacitor means a short on the DC rail. Then test the primary MOSFET drain to source and each diode in the bridge rectifier.
- Stage 4: Work through the parts cold. Go in signal order: primary switch, snubber diode, startup resistors (these can drift up to megohms and starve the controller), the controller's VCC capacitor, secondary Schottky diodes, and the output capacitors.
- Stage 5: First power-up through a dim-bulb limiter. Once the shorts are cleared and the damaged parts replaced, connect the unit to an isolation transformer with a 60W incandescent lamp in series. A healthy supply gives a brief flash as the bulk capacitor charges, then the bulb dims to dark or a faint glow. If it stays at full brightness, switch off at once: a short is still there on the primary side.
- Stage 6: Test under real load. A supply that shows 12V with nothing connected can still collapse at 2 A. Use an electronic load or a suitable resistor or lamp load, and check both the output voltage and the ripple before you return the unit to service.
4. Component Test Reference Table
Use this table as a quick reference when probing parts with a multimeter. Other components in parallel can distort in-circuit readings, so if a result looks wrong, lift one leg or remove the part.
| Component | DMM Mode | Healthy Bench Reading | Defective Reading | Bench Caveat |
|---|---|---|---|---|
| Input Fuse | Continuity (Ω) | 0.1 Ω – 0.5 Ω (Continuous) | O.L (Open Circuit) | Blackened glass suggests a violent primary short. |
| NTC Inrush Limiter | Resistance (Ω) | Typically 2.5 Ω – 15 Ω at 25°C (varies by part) | O.L (Blown open) or cracked | Resistance falls as the part heats in operation. |
| Bridge Rectifier | Diode Test (V) | 0.50V – 0.65V forward, O.L reverse | 0.00V (Shorted) or O.L both ways | Test all four internal diodes between the AC and DC pins. |
| Primary MOSFET | Diode Test / Ω | Drain-Source: 0.5V diode body drop, O.L reverse; Gate: O.L | Drain-to-Source shorted (0.00V) or Gate punctured | A shorted MOSFET often takes the gate resistor and controller IC with it. |
| Secondary Schottky | Diode Test (V) | 0.20V – 0.40V forward, O.L reverse | 0.00V (Short circuit both ways) | Secondary Schottky diodes are among the most common parts to fail shorted. |
| Startup Resistors | Resistance (Ω) | 100 kΩ – 470 kΩ (matches color code) | Megohms high or O.L | These resistors can drift high or go open, so the IC never gets startup current. |
| Electrolytic Cap | ESR Meter | About 0.03 Ω – 0.20 Ω (low-ESR type, depends on value and voltage) | > 2.0 Ω or high capacitance loss | A capacitance reading alone can miss a dried-out part; check ESR. |
| Optocoupler | Diode Test / Test Rig | LED side: ~1.15V forward; Transistor: O.L open until LED is lit | LED open or transistor shorted | A faulty optocoupler can cause output runaway or an instant OVP shutdown. |
5. Symptom-to-Cause Troubleshooting Guide
Most failures fall into a few recognisable symptoms. Here is what each one usually means.
Symptom A: Completely Dead (Zero Output, Fuse Remains Intact)
No output, no standby light and an intact fuse means power is reaching the bulk capacitor but the oscillator is not starting.
- Likely cause 1: open startup resistors. The controller needs a trickle of current from the DC rail to charge its VCC capacitor before the transformer auxiliary winding takes over. The high-value startup resistors often go open.
- Likely cause 2: a tired VCC capacitor. A small 10 µF to 47 µF capacitor holds up the controller supply. If its ESR is high, the chip cannot keep running after the startup burst and shuts down.
- Avoid: poking controller pins with the unit powered. Meter leads add stray capacitance to sensitive feedback and timing pins and can upset the switching enough to kill the MOSFET.
Symptom B: Blows the Mains Fuse Immediately Upon AC Plug-In
A fuse that blows the moment you plug in points to a dead short across the rectified DC rail, before switching has even started.
- Likely cause 1: shorted primary MOSFET. Voltage spikes or heat stress break down the silicon and short drain to source.
- Likely cause 2: shorted bridge rectifier diode. One diode failing short puts the AC input across the bulk capacitor.
- Likely cause 3: shorted MOV, snubber diode or bulk capacitor. A varistor that took a surge, or a failed snubber diode or capacitor, can short the rail.
- Avoid: a bigger fuse, a wire link or foil. Without a working fuse, a fault can overheat the board, burn traces and start a fire.
Symptom C: Clicks, Chirps, or "Hiccups" Continuously
A rhythmic click, often one to five times a second, means the controller is alive and trying to start but trips its over-current or short-circuit protection each time.
- Likely cause 1: shorted secondary Schottky diode. The secondary winding is effectively shorted, so each MOSFET pulse drives a large current through the sense resistor and the controller shuts down.
- Likely cause 2: shorted output capacitor. The electrolytic has broken down internally.
- Likely cause 3: open feedback optocoupler. With no regulation signal, the controller assumes the output is running away and cycles into protection.
- Avoid: leaving the unit clicking for long. The repeated start attempts heat the startup resistor and snubber parts.
Symptom D: Output Voltage Steady but Significantly Low
For example, a 12V rail that sits at 7V, or a 24V rail at 14V:
- Likely cause: the feedback reference (TL431) or its divider resistors. If the lower divider resistor drifts high, or the upper one drops, the sense voltage reads too high and the controller backs off the duty cycle, so the output settles low.
