Repair Worked Examples

This lesson is the payoff for everything in this module. Rather than introducing a new technique, it puts every technique to work together on four complete, realistic repairs — an audio amplifier with a dead channel, a bench power supply with a thermal fault, a powered speaker with worsening hum, and a microcontroller project that resets at random. Each example starts the way a real repair starts: with a vague complaint and no idea what is wrong. By the end of each, you will see the systematic method, divide and conquer, signal injection or tracing, substitution, in-circuit measurement and scope work combining into one confident diagnosis.

Diagnosing a silent amplifier channel: a large DC offset at the output keeps the protection relay open, and checking the parts around the shorted output transistor finds all the damage before power is reapplied.

View Larger

Example 1: Amplifier Channel Silent, Protection Relay Never Clicks

Initial complaint: A two-channel 100 W audio amplifier powers on, but the speaker protection relay never clicks in and there is no sound from either speaker. The owner says it stopped working while playing loudly at a party.

Tools used: Multimeter, oscilloscope, function generator, 8 Ω power load resistors, current-limited bench supply or a mains lamp limiter.

Diagnosis

Step 1 (systematic method): Gather information. A fault that appeared suddenly at high volume suggests an overstress event in the power stage rather than gradual aging. The relay staying open is itself a clue: in most amplifiers, the protection circuit holds the relay open if it detects DC at an output, to protect the speakers.

Step 2 (predict, then measure): With the speakers disconnected and the amplifier powered through a mains lamp limiter, the DC voltage at each channel output (before the relay) is measured. A healthy output sits within about ±50 mV of zero. The left channel reads 8 mV; the right channel reads −41 V — almost the full negative rail.

Step 3 (divide and conquer): An output pinned at a supply rail means either the output stage is shorted, or the driver stages are commanding full output. With the power off and the capacitors discharged, the right channel output transistors are measured with the meter’s diode test: the lower output transistor reads a near short between collector and emitter.

Step 4 (in-circuit measurement): Before replacing parts, the low-value emitter resistors and the driver transistors of that channel are checked: one 0.22 Ω emitter resistor has gone open, and one driver transistor is also shorted — collateral damage from the same event.

Step 5 (substitution): The failed output transistor is replaced together with its partner (output pairs should be replaced as matched sets), along with the open resistor and the shorted driver.

Root cause: Overdriving the amplifier into a low impedance at high volume overloaded the output stage, shorting a transistor and damaging the parts around it. The protection relay did exactly its job: it saw a large DC offset and kept it away from the speakers.

Repair: Replace the output pair, the emitter resistor and the driver transistor; check the bias network; set the idle current to the manufacturer’s value.

Verification: Powered up through the lamp limiter, the right output sits at 12 mV DC and the relay clicks in after its delay. On a load resistor with a 1 kHz tone, both channels deliver full power cleanly, confirmed on the scope, and the idle current is stable after an hour.

Lessons learned: The relay behavior pointed straight at a DC offset; one voltage measurement per channel found the faulty side; and checking the parts around a failed transistor, before powering up again, prevented the new parts from being destroyed by the same fault.

Example 2: Bench Power Supply Shuts Down After Twenty Minutes

Initial complaint: A 0–30 V, 5 A switching bench power supply works normally at first, but when it supplies 3 A or more for about twenty minutes, the output drops to zero and the display shows an over-temperature warning. After cooling for a while it works again.

Tools used: Multimeter, thermocouple or infrared thermometer, electronic or resistor load, a spare fan.

Diagnosis

Step 1 (systematic method): Gather information. The fault depends on load and time and recovers after cooling — the classic signature of a thermal problem. The over-temperature warning shows the supply’s own protection is doing its job, so the question is why it is getting hot.

Step 2 (most-likely-cause-first): In equipment with forced-air cooling, the fan and the airflow path fail far more often than the power semiconductors. With the supply running at 3 A, the fan is observed: it turns, but slowly, and the air from the vent is barely moving. Dust is visible in the inlet grille.

Step 3 (in-circuit measurement): The voltage across the fan is measured at 12 V, which is correct, so the fan’s drive circuit is healthy and the fan itself is suspect. The heat sink temperature, measured with a thermocouple, reaches 85°C after fifteen minutes — well above its normal operating range.

Step 4 (substitution): A known-good fan of the same size, voltage and airflow rating is connected temporarily in place of the original. With the same 3 A load, the heat sink stabilizes at 52°C and the supply runs for an hour without shutting down.

Root cause: The fan’s bearing had worn and the inlet was clogged with dust, so the fan was turning at a fraction of its rated speed. The supply’s thermal protection correctly shut it down before the semiconductors were damaged.

