By Steve Rose
Friday, October 25, 2002
Another exchange on the Extreme Tech Forum: I've included this as a challenge to anyone who can correct my apparent misconceptions about the operation of surge suppressors.
One could consider that all "surge suppressors" operate in a serial, inductive mode. When a "parallel" device shunts a transient between the affected line and ground, the transient is not dissipated in the device, it is dissipated in the wiring leading to the device by inductive reactance. The energy dissipated by the device represents the shunted current times the "breakover voltage". Breakover voltage is placed in quotes as the voltage that causes the breakover may not be the same as the voltage across the device once breakover has occurred, as you pointed out with gas devices.
Lets say that there was a source of a 5000V transient located outside. The "parallel" device will pass whatever current is required to clamp the voltage at something close to its breakover (well, maybe not that close, say 1000V). The amount of current that it is required to pass is equal to the amount of current that drops the remainder of the transient voltage across the resistance / inductive reactance of the power line between the source and the "parallel" device (4000V). The energy of the transient is dissipated (or stored and released, as you pointed out) by the power line.
As a result, it always seemed logical to me that any "parallel" device should be placed at the end of the smallest wire gauge extension cord that meets the needs of the circuit, rather than plugged in to the wall with its twelve gauge conductors. (Note -- not that I _do_ it, but it always seemed _logical_). It appears that APC has come to a similar conclusion with the two chokes shown in your picture. You mention insulation a lot -- those coils look pretty close, and are insulated with enamel. Perhaps they serve as a spark gap in case of a lightning strike.
Which brings up "whole house" surge suppressors, that are connected to the service entrance wiring!? Do they contain a serial inductive element?
You are so right about protecting all wires coming into a computer. Anything that is a conductor connected anywhere to the set of protected equipment (antenna, cable television, network, telephone, power, streetcar) can carry a nearby lightning strike into an otherwise protected system with dire consequences. (The phone company didn't ground their service entrances in my neighborhood, since there was no electricity and they wrongly assumed that there were no ground rods installed. This resulted in a fried photovoltaic system when lightning struck my neighbor's windmill and traveled to my cabin via the phone line, jumped from the phone outlet to the metal lath under the plaster, the out of the lath to the case of a properly grounded flourescent lamp in the next room. We could reconstruct what happened due to the charred holes in the plaster.)
Another source of transients within an office environment, that generally wouldn't be suspected, is cooling equipment that has frosted up -- the office refrigerator, especially during the holiday season, or an overworked or dirty air conditioner that has frosted over. In this case, the compressor kicks out under load due to a self resetting protective device, which generates a very large inductive spike. In one case, the spikes attacked a well protected hard drive causing random sector errors, and in another caused a timer controlled playback device to start at apparently random times.
I'll be looking forward to your test reports.
Evidently our understanding of the operating principle of MOVs is quite different. At breakover voltage, it is my understanding the effective resistance of an MOV becomes effectively zero, shunting the surge. It is not a voltage regulator as would be a zener diode. We are not dealing simply with DC circuitry, as pointed out in the article. Surges are effectively high-frequency signals that do not behave as you would expect for DC. Putting a surge suppressor at the end of a long, high-resistance cable would not be effective. -winn
Hi Winn, I'm honored by your reply. You have cut right to the heart of the matter. Assuming (in the extreme) that the resistance of an MOV becomes 0 ohms when breakover occurs, it dissipates no energy by definition (0 ohms, 0 volts dropped across MOV, any current * 0 volts = 0 watts of energy dissipated). The energy of the surge must be dissipated elsewhere, and the only "elsewhere" is the circuit between the point where the surge is introduced and the terminal of the MOV. Steve