A powered monitor speaker on my desk died. Not dramatically. It just went quiet one afternoon, the little power LED still glowing as if nothing were wrong, the other speaker of the pair perfectly happy. No smoke, no smell, no bang. The most annoying kind of dead, because everything looks fine.
The honest thing to do with a fifteen-year-old active speaker is to bin it and buy another. But it was a good speaker, the pair are no longer sold, and I had a free evening and a suspicion. Most consumer electronics do not fail in clever ways. They fail at a small number of boring, physical points, and if you know where to look you can fix a surprising amount with an iron and a couple of pence of parts.
Open it up and look before you measure
The first rule is to look before you reach for the meter. Get the board out, get a bright light on it, and just read it. Eyes catch a lot that instruments miss.
This board told me most of what I needed before I touched it. Near the power section, one of the electrolytic capacitors had a domed top where it should have been flat, and the faintest crust of dried electrolyte around the vent. That is a capacitor that has cooked. They dry out and bulge with age and heat, and the power supply section near a warm amplifier is exactly where they cook fastest. A 1000 µF 25 V can, slightly domed. The classic.
While I had the light on it I went over the rest. Solder joints around the heavier components are the other usual suspect, because heat cycling and the mechanical stress of a heavy part on a thin board crack them over the years. One of the joints on the audio connector had the tell-tale grey ring around the pin, a hairline crack where the solder had pulled away from the pad. That alone could explain a channel going silent: a connection that is there when cold and open when warm, or open when something nudges the desk.
So: one obviously failed capacitor and one suspect joint. A satisfying haul for five minutes of looking.
Desolder, replace, reflow
Removing the old capacitor is the fiddliest part on a double-sided board, because the plated-through holes wick heat away and grip the leg. The trick is more heat for less time, not less heat for more. A hotter iron, a quick wetting of the joint with fresh solder to get it flowing, then braid or a solder sucker to clear the hole. Patience here saves you lifting a pad, and a lifted pad turns a five-minute fix into an afternoon of repair.
The replacement matters more than people expect. For electrolytics you match three things and improve one. Match the capacitance, the 1000 µF. Match or exceed the voltage rating, so a 25 V part can be replaced by a 35 V without complaint but never by a 16 V. Match the physical size roughly so it fits the footprint. And improve the temperature rating if you can: a 105 °C part will outlast the 85 °C part that died, which matters precisely because this one sits next to something warm. Mind the polarity. Electrolytics care which way round they go, and fitting one backwards is how you get the bang I did not have the first time.
old: 1000 µF 25 V 85 °C (bulged)
new: 1000 µF 35 V 105 °C (low-ESR)
The cracked joint just needed reflowing. Clean iron, a touch of fresh flux, heat the pad and the pin together until the old solder melts and pulls back into a clean shiny cone, add a hair more solder if it looks starved. A reflowed joint should look wet and concave, like a tiny volcano, not a dull grey ball. If it looks dull, it is cold, and a cold joint is just a future fault you have signed up for.
Test before you trust it
Resist the urge to bolt it back in the case and call it done. I powered it on the bench first, well away from anything, watched the new capacitor for a few minutes to be sure it was not getting warm or doing anything it should not, then fed it a quiet signal. I keep a cheap signal source and a hand on the mains switch for exactly this moment, because a recapped power supply is the one place where a mistake announces itself loudly. Sound. Both channels, clean, no hum, no crackle when I tapped the board near the connector that had the cracked joint.
A quick word on the tools, because people assume this needs a bench full of kit and it does not. A temperature-controlled iron that reaches around 350 °C, some decent flux, a length of desoldering braid, and a multimeter that does continuity and capacitance. That is the entire arsenal. The flux is the part beginners skip and regret, because old joints are oxidised and solder will not flow onto them cleanly without it. A dab of flux turns a frustrating reflow into a one-second one. The multimeter earns its place by letting you check a suspect capacitor in circuit, roughly, and confirm a joint is actually connected before you reassemble and discover it is not.
Total cost: one capacitor I already had in a drawer, and an evening I would otherwise have spent watching something I would not remember. The speaker is back on my desk and has behaved ever since.
The wider point is the one worth keeping. A great deal of "dead" electronics is not dead, it has one failed part at one of a few predictable places: a dried-out capacitor near the heat, a cracked joint on a heavy or stressed component, a fuse, a tired connector. None of it needs a schematic or an oscilloscope to find. It needs a bright light, a willingness to look closely, and an iron that gets properly hot. The skill is not in the soldering, which is mechanical and learnable in an afternoon. It is in believing the fault is something small and looking until you find it.