Repair: Replace the fan with one of equal or greater airflow, and clean the inlet grille and heat sink fins.

Verification: Running at 5 A for two hours, the heat sink stabilizes at 58°C and the supply does not shut down.

Lessons learned: A load- and time-dependent fault that recovers after cooling is almost always thermal; checking the cheap, mechanical cooling parts first saved time; and a temporary substitution proved the diagnosis before any part was ordered.

Example 3: Powered Speaker Hum Getting Steadily Worse

Initial complaint: A ten-year-old powered speaker has developed a hum that is audible even with the volume at zero and no input connected. The owner says the hum has grown louder over several months.

Tools used: Oscilloscope, multimeter, ESR meter, isolation transformer.

Diagnosis

Step 1 (systematic method): Gather information. Hum with no input connected rules out a ground loop with other equipment, because there is no other equipment connected. A gradual increase over months points to a component aging slowly rather than a sudden failure.

Step 2 (scope): The amplifier’s main supply rails are checked with a scope (powered through an isolation transformer). The rails carry a sawtooth ripple of 3.5 V peak-to-peak at 120 Hz — twice the mains frequency, so the rectifier is working on both half-cycles, but the reservoir capacitors are not smoothing the output nearly as well as they should.

Step 3 (in-circuit and component measurement): With the power off and discharged, the two large reservoir capacitors are checked with an ESR meter. Both read several ohms instead of the expected few tens of milliohms, and one has a slightly domed top. They sit right beside the amplifier’s heat sink.

Step 4 (substitution): Both capacitors are replaced with new ones of the same capacitance and voltage and a 105°C rating.

Root cause: Years of heat from the nearby heat sink had dried out the electrolyte in the reservoir capacitors, raising their ESR and reducing their effective capacitance, so ripple on the supply rails grew until it was audible as hum.

Repair: Replace both reservoir capacitors with 105°C, long-life types; check the smaller filter capacitors in the same area at the same time.

Verification: Ripple on the rails falls to 0.3 V peak-to-peak under normal listening levels and the hum is no longer audible with the volume up and no input.

Lessons learned: Hum with nothing connected pointed inside the unit; a scope on the supply rails identified ripple at twice the mains frequency; and an ESR meter confirmed aged capacitors that a simple capacitance check might have missed.

Example 4: Microcontroller Project Resets When the Relay Switches

Initial complaint: A greenhouse controller built around a microcontroller board reboots at random. The owner has noticed that resets seem to happen when the irrigation pump relay switches.

Tools used: Oscilloscope, multimeter.

Diagnosis

Step 1 (systematic method): Gather information. Resets associated with a load switching suggest a power or noise problem rather than a software bug. The relay coil and the microcontroller share a 5 V supply, and the pump is powered through the relay contacts from a separate 12 V supply.

Step 2 (scope): The scope is connected to the microcontroller’s 5 V pin and set to trigger on a dip below 4.5 V. Each time the relay turns off, the trace shows a sharp spike and then a dip to about 3.9 V lasting a few microseconds — below the brown-out threshold of the microcontroller.

Step 3 (in-circuit inspection): The relay driver transistor has no flyback diode across the relay coil. When the transistor switches off, the coil’s stored energy produces a large voltage spike, and the coil current and the pump wiring share a thin ground wire with the microcontroller board.

Step 4 (repair and test): A 1N4148 diode is fitted across the relay coil (cathode to the positive supply), a 470 µF capacitor is added on the 5 V rail near the relay, and the relay and pump grounds are rewired to return directly to the supply rather than through the microcontroller board.

Root cause: The missing flyback diode let the relay coil generate a large spike on turn-off, and the shared ground wiring coupled that disturbance into the microcontroller’s supply, triggering brown-out resets.

Repair: Fit a flyback diode across the relay coil, add local bulk capacitance, and separate the high-current return paths from the logic ground.

Verification: With the scope still connected and triggering on the same threshold, the relay is cycled a hundred times without a single trigger event, and the controller runs for a week without resetting.

Lessons learned: Linking the symptom to an event (the relay switching) led straight to a power problem; a triggered scope caught a microsecond dip that no meter could show; and the cure combined a component fix with a wiring change.

Lessons Learned Across All Four Examples

Two common threads run through these examples. First, the power supply — its rails, its cooling, and its wiring — was involved in three of the four faults, which is why checking supplies early is such a reliable habit. Second, in every case the decisive step was a measurement chosen to test a specific hypothesis, not a part replaced in hope. The protective features built into the equipment — the speaker relay, the thermal shutdown and the brown-out detector — were also doing their jobs in three of the four cases; understanding what a protection circuit is reacting to is often the fastest route to the real fault.