Capacitor Failure Symptoms

Capacitors filter power rails, decouple logic, set timing and couple signals in every vintage computer, console and monitor, and several kinds change value or fail outright with age. The signs of capacitor failure, the tests that confirm it, and the capacitor makers' guidance on replacement, cleaning and storage are below. Values for particular machines are on the guides in Category:Capacitor Replacement Guides.

Capacitor families fail in different ways. An aluminium electrolytic loses electrolyte, and its capacitance falls while its ESR rises. A solid tantalum has no wear-out mechanism but fails short under overstress, which can lead to overheating and ignition. A surface-mount ceramic can crack when the board flexes, and an old epoxy-cased paper mains suppression capacitor can burst and smoke when switched on. Start by identifying which kind of part you are looking at.

Common symptoms

Symptoms of failed capacitors

Symptom	Usual cause
Bulging or domed top, or a split vent	Gas generated inside the can has opened or lifted the pressure relief vent. Nichicon lists the causes as reverse voltage, over-voltage, extreme ripple or AC voltage on a polarised part. Goldwasser's notes add that bulging electrolytics are leaky and usually run hot.
Electrolyte at the base, or corrosion around the leads	The seal has failed and electrolyte has escaped onto the board.
Hum bars in the picture, hum in the audio, a flickering display, or a low B+ supply in a monitor	A reservoir or filter capacitor that has lost capacitance or gone open. After a full-wave rectifier the ripple is at twice the mains frequency, so the hum is at 100 Hz on 50 Hz mains (120 Hz on 60 Hz).
A switch-mode supply that will not start, cycles, or runs unstably; deflection faults in a monitor	High ESR. Goldwasser notes that raised ESR can cause "totally incorrect or unstable operation" of switch-mode supplies and of TV and monitor deflection circuits, and that an electrolytic can develop high ESR while still reading close to its marked capacitance on an ordinary meter.
A dead rail, a blown fuse or a burnt patch on the board	A capacitor that has failed short. Nichicon calls shorts "very rare" in aluminium electrolytics in the field; Vishay notes that a solid tantalum fails short unless it has a built-in fuse.
A loud crack and acrid smoke from the mains input at switch-on	A failed metallised paper X or Y suppression capacitor.

Failure modes by capacitor type

Aluminium electrolytic

A vented electrolytic capacitor beside others that are bulging (photo: WizardTim, CC BY 4.0)

Nichicon divides failure into catastrophic failure, where the capacitor has completely lost its function through a short or open circuit, and degradation failure, a gradual drift outside the limits in the part's own specification. In the field it lists four modes:

• Short circuit: "very rare", caused by vibration, shock or stress on the leads, or by over-voltage, extreme ripple or pulse current.
• Open circuit: from mechanical stress at mounting, from halogen-containing cleaning agents or coatings corroding the leads and foils, or from the electrolyte drying out when the seal deteriorates or the part runs hot.
• Capacitance drop and high loss (high ESR): from continuous reverse voltage, ripple above the rated maximum, or extreme charge and discharge.
• Vent operation: gas generated by reverse voltage, over-voltage, extreme ripple or AC voltage.

Visible signs are a bulging or split vent on the top, a domed base on a surface-mount part, fluid or dried crust at the base, and corroded leads or discoloured board around the part. A capacitor that looks perfect can still have dried out, so measure as well as look.

When a vent operates the "smoke" is hydrogen gas and vaporised electrolyte, not combustion. It is over 100 °C. Switch off, keep your face away, and wash any electrolyte off skin with soap and water.

Tantalum

Dipped tantalum capacitors from a Heathkit oscilloscope that failed short circuit. They show little outward damage (photo: WizardTim, CC BY 4.0).

Vishay states that solid tantalum capacitors have no known wear-out mechanism, but that excessive voltage, current and temperature reduce their reliability, and that overstress can lead to overheating and "possibly ignition". Vishay normally recommends 60 % to 50 % voltage derating, so the applied voltage is no more than 40–50 % of the rating. By that rule a tantalum on a 5 V rail should be rated at least 10 V, and preferably more. Vishay also notes that a fuse in series changes the failure mode from a short to an open circuit.

