Consider other key factors when diagnosing:
If you want to take a good picture of the moon’s craters, it is best to do it when the moon is only partially lit by the sun, rather than when it is full. This seems counterintuitive, as one would think that the brightest moon would make the best picture subject. But when the sun is at an angle to the moon, rather than straight on, the contrast of the shadows makes the craters look much clearer. So what does this have to do with power quality, you ask?
It is important to know the phase angle on the voltage sine wave that a power quality disturbance originates. Most people who deal with power quality-related problems are aware of other key factors, such as what type of disturbance, what time of day, the magnitude and duration, etc. In Figure 1, it appears that the sags are occurring most workday mornings, either two or three times, with about the same magnitude and duration. This might lead to the conclusion that the source would be some sort of load that turns on at those times. Closer examination shows that they don’t occur every day, and the times are not precisely the same, which would tend to discredit any timed-based type of load turning on.
The key to explaining the differences and providing a better clue as to the type of load causing these sags—which were measured at a 120V duplex outlet in an office environment—is seen in the waveform cycles of the voltage sags in Figures 2 through 4. The sag can originate at a multitude of phase angles, both in the positive and negative halves of the waveform. Figure 2 shows the sag beginning with a 70V negative unipolar transient (one that subtracts energy from the curve) and then the voltage returns to a sinusoidal waveshape but at a reduced voltage amplitude (101Vrms) that recovers back to near the original amplitude in approximately 9 to 10 cycles (see Figure 5). Figure 3 is a similar sag at a different phase angle, but originating with a more severe 120V negative unipolar transient. And lastly, the third sag in Figure 4 has a similar rms plot, though not quite as severe (106Vrms), which was just enough to cross the 90 percent of nominal threshold for a sag with no transient.
If the same load is turning on, why the differences in the magnitude of the sag and transient? The load itself holds the key. The given that the sag occurs nearly every workday morning but at slightly different times would eliminate any utility-based operation from being the source. A power factor correction capacitor switching in to handle the increased loads in the daytime would most likely do it much earlier in the day, and it would not happen multiple times. In addition, the sharpness of the transient (fast rise time) would point to the source of the problem being relatively close to the monitoring point, rather than the half-mile-away PF cap bank. In addition, there is not the typical oscillatory transient that occurs when the capacitance and inductance of the distribution system react to the sudden energization of the cap bank, nor do they normally cause sags.
In order to get such an abrupt change in the voltage, the problem is usually something that can draw a lot of current rapidly, resulting in a sudden voltage drop in the source impedance, leaving less voltage left for the loads, aka, a sag. A capacitor does exactly that, as it does not like to have its voltage change instantaneously. When trying to take the 0 volt uncharged capacitor and connect it to the 120Vrms circuit will cause a current flow that is proportional to the change in voltage across the terminals of the capacitor.
At 90 and 270 degrees on the sine wave, the voltage is at its 170V peak, or the approximate square root of 2 (1.414) times the rms voltage (120vrms) in a low distorted system. At phase angles other than these, the voltage is equal to the peak voltage times the sine of the phase angle, which goes from 0 at 0 degrees to 1 at 90 degrees, then back to 0 at 180 degrees and so on. In Figure 3, the 260-degree phase position of the origin of the sag would be a voltage of approximately –146Vpk, whereas it would be approximately 119Vpk for Figure 2. Hence, the larger step change in the voltage that the capacitor experiences, the larger the transient. Once the capacitor is charged, then the normal current draw of the load becomes the dominate factor in determining the amplitude of the sag.
So why is there no transient in Figure 4? It appears that the random turn-on point of the load coincides with a phase position of near zero degrees. At that point, there is no voltage on the circuit and no voltage on the uncharged capacitor. Hence, there is no difference between the two, and there is no sudden current flow to charge such rapidly. Instead, it just charges as the voltage of the 60Hz sine wave increases. If the nominal voltage had been slightly higher, the load turning on at 0 degrees would not even cause an rms variation that would be classified as a sag, as appears to be the case on Tuesday morning.
Hence, the phase angle that the load turns on has a significant difference in the picture of the sag, just as it does in the picture of the moon. ECBINGHAM, a contributing editor for power quality, can be reached at 732.287.3680.