- Also check: output capacitors with high ESR, which let the output sag under even a light load.
6. Why Electrolytic Capacitors Fail So Frequently (ESR & Heat)
Aluminum electrolytic capacitors are among the most common wear-out parts in a power supply. Inside is a roll of aluminum foil and paper soaked in liquid electrolyte, closed with a rubber seal. Ripple current flowing through the capacitor's internal resistance heats it (I²R loss).
A widely used rule of thumb, based on Arrhenius behavior, is that every 10°C rise in temperature roughly halves a capacitor's life. A sealed adapter with no airflow can run its capacitors very hot, and the electrolyte slowly escapes through the seal.
As the electrolyte dries out, Equivalent Series Resistance (ESR) climbs. A healthy low-ESR 1000 µF 16V capacitor might measure around 0.03 Ω. A dried-out one can reach several ohms while a basic meter still reads, say, 850 µF. The capacitor then stops filtering and just gets hot. A capacitor does not have to bulge or leak to be dead, so check ESR.
7. Repair or Replace? An Honest Bench Checklist
Repair saves waste and can be satisfying, but some boards are not worth the time or the risk. Use this filter before you sink hours into one.
✓ Strong Candidates for Repair
- Quality branded ATX computer supplies (Gold or Platinum rated, worth over $100).
- Discontinued or proprietary industrial power boards with long replacement lead times.
- Supplies inside high-end audio amplifiers and studio monitors.
- Television power boards that mostly use standard through-hole parts.
- Faults limited to the secondary side, such as bad capacitors or a shorted Schottky diode.
✗ Scrap / Replace Immediately
- Low-power USB wall chargers (under 20W) that cost a few dollars to replace.
- Ultrasonically welded or potted supplies that cannot be resealed safely.
- Boards with scorched, carbonized laminate between high-voltage traces (carbon conducts and can cause further faults or fire).
- Supplies with shorted transformer windings, since custom transformers are rarely sold separately.
- Medical or other safety-critical equipment.
8. Seven Common Beginner Mistakes
- 1. Replacing only the blown fuse. The fuse did its job; something is shorted. A new fuse in the same fault usually blows again, and can take more parts with it.
- 2. Swapping the MOSFET without checking its gate drive. When a MOSFET fails, the high drain voltage often reaches the gate resistor and controller IC. Fit a new MOSFET without checking those, and a damaged driver can destroy it at the first power-up.
- 3. Shorting a capacitor with a screwdriver. Besides the arc and the molten metal, the surge can damage the capacitor internally.
- 4. Powering up straight from the wall. A repaired supply should first run through a series incandescent lamp or a variable transformer. Switching on at full mains is how repairs turn into smoke.
- 5. Fitting general-purpose capacitors where low-ESR types belong. A standard 85°C part in a high-frequency filter position will run hot and fail early. Use 105°C low-ESR capacitors.
- 6. Replacing X and Y capacitors with ordinary ones. Safety capacitors are designed so that a failure does not leave a shock or fire hazard. An ordinary capacitor can fail short and put mains voltage where it should not be.
- 7. Clipping a scope ground to the primary side. Unless the scope or probe is isolated, this can short the primary to earth, destroy the probe and trip the breaker.
9. Tools Worth Owning for Power Supply Repair
You do not need expensive lab gear to diagnose most SMPS faults. These four tools cover most of it.
10. Frequently Asked Questions (FAQ)
Is it safe to repair a switch mode power supply at home?
Why does my power supply blow the fuse instantly on plug-in?
How do I discharge a bulk capacitor safely?
What is a dim-bulb tester?
Why does my supply click or chirp continuously without output voltage?
Can I replace a blown MOSFET with one that has higher voltage and current ratings?
Why do electrolytic capacitors fail without bulging or leaking?
What are safety X and Y capacitors?
11. Key Takeaways for the Bench
- Discharge first, then measure. Put a 10 kΩ 10 W resistor across the bulk capacitor and confirm below 12V DC before you touch anything.
- Find the cause of a blown fuse. It usually means a shorted MOSFET, bridge diode or similar part. Locate it before fitting a new fuse.
- Power up through a dim-bulb limiter. A series 60W filament lamp limits fault current and often saves the new parts.
- Check capacitors for ESR, not just capacitance. High ESR from dried-out electrolyte is a very common cause of instability and clicking.
- Respect the isolation barrier. Keep creepage gaps intact and use only safety-rated X and Y capacitors.
12. Sources and Standards
The safety ideas in this guide follow the thinking behind these standards. They are listed for reference, not as a compliance guide.
- IEC 62368-1: Audio/video, information and communication technology equipment – Part 1: Safety requirements (Galvanic isolation, creepage, and clearance).
- IEC 61010-1: Safety requirements for electrical equipment for measurement, control, and laboratory use (Measurement Category ratings and probe safety).
- IEC 60384-14: Fixed capacitors for use in electronic equipment – Part 14: Safety suppression capacitors (Classes X and Y).
- IEC 60065: Audio, video and similar electronic apparatus – Safety requirements.
Standards are cited by title for general reference. Electrical codes and safety rules vary by country. Work on mains-powered equipment carries real risk, so follow your local rules and know your limits.