Capacitor plague: spot a swollen or leaking cap

If you've got a PC motherboard that won't POST, a monitor that clicks and dies, or an amplifier that hums where it didn't used to, the first thing to check before you condemn the board is the electrolytic capacitors. The "capacitor plague" of roughly 2000-2007 was a genuine industrial-scale failure — a stolen and incomplete electrolyte formula from a Japanese manufacturer ended up in cheap Taiwanese caps, which then ended up soldered onto millions of boards. The electrolyte vented hydrogen, the cans bulged, and the gear died. Dell GX270 and GX280 desktops are the textbook example — entire corporate fleets failed within warranty. Apple iMac G5s. Early flat-panel TV PSU boards from pretty much every brand. A pile of Asus P4P800 and A7N8X boards. HP and IBM workstations of the same vintage.

The plague proper is long over, but it had two lasting effects. First, it trained a generation of repair people to look at the tops of caps before they look at anything else, which is genuinely the right instinct. Second, it normalised the failure mode so much that lazy capacitor manufacturing still happens — modern boards from no-name PSU brands, cheap LED drivers, monitor power boards, and 10-year-old appliance control PCBs all fail the same way for the same reason. Heat plus time plus a marginal electrolyte equals a dead cap.

What follows is what to actually look for. None of this requires expensive kit, but an ESR meter helps.

What you're looking for visually

There are four classic signs and they're all on the top of the can.

1. Bulging top vent. Healthy aluminium electrolytic caps have a flat or very slightly domed top with a stamped cross or K-shaped score line — that's the pressure-relief vent. A failing cap pushes that top up into a visible dome. Compare against neighbours of the same series on the same board; if one is taller or rounder than the others, it's gone. 2. Electrolyte residue around the base. Look at the PCB right where the cap meets the board. Crusty brown or reddish-brown deposits, sometimes flaky, sometimes shiny, means the cap has leaked from the bottom seal. Often the cap looks fine from the top in this case. 3. Brown crust on the top. Dried electrolyte that has wept up through the vent without the can rupturing. Looks like dried gravy. 4. Top cracked open. The vent has fully opened. You'll see a star-shaped split. This one is unambiguous.

One important distinction: aluminium polymer caps (solid polymer, usually shorter and stubbier, often with a flat unscored top in dark colours) fail differently. They don't bulge or weep electrolyte in the same way because there's no liquid electrolyte to vent. They go short-circuit or lose capacitance silently. They are markedly more reliable than wet electrolytics in equivalent service, which is why high-end motherboards moved to them, but they aren't immortal. Standard aluminium electrolytics — the taller cans with the scored top — are the ones doing the visible failing.

What you're looking for with a multimeter

The right tool is an ESR meter. Equivalent Series Resistance climbs dramatically as an electrolytic cap dries out, well before capacitance drops enough to flag on a regular meter. A cheap MESR-100 or the Bob Parker-derived clones will tell you in seconds whether a cap is good, marginal or dead, and most will do it in-circuit on a powered-down board.

Without an ESR meter, a regular DMM with a capacitance function is a poor second. You can probe a cap in-circuit and get a reading, but in-circuit measurements lie — you're measuring the cap in parallel with whatever else is on that rail. A reading of "270µF" on a 220µF cap might mean it's fine, or it might mean you're reading the cap plus the rest of the bulk capacitance on the 12V rail. To get a real number you have to lift one leg of the cap, which means desoldering anyway, at which point you may as well just replace it.

The crucial point: a cap that looks visually perfect can be electrically dead. High ESR with a flat top is common, especially on caps that have spent years near a hot heatsink or VRM. If a board's symptoms point at a power rail and the caps on that rail look fine, test them anyway.

Brands and series to be suspicious of

Bad-batch series worth distrusting on sight:

• Capxon — any series, especially KF and GF. The poster child of the plague era and still produces marginal stock.
• GSC / Evercon — same parent, same problems. Common in cheap PSUs.
• Lelon — particularly the RXA and RXG series.
• Sacon — FZ series in particular, infamous for bottom-seal leaks.
• Nichicon HM and HN batches from roughly 2001-2005 — yes, even Nichicon shipped bad stock during the plague. The HM and HN series of that era are suspect; modern Nichicon is fine.
• OST, Fuhjyyu, Su'scon — generally low-grade Taiwanese stock.

Brands and series to replace WITH:

• Panasonic FR, FM, FC — workhorse low-ESR 105°C parts.
• Nichicon HE, HZ, HW — modern low-ESR series, excellent.
• Rubycon ZLH, YXF, MCZ — also excellent.
• Chemi-Con KZE, KY — solid.

If you can't read the brand because the sleeve is partly burned or the print is faded, replace the cap regardless. Mystery caps in a known-dead board are a waste of time to investigate.

Replacement rules

• Voltage: match or exceed. A 16V cap can be replaced with a 25V cap, never the other way round. Don't go more than one step up — physical size matters.
• Temperature: 105°C only. Never replace with an 85°C part, even if the original spec called for it. The cost difference is pennies and the life difference is years.
• Capacitance: match. Don't "upgrade" 1000µF to 2200µF on a switching supply — inrush current and loop stability were designed around the original value.
• ESR: equal or lower. Higher ESR will cause ripple problems and bake the cap. This is the spec people most often get wrong.
• Physical size: check diameter and height. Tight VRM areas on motherboards will not accept a fatter can.
• Lead spacing: 2.5mm or 5mm for the common sizes. Don't bend leads aggressively to fit.

On lead-free boards (everything since roughly 2006), the ground plane sinks heat aggressively. A 25W iron will fight you. Use a 60-80W iron, preferably temperature-controlled, and add a touch of leaded solder to the joint before you try to remove the cap — the eutectic mix melts at a lower temperature and breaks the joint loose. Don't lever the cap out while the solder is still solid; you'll lift the pad and the pad is worth more than the cap.

When it's not the caps

Most cap-failure boards present with one or two obviously-bulged caps plus a handful of borderline ones nearby — heat damage is local. If you can't find any visibly bad caps and ESR testing comes up clean, the caps probably aren't the fault. Look elsewhere.

On PSUs specifically, don't forget the primary-side bulk caps — the big 400V cans near the mains input. They fail less visibly than the secondary-side low-voltage caps but they do fail, especially in PSUs that have spent years running near rated load in a hot case. A primary cap with degraded capacitance will give you mains-frequency ripple, poor hold-up time, and sometimes startup failure.

Cases where cap replacement will NOT fix the board:

• Failed switching MOSFET on a PSU primary — usually shorted drain-source, often takes the gate driver with it. Caps may be fine.
• Dead PWM controller IC — the cap on the Vcc pin may have failed and killed the chip, but replacing only the cap leaves you with a dead chip.
• Cracked solder joints under BGAs or heavy connectors — common on monitors and TVs around the LVDS connector and PSU transformer pins.
• Failed inductor — rare, but a shorted-turns choke on a VRM will mimic a cap failure perfectly.
• Blown fuse with no obvious cause — something downstream killed it. Replacing caps and the fuse will just blow the fuse again.

If you've identified the bad caps but don't fancy the soldering — particularly on a densely-packed motherboard or a lead-free PSU board — send the board in for repair and we'll do it at the bench with proper desoldering kit and a fresh stock of Panasonic FR.

See also

• PSU faults: capacitor vs MOSFET failure
• Vintage amp hum or hiss when warm: recap symptoms