THE SURGE TO KILL
I confess: I'm a guy who rides large-displacement motorcycles too fast for complete control. And yet I'm pretty careful when it comes to electronic equipment. Maybe it's because it seems so helpless when it expires suddenly at the cruel hands of fate. At any rate, electronic equipment (particularly digital audio gear) is vulnerable to both annoying and catastrophic power-line problems. Even if you don't have any respect for your own mortality, your stereo gear should have line spike and surge protection, with hash filters thrown in too.
Partly it's a question of probability. For example, lightning strikes can occur anywhere, but they tend to favor certain unlucky areas. (There is no truth to the rumor that South Florida became a high-risk area only after I moved here, a few years ago). One lightning discharge might measure a million amperes and several million volts, providing instantaneous power of a trillion watts. A direct hit isn't required for damage; a strike anywhere in a local utility network can cause a spike of up to 5,000 V, lasting 100 µS, to head right up your power cord.
The other odds which must be considered are man-made fluctuations in line voltage. For instance, spikes can occur when inductive loads are switched off; when a transformer or motor coil is de-energized, the collapsing magnetic field sends a voltage spike of perhaps 3,000 V, or more, back through the line. This spike in the circuit can affect other circuits as well; wiring capacitance can couple transients from one wire to another or from a wire to ground. A vacuum cleaner on one circuit may affect your CD player on another.
Other man-made line variations proliferate. Aside from blackouts and brownouts (5 to 15% voltage reduction), surges above nominal line volt age, and sags below it, can occur.
Surges and sags are long-duration problems (as opposed to transients) stemming from load fluctuations in the network; their magnitude depends on the size of the load. The starting cur rent of a motor, a refrigerator, or an air conditioner, for example, can cause a sag, manifested by a momentary dimming of lights. A sag might measure as a 20% reduction in line voltage, lasting hundreds of milliseconds. Surges typically are caused by a sudden decrease in utility demand, resulting in an over-voltage until the system can compensate. Both sags and surges might also occur as a result of a lightning strike as the system tries to cope. The farther away you are from a power plant or substation, the greater the chance of fluctuation.
What kinds of problems can occur from power-line fluctuations? Anything from flakiness to annihilation. Integrated circuits are thoroughly checked before leaving the fabrication plant, but a certain percentage inevitably appears at the low end of the reliability scale. A marginal chip may fail (perhaps much later, intermittently) when subjected to the stress of power-line fluctuations.
Aging due to normal wear and tear is accelerated by over-voltage conditions caused by surges and spikes. More severe over-voltages damage power-supply components as well as chips. If there is no filter, or if there is some low-impedance path to ground, the surge or spike will use the device as its sink.
A rectifier diode might fail while handling a fraction of its rated current, voltage-regulator ICs and transistors might blow while not even running warm--all thanks to surges or spikes.
Even a powered-down device can be critically damaged; the surge may arc across the power switch, applying high voltage to the device. A lightning strike may cause a common-mode voltage surge in which both power lines are raised to a high potential relative to earth. The high voltage may cause arcing across conductors and ground, destroying anything which happens to be between them. The result: Barbecue.
Unquestionably, lightning is potent stuff, but some simple measures can go a long way. Lightning rods are de signed to provide an attraction point and current shunt. The rod itself is a sharp-pointed copper shaft connected by a copper cable to a copper-plated steel rod buried in the ground. When mounted at the highest point on a structure, the rod offers a zone of protection around it; larger structures re quire multiple rods. Because of the extreme voltages present in a lightning strike, flashover can carry high voltage to conductors 1 foot away. Further more, induced surges can spring up in any nearby conductors.
Fig. 1--A simple a.c. line interference filter.
Fig. 2--Installation of MOVs in an a.c. outlet box.
For increased protection against lightning strikes or any kind of power surge, transient suppression is required. Lightning arrestors such as zeners, thyristors, or varistors should be placed across the a.c. line and be tween a.c. lines and ground. A high-power zener network can be located across power-supply secondaries or between rectified outputs and ground.
