These days power-quality trouble-shooters need to do more than decipher data from three-phase monitors. They need to be skilled at using test tools to troubleshoot problems in the premises wiring system, from the receptacle on up. Relatively low-cost hand-held test tools are all that is required for measurements.

"Bottom-up" troubleshooting makes sense for a number of reasons: * Comprehensive studies in the early to mid-1990s demonstrated that relatively few sags and other voltage events were caused by the utility, with the majority being originated in the facility.

*Roughly three-quarters of power-quality problems arise from premises wiring and grounding issues.

* Power quality is usually best at the service connection and gets worse moving toward the most remote point of the distribution system-the receptacle outlet. That is also where the sensitive loads (and users) tend to be located, and where the first call for help often originates.

Measurements at the outlet

In a three-pronged outlet, there are only three possible voltage measurements, and only two of these are useful: Line-Neutral (L-N) Voltage and Neutral-Ground (N-G) voltage. A skilled troubleshooter takes measurements at the outlet, and isolates whether the problem came from the power system or has its roots in the equipment itself . Let's say the manufacturer's service tech has already swapped boards a couple of times and all eyes are now glaring at the gremlin in the wall socket.

The first power-quality question we now need to answer is whether the L-N voltage is of sufficient and consistent amplitude to serve the needs of the load.

Line-to-neutral and source impedance The VRMS (root mean square) value of the L-N voltage is the first measurement to take. If it's too low, that could be an immediate indication of excessive source impedance. Source impedance is simply the total impedance looking back from the load to the source. There are two major contributors to this source impedance. One is the wiring; the longer the conductor and the smaller the diameter (higher gauge), the higher the impedance. The other factor is the internal impedance of the transformer (or other source equipment). This internal impedance is a way of saying that a transformer of a given size/rating can only supply so much current. One of the most basic principles in "Performance Wiring," i.e., wiring that exceeds Code minimum requirements and helps assure power quality, is to minimize source impedance. In other words, assure adequate transformer capacity and install relatively short wire runs . Also, just so we don't overlook the obvious, when you eventually get to the panel, make sure the connections are torqued to spec but not over-torqued; loose connections will always be the No. 1 suspect for excessive source impedance.

Many electricians think more in terms of IR drop. Source impedance can be thought of as the "R" in IR drop (or, strictly speaking, the Z in IZ drop). Any IR drop is subtracted from the L-N voltage available to the load. High-source impedance increases the potential for significant IR drops, possibly exceeding the NEC- recommended 5% maximum.

Peak voltage and flat-topping We can't stop at the VRMS measurement of L-N voltage. Just as important is the V-peak value. We are trying to see if the voltage waveform suffers from excessive distortion, which at the branch circuit takes a characteristic form called flat-topping. A scope display will immediately show if the waveform is flat-topped. If a scope is not available, a multimeter with the ability to read the peak value of the wave (sometimes called the Peak MIN/MAX function) can be used. The ratio of VPEAK to VRMS is called the Crest Factor. For an ideal, pure sinewave, the Crest Factor is 1.414, but the flat-topped voltage waveform typically found in many commercial buildings will have a lower value.

If we understand what causes flat-topped voltage, we'll also see why it can be a problem. The first stage of the power supply in electronic equipment has a diode bridge that turns ac into pulsating dc, which then charges a capacitor. As the load draws the cap down, the cap recharges. However, the cap only takes power from the peak of the pulsating dc wave to replenish itself because that's the only time the supplied voltage is higher than its own voltage. The cap ends up drawing current in pulses at each half-cycle peak of the supplied voltage. This is happening with virtually all the electronic loads on the circuit. The voltage peak tends to get dragged down as a result-in other words, flat-topped.

The greater the source impedance, the greater this voltage distortion. The source impedance is naturally greatest at the end of a branch circuit, the farthest point from the source. That's the same place where all those electronic loads are demanding current at the peak of the wave.

Intermittent sags

Low L-N VRMS combined with flat-topping is a deadly duo. It prevents caps from getting charged to their designed capacity. But so far our measurements have only been static. What happens when we add the dimension of time? Now the possibility for problems increases. When an intermittent low-voltage event, a sag, comes along, the ride-through capability of the power supply is diminished. What this means to the operator is unpredictable resets, crashes, lock-ups, etc.

Sags are common events at the receptacle. Sags can be caused by other branch circuit loads. Laser printers and copiers are notorious for cyclical sags. They are good candidates for their own dedicated circuits. Even a fan motor starting up on a circuit heavily loaded with electronic equipment might cause enough IR drop to crash one of its sensitive neighbor loads. Sags could also originate "upstream" on the three-phase feeder circuit.

Three-phase monitoring equipment, whether portable or installed, is not designed to isolate events on the branch circuit level. However, low-cost palm-sized recorder modules are now available to insert into the receptacle. A troubleshooter can thus see the same voltage the sensitive electronic load sees. One model has user-settable thresholds and can capture up to 4000 events. Spreadsheet and graphical software display the results, so the field technician is spared the tedium of scanning rolls of paper.

