What about surges, spikes, zaps, grounding and your electronics?
Theoretically, the power coming into your house is a perfect AC sine wave. It is usually quite close. But occasionally, it won’t be. Lightning strikes and other events will affect the power. These usually fall into two general categories: very high voltage spikes (often into 1000s of volts, but usually only a few microseconds in length) or surges (longer duration, but usually much lower voltage).
Most of your electrical equipment, motors, transformer-operated electronics, lights, etc., won’t even notice these one-shot events. However, certain types of solid-state electronics, particularly computers with switching power supplies and MOS semiconductors, can be damaged by these occurances. For example, a spike can “punch a hole” through an insulating layer in a MOS device (such as that several hundred dollar 386 CPU), thereby destroying it.
The traditional approach to protecting your electronics is to use “surge suppressors” or “line filters”. These are usually devices that you plug in between the outlet and your electronics.
Roughly speaking, surge suppressors work by detecting overvoltages, and shorting them out. Think of them as voltage limiters. Line filters usually use frequency-dependent circuits (inductors, capacitors etc.) to “tune out” undesirable spikes – preventing them from reaching your electronics.
So, you should consider using suppressors or filters on your sensitive equipment.
These devices come in a very wide price range. From a couple of dollars to several hundred. We believe that you can protect your equipment from the vast majority of power problems by selecting devices in the $20-50 range.
A word about grounding: most suppressors and EFI filters require real grounds. Any that don’t are next to useless.
For example, most surge suppressors use MOVs (metal oxide varistors) to “clamp” overvoltages. Yes, you can have a suppressor that only has a MOV between neutral and hot to combat differential-mode voltage excursions, but that isn’t enough. You need common-mode protection too. Good suppressors should have 3 MOVs, one between each pair of wires. Which means you should have a good solid ground. Eg: a solidly connected 14ga wire back to the panel. Not rusty BX armour or galvanized pipe with condensation turning the copper connection green.
Without a ground, a surge or spike is free to “lift” your entire electronics system well away from ground. Which is ideal for blowing out interface electronics for printer ports etc.
Secondly, static electricity is one of the major enemies of electronics. Having good frame grounds is one way of protecting against static zaps.
If you’re in the situation of wanting to install computer equipment on two wire groundless circuits take note:
Adding a GFCI outlet to the circuit makes the circuit safe for you. But it doesn’t make it safe for your equipment – you need a ground to make surge suppressors or line filters effective.
Are you sure about GFCIs and ungrounded outlets? Should the test button work?
The NEC, section 210-7(d), and CEC, section 26-700(9), are quite explicit that GFCIs are a legal substitute for a grounded outlet in an existing installation where there is no ground available in the outlet box.
But your local codes may vary. As for the TEST button — there’s a resistor connecting the LOAD side of the hot wire to the LINE side of the neutral wire when you press the TEST button. Current through this resistor shows up as an imbalance, and trips the GFCI. This is a simple, passive, and reliable test, and doesn’t require a real ground to work. If your GFCI does not trip when you press the TEST button, it is very probably defective or miswired. Again: if the test button doesn’t work, something’s broken, and potentially dangerous. The problem should be corrected immediately.
The instructions that come with some GFCIs specify that the ground wire must be connected. We do not know why they say this. The causes may be as mundane as an old instruction sheet, or with the formalities of UL or CSA listing — perhaps the device was never tested without the ground wire being connected. On the other hand, UL or CSA approval should only have been granted if the device behaves properly in *all* listed applications, including ungrounded outlet replacement. (One of us called Leviton; their GFCIs are labeled for installation on grounded circuits only. The technician was surprised to see that; he agreed that the NEC does not require it, and promised to investigate.)