First post, by songoffall
- Rank
- Member
This will be a short explanation on certain failure modes of capacitors, for people to understand the process better. Also to explain the "capacitor plague". Feel free to add to it - I don't claim to know it all or have the ability to put it into a single post.
What is a capacitor?
A capacitor is essentially two electrically conductive surfaces, called plates, separated by a non-conductive dielectric. When you apply different charges to these two surfaces, those charges are drawn to each other, so the capacitor can hold charge.
Filtering capacitors, for example, will absorb charge when the current goes up and give up charge when the current goes down, thus filtering out changes or noise in currents.
There are polarized and non-polarized capacitors: polarized capacitors have designated positive and negative plates, called anode and cathode.
If you take the simplest capacitor, where two plates are separated by air, glass, vacuum, silicon, these are non-polarized, and it doesn't matter which plate takes positive or negative charge.
Ceramic capacitors and film capacitors work the same way.
Electrolytic capacitors are different.
E.g. alumnium electrolytic capacitors are two aluminum plates coiled around each other with paper between them. Then the paper is soaked with conductive electrolyte. Then low voltage is given to the plates - at this point what we have is not a capacitor and the current can flow freely from plate to plate through the electrolyte. But the process causes the electrolyte to react with the positive - anode - plate, and build a non-conductive aluminum oxide layer on it. Then the current is raised to the operating voltage of the capacitor.
The thicker the oxide layer - the higher the voltage, at which the capacitor can operate, but also the lower the capacitance - how much charge the capacitor can hold. Connecting electrolytic capacitors backwards will swap the anode and the cathode, but because the conditions of the process are not the same as in the factory, it will be too fast, consume a lot of power and transform it into heat, there will be gas buildup from the chemical reaction and the capacitor will fail. And even if it doesn't, the end result will not fit the original specs.
While there's still electrolyte inside the capacitor and the capacitor is operated within specs, the oxide layer will keep reforming, which is why aluminium electrolytic capacitors are self-healing.
Tantalum caps are also electrolytic capacitors, except a different method and different materials are used when creating them, but while they might look like ceramic or film caps, they are polarized.
How capacitors fail
1. Dielectric breakdown. Air is a dielectric, but if you charge two plates with opposing charges and keep raising the voltage, an electric arch will form between those plates, and air will start behaving like a conductor. Every dielectric has its own dielectric breakdown voltage, and a capacitor's dielectric breakdown voltage depends on the dielectric material, its thickness and the surface area of the plates. Which in turn may be affected by the conditions in which the capacitor operates, and the age of the capacitor.
Because even at breakdown dielectrics have a lot of resistance, a lot of heat is released, and causes the capacitor to lose its integrity and fail.
In common capacitors, it may cause the plates to make contact and short. This is what happens with most capacitors when they fail due to dielectric breakdown.
In safety capacitors, the failure does not cause a short but an open circuit - they are built so when the capacitor fails, the plates are drawn apart. REEFA caps, known for releasing their magic smoke and wonderful aroma at failure, are safety caps.
2. Electrolytic capacitors have additional modes of failure. In aluminium electrolytic capacitors, the oxide layer will keep deteriorating without the self-healing while in operation. And when you turn on an old device that has been in storage for years without safely reforming the capacitors, they will reform rapidly, which will cause a buildup of hydrogen gas and the temperature will raise inside the capacitor. This, in turn, may cause the capacitor to bulge and even bubbles to form between plates driving them apart. Some capacitors can even explode from high internal pressure or leak electrolyte on the PCB.
If kept in operation, electrolytic capacitors will keep self-healing until the electrolyte is spent and the capacitor starts drying up. It will drift out of spec and eventually fail, but in a non-violent way.
The capacitor plague
The capacitor plague refers to the earlier than expected failures of electrolytic capacitors in the first half of 2000s. Usually these are capacitors of Taiwanese origin.
Bad electrolyte causes corrosion inside the capacitor and a buildup of gas which causes bulging, ruptured or leaking capacitors even in normal operation. I have a bunch of Radeon 9200SE cards from the same era and every single capacitor on them has failed. Same often goes for motherboards from early 2000s.
There's a theory that Taiwanese manufacturers tried to steal the electrolyte formula from Japan and the formula was miscopied, but I don't really know the details of that story.
But, in the end, the capacitor plague was one of the factors that cemented the solid reputation of Japanese capacitors.
Compaq Deskpro 2000/P2 300MHz/384Mb SDRAM/ESS ES1868F/Aureal Vortex 2
Asus A7N8X-VM400/AMD Athlon XP 2ooo+/512Mb DDR DRAM/GeForce 4 MX440/Creative Audigy 2
Asus P5Q Pro/Core2 Quad Q9400/2Gb DDR2/GeForce 8800GT/Creative X-Fi