Why does a two-cycle power outage create havoc with electronic equipment when, sometimes, a three-cycle outage causes no problems?

We seem to know so little about how power grid values that are higher or lower than guaranteed nominal values affect unprotected electronic equipment. Equipment that process intelligence (computers, communications, etc.) rather than power (adjustable speed drives, electroplating rectifiers, etc.) have a common denominator. The incoming grid power "sees" the equipment's DC power supply, which, in turn, bears the brunt of any AC grid variations.

The DC power supply

As shown in the accompanying diagram, the DC power supply is a sophisticated piece of equipment. Its basic function is to deliver stabilized low-voltage DC to the digital logic it feeds. Based on a fast-switching DC-to-DC converter, the unit converts rectified 60 Hz into the low-voltage DC (typically 5V) required by computer logic. The nominal 5VDC output is compared to an accurate 5V reference so that an error-correcting feedback signal can be developed. This signal adjusts the relative ON and OFF durations of the DC-to-DC converter, thereby holding the output at the required 5V.

In principle, switch-mode power supplies (SMPSs) have the ability to bridge a total power outage for periods up to three complete cycles. The key requirement for maximum immunity against power outage lies in the filter capacitor (labeled "C1"in the diagram) being fully charged to design voltage. This capacitor acts like a short-term battery; during a power outage, the power supply's DC-to-DC converter is kept running by drawing current from this capacitor.

Stored capacitor energy is equal to one half of capacitance times peak voltage squared (1/2)(C)([V.sub.p])[sup.2]. Because the amount of the stored energy varies with the square of the peak 60-Hz voltage, the capacitor's ability to sustain inverter operation during an outage drops off twice as fast as reductions in line voltage. As such, the power supply ride-through capability depends strongly on the capacitor being charged to full design voltage immediately prior to the power outage. This is why one 3-cycle outage may not affect electronic systems while another such outage will shut the equipment down.

Impact of voltage sag

Many electronic equipment manufacturers assume that their hardware will operate from a distribution network with zero internal impedance, receive a pure sinewave, and never be subject to line voltage variances of [+ or -] 5% from nominal. As such, most power supplies are designed to accommodate prolonged line voltage sags to these levels. However, the combination of utility and locally generated disturbances knows no such modest bounds. Utilities are permitted line voltage reductions (brownouts) to cope with seasonal demands. Large motors that accelerate high inertia loads, spot welding, and many other loads act to further drop the voltage level available at electronic power supplies in a typical facility.

Computer shutdown and sag-induced logic errors are not the only problems. Actual damage to the DC power supply is an even greater danger. Reduced input voltage can cause excessive power supply heat dissipation, resulting in short equipment life.

Why this overheating? To maintain constant DC output as line voltage declines, the DC-to-DC converter circuit must draw from the reservoir capacitor. Consequently, with line voltage reduced, this capacitor undergoes deep discharges between the twice-per-cycle charging periods.

Electrolytic capacitors are not designed for deep discharge, nor for the large terminal voltage variations that result. Excessive capacitor charge and discharge currents cause internal heat dissipation, producing dielectric stress and reduced mean-time between failure (MTBF). Rectifiers and DC-to-DC converter switching transistors also draw high peak currents, which raise their junction temperatures and, consequently, take a toll on semiconductor longevity.

Impact of overvoltage, surges, RFI, and hormonics

Short-term voltage surges 10% beyond nominal usually are not harmful; however, higher input voltages may overwhelm the power supply's voltage regulating capability, feeding damaging voltage levels to the electronic circuits.

High input voltage also can puncture the power supply's rectifier and switching transistor junctions, causing MTBF degradation and breakdown. High-voltage transients lasting mere microseconds can permanency wreck both the power supply and its electronic equipment load as well.

Digital logic circuits that define "zeros" by voltages in the 0V-to-0.5V region, and "ones" by 4.5V-to-5V levels, are highly susceptible to inductive "kicks" that are directly impressed on their 5VDC power supply.

The power supply's reservoir capacitors (noted as "C1" and "C2" in the diagram) don't absorb transient energy because their wiring inductance (negligible at 60 Hz) introduces significant isolating impedance at the MHz-equivalent frequencies of fast-rise transients. As a result, transient energy follows the line of least resistance, which is to the power supply's output terminal.

Line-borne noise (RFI and low-voltage transients created by high-current logic circuits) is unlikely to damage the power supply. However, few power supply designs have careful component shielding and placement. As such, line noise can be coupled by stray capacitance to the DC output, where it disrupts communications and computer circuits. Because this noise may be intermittent and is usually beyond the frequency range of many measuring instruments, diagnosing this source of equipment malfunction is difficult and time-consuming.

Harmonic voltages of the 60-Hz line frequency that are impressed on the AC power line are unlikely to damage the power supply. Nevertheless, higher harmonics of the 60-Hz power supply can fool certain control circuits into erroneous operation. Timing operations that are initiated by a sinewave's zero crossing can be falsely triggered by the more numerous zero crossings of higher harmonic frequencies.