Transients

The next step on the time dimension is the realm of very fast events: transients that occur on the sub-cyclical time scale. Transients are caused by load switching and outside events like lightning. Transients at the receptacle are generally not the high-energy type that service entrance, panel, and outdoor surge arresters are designed to protect against. Nevertheless, these relatively low-energy transients can cause damage to solid-state devices by, for example, puncturing semiconductor junctions. They are also a potential source of noise because their fast rise times facilita te coupling into signal bearing circuits.

A scope will determine if transients are present. If all you have is an analog scope, you can try to set the trigger level above the power-line voltage signal, tweak the time base (fast, faster, fastest), and, under cover of darkness, look for blips. If you have a digital scope, which has waveform storage, you can use the single-shot function: set the trigger level above the wave and catch transients one at a time. Both the analog and traditional digital scope have a limitation; they only catch so-called impulsive and possibly oscillatory transients, the kind that spike out from the wave, and only if they exceed the peak. Notches, however, cannot be captured. These notches can be a problem if they cross the zero line of the sine wave because they then confuse microprocessor clocks and generator frequency controls, which rely on counting the zero crossing point.

Neutral-to-ground

N-G voltage may seem mysterious, but let's try to demystify it by saying it's a handy measurement for two things-circuit loading and some common grounding errors. First, let's consider circuit loading. At the receptacle, it's not easy to tell if a circuit is heavily loaded. Unlike at the panel, there is no convenient access to circuit conductors to measure current. Also, we often don't know the voltage at the panel breaker, so we can't compare the L-N voltage at the receptacle with the L-N at the breaker. But N-G voltage can give us a quick "go/no-go" reading on circuit loading (Fig. 1).

Circuit loading takes time to learn, but it is simple, and worth the time to learn. Imagine the simplest branch circuit, a dedicated circuit with L-N-G to a single outlet.

When loads are ON, an IR drop occurs on both the hot wire and the neutral wire, and is roughly equal on each wire. The total IR drop is subtracted from the voltage available to the load. For a given value of R (length and gauge of wire), the greater the current (I), the greater the IR drop. Ohm's Law is still the law of the land.

Now let's make a big initial assumption, that the Ground connection is correct, namely that it goes back to the source and connects to the Neutral there, and that there are no sub-panel or receptacle N-G bonds in between. In this case, there is no or minimal ground current, and the ground connection looks like a big test lead extension back to the N-G connection.

If I now measure the N-G voltage at the receptacle, using this long test lead extension, my meter is looking at the IR drop on the neutral conductor. Under normal load conditions, if it's a couple of volts, you're in good shape with this branch circuit. If it's high, let's say in the 5-V and over range, you will want to refrain from adding additional load or better yet, cut down the load on the circuit. Remember, the IR drop on the neutral (N-G voltage) adds to the IR drop on the hot conductor as well, and the combined IR drop subtracts from the L-N voltage.

Shared branch neutrals

The predominance of electronic loads, each generating third-harmonic currents, which add up on the neutral, has changed the situation. Third-harmonic currents result from the pulsed, "peaky" current draw of the capacitor input circuits, the same ones causing voltage flat-topping. In the above scenario of 10 A per phase, you could easily have 20 A or more on the neutral. This is 180-Hz current, with higher heating effects than the equivalent current at 60 Hz. First and foremost, this creates a fire hazard because there is no circuit breaker to protect the neutral. Also the L-N voltage available to all three branch circuits is reduced, and load stability could suffer. So if N-G voltage is high, suspect shared neutrals. Proceed to the panelboard, count the number of hot conductors and the number of neutrals, looking for fewer neutrals than hot conductors. Measure neutral currents. Maybe in the short run, you can shift loads to minimize neutral overloading, but ultimately, you'll have to consider rewiring. Shared neutrals, with today's electronics, are an accident waiting to happen. A high N-G voltage measured at the receptacle could also be due to an overloaded feeder neutral in the panel. Maybe the neutral cable is undersized. Third harmonics from all three phases obviously add up in the feeder neutral as well as in a shared branch neutral. The 1996 NEC for the first time required a feeder neutral equal in size to the phase conductor, but the consensus from the power-quality community is to double the size of the neutral. The point is that a high N-G voltage is like a fever: it warns you of symptoms.

Neutral-ground bonds What if N-G voltage is near zero? With normal circuit loading, there should always be a couple of volts or so of N-G voltage. A reading near zero indicates the likely presence of an illegal N-G bond. This bond could exist in the sub-panel, or it could be the result of miswired receptacles (N-G switches, or stray copper strands making contact). These illegal N-G bonds are common and should be removed. There is also the rarer occasion when N-G voltage is extremely high and unstable. This could indicate a high-impedance loose N-G connection. Usually, this will cause system-wide problems showing up as problems on many loads on different circuits.