A shorted tantalum can take a regulator, a fuse or series resistors with it. It may show no outward damage, or a cracked or blackened case.

Ceramic

Ceramic capacitors have no electrolyte. Two behaviours are easy to mistake for faults or to miss:

• Flex cracks in surface-mount multilayer parts. KEMET calls flex cracks "the primary failure mode of MLCCs", caused by tensile and shear stress during board flexure and thermal cycling, and says they "can result in low IR or short circuit failures". A cracked part usually shows nothing on the outside. It is found by measurement, or by finding a rail pulled down.
• Class II dielectric behaviour. High-permittivity dielectrics such as X7R change capacitance with voltage and with time; KEMET's X7R data gives an ageing rate of up to 3 % capacitance loss per decade hour. Nichicon's comparison of dielectrics gives barium titanate, the high-permittivity ceramic, a dielectric constant of 500 to 20,000, against 15 to 250 for the titanium-oxide temperature-compensating type. A Class II decoupling capacitor that reads below its marked value can be behaving normally.

Film and mains suppression capacitors

A RIFA 68 nF 250 V X2 capacitor that has burst in a Tektronix 2445 power supply (photo: J. C. Barros, CC BY 2.0)

Goldwasser's notes describe film (Mylar) and disc ceramic capacitors as rugged and rarely found bad. The exception is the mains suppression capacitor across the line (Class X) or from line to earth (Class Y) in the power supply input.

The usual problem part is RIFA's PME series of epoxy-encapsulated metallised paper capacitors. TekWiki describes the failure: in examples over 20 years old the epoxy cracks, humidity enters while the equipment is switched off, and at the next switch-on the capacitor fails, "releasing foul-odoured smoke and sometimes even catching fire". TekWiki notes that equipment of many makes from the 1970s to the 1990s is affected, typically equipment with switch-mode power supplies, and recommends replacing these parts proactively with a modern part of the right rating and class. Jerry Walker's tests, reported by Hackaday, found that parts which had been used across AC mains were more likely to fail than parts used in DC circuits.

The PME271M is still made by KEMET. Its datasheet describes multilayer metallised paper encapsulated and impregnated in self-extinguishing material meeting UL 94 V-0, rated 275 V AC and qualified to EN/IEC 60384-14, including the active and passive flammability tests. Replace a suppression capacitor with a modern part of the right rating and safety class: Class X across the line, Class Y from line to earth. An ordinary film capacitor of the same value and voltage is not a substitute.

Testing a suspect capacitor

Capacitor tests

Method	Tool	What it shows
Visual inspection	None	Vented, leaking, bulging and cracked parts only.
ESR	ESR meter	Raised ESR, which a capacitance meter can miss. ESR meters use a low test voltage, so they can usually test in circuit; a very low resistance in parallel can make a bad capacitor read good. Put the probes on the capacitor's own terminals.
Capacitance	Capacitance or LCR meter	Lift one leg for an accurate reading. Standard tolerance on an aluminium electrolytic is ±20 % (M).
Leakage	Current-limited bench supply, or a meter with a leakage function	Charge to rated voltage through a resistor and watch the current settle. Leakage current rises during long unpowered storage and falls back when voltage is reapplied.
Resistance	Multimeter	Hard shorts and, on large electrolytics, an open part. It will not find high ESR.
Ripple on the rail	Oscilloscope	Ripple on a DC rail under load, the direct evidence that filtering has failed.

Discharge a capacitor before removing it. In a monitor or an all-in-one machine, read CRT Discharge Procedure first.

Replacing capacitors

Match or exceed the voltage rating of the part removed. A rating at or below the rail the part sits on is wrong. Where a board was built in different versions for different mains voltages, check which version you have before ordering: Console5's lists for the Macintosh Classic power/sweep board 630-0560 give CP1 and CP13 as a pair of 220 µF 250 V parts on the 120 V version, and CP1 as a single 220 µF 400 V part, with CP13 replaced by a wire link, on the 240 V version.

