We were discussing lightning suppression last time, and that just seems to lead logically to surge suppression techniques in electronic equipment. It's a huge industry, and the continuing popularity of fragile computer equipment means it's getting bigger all the time. There's a lot of black art and pseudoscience involved, too.
The surge we're trying to protect our precious equipment from is an above-normal voltage. For the time being we don't need to worry about where it came from: maybe a direct lightning strike, more likely an inductively- or capacitively-induced spike, or a wallop from switching action on the grid (are you listening, Ontario Hydro?). The basic building block of all surge suppression is the transient suppressor. There are two basic flavours: some devices change their impedance exponentially as voltage is raised, others have a threshold voltage where they suddenly change behaviour. Your thyrites and MOV's (metal-oxide varistors) are in the first category; gas discharge tubes and zener diodes are in the second.
A thyrite is usually a stack of disks of silicon carbide, often in a high-voltage power supply. They've been around since the 1930's, originally for protecting high-voltage transmission lines. They drop in resistance when the voltage is raised. They can handle large amounts of power, but you don't see them much in today's designs… one of the reasons being that they draw a significant amount of current even under normal voltage conditions. So they're quite big, and can get quite warm. But they were one of the earliest forms of suppression, and they led to the MOV.
MOV's are ubiquitous today. They're mostly made of zinc oxide, with a few trace elements thrown in. They have a much sharper "knee" and leap into action more sharply than thyrites. They're cheap and reliable and can handle a fair amount of energy, and when they fail, they short-circuit. That can be a good thing, since they'll continue to provide circuit protection even after they're cooked. Unless they explode.
Which they do quite often.
MOV's have gotten a bit of a bad reputation (apparently unearned) amongst the so-called experts, though. There have been claims that they are slow to react, and that their voltage threshold (the location of the "knee") drifts after they've been used. Further research has shown that the basic electrochemical process in the MOV takes place in about 500 picoseconds (that's pretty fast!) The culprit in the slowdown, of course, is the inductance of the component leads, and we can minimize that by using good RF techniques and keeping leads as short and direct as possible. And it turns out that the threshold does change with use, but as the component ages (after a few more "hits") it returns to its nominal value.
Gas-discharge tubes are used in telecom circles, along with carbon contacts ("carbons"). They consist of a couple of closely spaced contacts in a metal tube. Not much call for them in power supplies, since once they start arcing, they won't stop until the voltage is near zero. Good potential crowbar, though. Some small transmitters (Telefunken is one) place them across the output terminals.
Zener diodes can make an effective crowbar, too, but they are somewhat frail. Overvoltage conditions create a very small active hot spot inside them, and this is where they tend to fail. When they fail they may go short, or open, or somewhere in between. Some manufacturers claim zener action can take place in one or two picoseconds, which may be true at the molecular level, but defies belief for any leaded component (read: in the real world). Which is why 99 times out of 100 you'll find the MOV doing the job.
In addition to the transient suppressor, which is placed as a shunt to take the surge away from the load, many devices include a series low pass filter to delay the surge's passage to the load, and give the suppressor time to work. Sometimes a current-limiting device (a resistor or fuse, perhaps) is placed in series with either the shunt or the load to prevent its destruction.
RESULTS OF LAST MONTH'S QUIZ: Here's a circuit to convert a dual linear potentiometer into a single logarithmic or almost-audio-taper pot. Hey, it might come in handy some day!