Yesterday, during my office hours in the lab, a couple of the students finally did something I’ve been expecting all quarter: blew up an electrolytic capacitor. Luckily no one was hurt. The problem was the expected one: hooking up the capacitor backwards. In this case it was a little more subtle: the capacitor was being used in the low-pass LC filter at the output of their class-D power amplifier, and they had hooked the negative lead up to ground, instead of to their negative power rail. They actually managed to blow up two of the capacitors (the second much less dramatically) before getting the right power connection.
I looked around on the web to find a good explanation to give the class about electrolytic capacitors (most just say “don’t hook it up backwards” but don’t explain what is really going on)—I thought I knew how the electrolytic capacitors worked, but was not 100% certain, so I was glad to find the following explanation that matched my understanding exactly:
Electrolytic capacitors by Hans Egebo. The following is just an excerpt from http://www.hans-egebo.dk/Tutorial/electrolytic_capacitors.htm, which has more to say about the capacitors—Hans does make the mistake of claiming that the US uses MF for µF and MMF for pF, but I don’t know any electrical engineers who still use those archaic abbreviations—I’ve seen them in some old books, but I think that the usage went out of style in the 1950s. I’ve taken the liberty of Americanizing the spelling in the excerpt.
So, how to build a large capacitor?
One way to make a large capacitor is to take two long strips of aluminum foil (=large plates), put strips of isolating materials between them, and make a nice compact roll. Capacitors up to around 1µF can be made this way, but they are physically big, so if we want even higher capacity, we need to look for other things than plate area. It happens that we know of a very thin and very voltage resistant type of isolation material: Aluminum oxide. If we cover a strip of aluminum foil with a thin oxide layer, we have one plate and a very thin dielectric. Problem now is to make the other plate come close enough to the other side of the oxide layer. The thing that comes really close to anything is a liquid, so if we submerge our oxide covered plate in a conducting liquid, the liquid forms the other plate, and we can make a very large capacitor. A conducting liquid is called an electrolyte, see?
But there are problems
Nothing comes for free, so this type of capacitor has its drawbacks. Some have practical solutions: Instead of having liquid sloshing around inside the capacitor, an electrolyte-soaked paper is used, some modern types are even virtually solid. Others become restraints we have to live with:
The oxide layer is made by an electrolytic process; the foil is submerged in some liquid and current is passed through the liquid into the metal, forming the oxide layer. This is an advantage and a disadvantage: The good news is that the dielectric layer has self-healing capabilities, so if a weak spot occurs, the resulting leakage current will more or less rebuild the isolation. This is the reason electrolytic capacitors can regenerate if you raise the voltage over them slowly. The bad news is that the process is reversible! If you reverse the polarity, even the slightest leakage will begin to tear the dielectric down, resulting in more leakage, which tears away more dielectric, which—well, you get the picture. This is the reason you need to observe polarity strictly when using electrolytic capacitors.
Another problem is the presence of the electrolyte itself: Excess heat, either from inside or outside sources, will eventually start to evaporate the liquid, building up pressure, which may very well result in a violent leak, even an explosion. And as if that was not bad enough, once the electrolyte escapes, it will interact corrosively with other parts of the equipment, because all electrolytic liquids are more or less corrosive.
Finally, even if our electrolytic capacitor escapes a violent demise, its very construction gives it a limited life-span. Given time, the dielectric may deteriorate beyond regeneration, resulting in a high leakage current, or the electrolyte will eventually dry away, reducing the capacity by several orders of magnitude. This is the reason that people who restore antique radios will often be faced with the need to replace electrolytic capacitors.
Three sure ways to kill an electrolytic capacitor:
- Overvoltage: If the specified voltage is exceeded, current will leak through the isolation, not in a slow way that might regenerate weak areas, but violently, creating hotspots where additional break-down occurs. The danger of explosion is imminent.
- Reversed polarity: As described, the inverse of regeneration = self destruction, will occur. If the applied voltage is near the normal (right polarity) working voltage, break-down is quick and violent. The effect of a low inverse voltage might be reversible.
- Heat. Heat shortens the life of an electrolytic capacitor. A good rule of thumb is that every 10° C over 85º will cut the life expectancy in half.
Electrolytic capacitors are a very cheap and easy way to get high capacitance, but their somewhat low reliability and limited lifespan have caused problems, particularly if poor manufacturing practices are used in making them. Almost all computer manufacturers ran into trouble between 2002 and 2007 with improperly made electrolytic capacitors from Taiwan (see the Wikipedia article on “Capacitor plague”) which used the wrong electrolyte.
It is possible to get bipolar electrolytic capacitors (where the aluminum oxide coats both plates) that are not subject to the reversed polarity explosion (though they can still fail due to heat or overvoltage). Using bipolar electrolytic capacitors in the circuits course would add several dollars to the parts cost, since the larger bipolar ones cost about $1 each, and would not reflect normal practice, as bipolar electrolytic capacitors make up only about 1% of the electrolytic capacitor market (based on the number of parts listed at DigiKey).