Electrolytic life roughly doubles for every 10 °C drop in temperature (the Arrhenius rule), so a 105 °C part outlasts an 85 °C part of the same endurance hours at the same temperature. Jeroen Stessen notes in Goldwasser's FAQ that 105 °C types tend to have higher ESR than 85 °C ones, so where the heating comes from the part's own ripple current rather than from a nearby heatsink, check the ESR and ripple ratings as well. Exceeding the ripple rating is one of Nichicon's listed causes of capacitance loss and rising ESR.

Buy parts whose ESR, ripple current and endurance figures are published for the exact part number. See Battery Explosion, Capacitor or Corrosion Damage for choosing parts and suppliers.

Cleaning up after a leak

Remove the failed parts first, then clean the board. Nichicon's recommended cleaning agents for boards carrying aluminium electrolytics are isopropyl alcohol, water-based alcohol cleaners and surfactant cleaners. It limits cleaning to five minutes at under 60 °C, and asks for the board to be dried with hot air for at least ten minutes, below the sleeve's softening temperature of 80 °C.

Nichicon also warns against:

• halogenated cleaning agents, coatings and adhesives, which corrode the capacitor from inside;
• fluxes containing non-ionic halides, which have the same effect, so clean flux residue off;
• acidic or alkaline solutions on or near a capacitor.

Inspect every track and via in the affected area under magnification before applying power.

Normal behaviour that is not a fault

• A Class II ceramic decoupling capacitor reading below its marked value.
• An electrolytic reading up to 20 % off its marked value.
• A non-polarised (bipolar) electrolytic with no polarity stripe. Nichicon notes that bipolar types are for circuits that may see reversed polarity, and that even they "cannot be used in continuous AC applications".

Storage and re-forming

Nichicon states that an aluminium electrolytic stored unpowered for a long time develops a slightly raised leakage current, because the anode oxide layer reacts with the electrolyte. Applying voltage re-forms the oxide and the leakage current returns to its initial level. Its procedure for a loose capacitor is to charge it to its rated voltage through about 1 kΩ for about 30 minutes. For a capacitor already in a circuit with an adjustable supply, run at about half the rated voltage for about ten minutes, then raise the voltage gradually while watching the equipment; with a fixed supply, run for about thirty minutes and switch off. A capacitor stored for less than two years at 5 °C to 35 °C can be used without this treatment.

Re-forming treats leakage current only. It does not restore lost capacitance or lower a raised ESR. Nichicon recommends storage at 5 °C to 35 °C and 75 % relative humidity or less, out of direct sunlight.

Examples from manufacturers' documents

• Sega's parts lists encode the capacitor type in the part number. The Genesis 32X / Mega Drive 32X service manual lists electrolytics as 150-xxxx, ceramics as 151-xxxx and tantalums as 153-xxxx. CE18 on the main board is 150-0498, "CAP OS 100UF 10V", an OS-CON solid-polymer part (10SA100M), so a general-purpose electrolytic there is an ESR downgrade. CE1 and CE2 on the sub board are 47 µF 10 V tantalum chips (153-0119). See Sega 32X.
• On the Macintosh 128K and 512K analogue board, Larry Pina traces poor horizontal linearity (the left of the screen wider than the right), then bright diagonal lines a few minutes after switch-on, a darkening display and smoke, to C1, a 3.9 µF non-polarised high-frequency capacitor in the horizontal sweep circuit. He recommends a 100 V replacement and checking the yoke plug J1, which may be burned. The Macintosh SE has the same fault at C15.
• Pina's parts lists for the U.S. and International versions of that board use the same designators for different parts: C13 is a 33 µF 16 V electrolytic on the U.S. board and a 0.01 µF 1.5 kV ceramic disc on the International one. Both lists identify the "molded minibox" RIFA suppression capacitors and the 100 µF 200 V mains-side electrolytics. See Macintosh 128K/512K Capacitor Replacement Guide.

Related pages

• Battery Explosion, Capacitor or Corrosion Damage
• Category:Capacitor Replacement Guides
• CRT Discharge Procedure
• Recommended Tools