A heavy-duty solution to recurring problems is the installation of a constant-voltage transformer, which can minimize the effect of both surges and sags. Isolation transformers provide an ideal way to protect equipment from power fluctuation.
In addition to power surges, noise (often called hash) on the line can cause nonlethal yet annoying glitches. Sources of hash include motors, small electrical appliances, corroded light sockets and cords, fluorescent lights, SCR or triac dimmers, welders, X-ray machines, photocopying machines, d.c. switching power supplies, and internal combustion engines. We may classify the noise as radio frequency interference (r.f.i.), electromagnetic interference (EMI), or electromagnetic pulse (EMP).
Digital equipment itself (e.g., your CD player) is a prime source of electrical interference, FCC requirements notwithstanding. With system clock frequencies of 1 to 10 MHz, and the resulting harmonics, broad-band interference up to 54 MHz (channel 2) or beyond can be anticipated. Digital gates open and close at high speed a rate equal to r.f.i.; that confusion be tween normal digital signals and generated noise can cause internal errors as well as errors in surrounding equipment. Ironically, not only is digital equipment a prime culprit, it is also more susceptible to noise than ordinary appliances.
A simple AM radio can be used for EMI testing; tune it to the loudest noise and then wander around looking for sources. Not only will digital equipment radiate noise, but connecting cables are culprits as well. An a.c. cord can radiate noise (especially if it's long and ungrounded).
Low-pass filters are effective against both transient and noise problems; they allow 60 Hz to be passed whereas high frequencies are attenuated. This is accomplished by a high series input impedance and a low shunt impedance to ground. A parallel capacitor or series coil is the simplest example; veteran digital designers are well familiar with the 0.01-uF capacitor placed be tween the power buses and ground to suppress switching transients. More typically in protection devices, L, T, or pi sections are employed; although the filter should be installed at the point where the line enters the equipment's casing, it is usually inserted at the power receptacle. An example of an r.f.i./EMI filter is shown in Fig. 1.
Hash filters can be installed in the device under attack. However, a more enlightened approach counterattacks by eliminating hash at the source. Any defective wiring should be replaced, filters should be put on noisy tools or equipment, all covers and shields should be securely fastened, and grounding rules should be scrupulously followed to minimize interference. A list of rules for r.f.i./EMI protection includes the following: Enclose sources and receivers within a shield, filter all incoming and outgoing leads, use shielded coaxial cable for high frequencies and twisted leads for lower frequencies, use single-point grounding, eliminate ground loops, and keep signal-sensitive leads short.
In addition to filters, crowbar circuits and voltage clampers can be used to suppress transients. Crowbars use a thyristor or spark gap to divert transients. It is critical to ensure that the protection device can operate quickly enough to catch spikes; SCRs and triacs, for example, are often too slow.
A metal oxide varistor (MOV) is a volt age-clamping device; it operates in a way similar to that of a back-biased zener diode. Below the threshold volt age the MOV is an open circuit; above the threshold it conducts and thus absorbs the transient, dissipating it as heat. An MOV can respond to a transient in a few nanoseconds, with peak current capacities up to 50,000 amperes with a wide variety of operating voltages.
In terms of bang for the buck, the MOV is perhaps the most effective insurance you can buy. It guards against high-energy power-line transients and can be installed in existing power strips. For example, the General Electric V130LA10A MOV will clamp to 340 V at 50 amperes in 35 nS; it is available from Radio Shack. Three of them should fit neatly inside a power strip, wired between the hot, neutral, and ground wires as shown in Fig. 2. An MOV does not guard against hash; a separate line filter must be employed.
When specifying power-protection devices, you must determine how many output watts your application re quires. Add up the ampere ratings of the equipment to be protected on a line, multiply by the voltage (120 V), and throw in some headroom; the resulting figure is an approximation of the wattage required to operate the devices and the rating required of the protection device. A good protection device should offer fast response to repeated voltage spikes and high-frequency filtering for incoming and out going noise.
Line protection--you can pay a little for it now, or you can pay a lot for it later.
(by KEN POHLMANN; adapted from Audio magazine, Nov. 1987)
= = = =