Engineers in a large electric utility’s power quality (PQ) department had installed PQ monitoring to determine the source of a system component’s failures. They needed to see the voltage and current waveforms before, during and after the cap banks were switched in and out. The engineers had deployed the PQ system in thousands of locations, but early commission test results were not what they wanted. Thousands of events were being captured. In between those undesired events were the waveforms that they wanted, but not always. They also saw other unexplained data during the system overload.


In a teleconference between the engineers and the PQ instrument vendor, it was obvious both sides would benefit from understanding the other’s perspective. With most PQ instruments, their primary triggering mechanism to capture data is based on voltage variations, particularly the rms value of each cycle in half-cycle steps, per the IEEE 1159 and IEC 61000-4-30 standards. 


Above the high limit is a swell. Below the low limit is a sag (or dip). If it goes too low, it’s an interruption. But the voltage’s rms didn’t change significantly during the event that the utility wanted to capture. The voltage wave shape itself changed during the switching operation, which could be captured by a voltage transient triggering algorithm (Figure 1). However, the voltage wave shape had almost no change when the component was switched out—an event the utility also wanted to record.


Another voltage signal is available from the voltage supplied to switch itself. However, it is not present until the instant it is energized. PQ instruments designed to meet these standards require a phase-locked-loop (PLL) to synchronize the voltage sampling precisely x samples per cycle. As the fundamental frequency changes, so must the PLL to keep the sampling at the same waveform point. When there is no voltage, there is no frequency for the PLL to track, and it takes some time to acquire the signal when it first starts up. So this trigger mechanism also wasn’t going to work.


The key was to trigger on the current, as that changed when the cap bank was switched in and out. But it didn’t only turn on and off. It could step up, then up to another level, then back down, maybe back up, then back to zero. With multiple levels and unpredictable change pattern, a simple high/low threshold of the rms current wasn’t going to work. So, the engineers used a transient trigger method to check how the previous current cycle changed from the previous, referred to as wave shape triggering.


This seemed like a reasonable approach and worked in the lab. However, in the real world, the current wave shape is always changing as loads turn on and off. The net result was “billions and billions” of events, when they needed just a few per day.


Instead, the engineers used a current transient triggering method called “rms deviation.” It looks for a change in the rms value of consecutive cycles to change more than x percent. In this application, the current would double with each step up, or drop in half or more with each switch-out. With this trigger mechanism, the desired data was captured.


However, several things appeared in the waveforms that seemed unusual from the PQ instrument engineer’s perspective. Six cycles after the initial transient capture was another set of transients, much smaller but always detectable in the current and barely detectable in voltage (Figure 2). Restrike events typically don’t occur that far out in time.


In addition, closer examination of the voltage transients in the initial event showed a very high order harmonic ringing or oscillatory signal on each phase as the component was switched in. Normally, these ringing signals have a frequency in the 400–1,600 hertz (Hz) range, not more than 2,400 Hz, as seen in the harmonic spectrum graph of Figure 3.


The meeting shed light on both of these seemingly anomalous events, which the engineers deemed quite normal. Their switching system uses a two-stage mechanism to minimize inrush current. The second stage occurs approximately six cycles later. With their system impedance, higher order harmonic resonance is not uncommon.


To make this PQ monitoring application work, both the engineers and vendor had to learn the application, establish the desired information, understand how the PQ instrument works and discover how to best set it up to capture only that desired information. And they did.