In my August column, I reviewed some basic rules and procedures for troubleshooting power quality problems. Like water that flows downhill, PQ disturbances must obey the laws of physics. When reviewing data from customers who claim their problem defied these laws, it has been either user misunderstanding, incorrect instrument setup, instrument malfunction, missing or overlooked data or a combination of these.
It is important not to overlook any plausible cause of a problem, despite information provided by the customer. Let’s take the case of the phone system crashing and not restoring for some time, causing lost revenue at a sales call center for a well-known office supply company.
The office space was in a commercial building, occupying about 25% of the building’s third floor. The facility electrician stated that no other tenants had any power issues.
The electrician hired by the office supply company to investigate concluded that the power cable runs to the office area from the distribution panel were too long, causing a low-voltage condition. They said it being summertime when the problem started put an extra strain on the electric utility supply. After the expense of running new cables from the distribution panel to their electrical panel, the problem returned in a week or two. No one had yet observed that the problem only occurred on sunny Friday afternoons.
Putting a PQ monitor on the circuit that fed the PBX, at the receptacle for the PBX and another “unaffected” receptacle in another part of the office, it didn’t take long to see the likely cause. The voltage at the panel and the PBX would go into an interruption instantly, no slow decay and no huge current increase that would have resulted in a overcurrent breaker trip. The monitor at the unaffected receptacle did not detect any voltage sag that would typically occur on another circuit fed from the same panel.
The conclusion was that someone wanted to get an early start on the weekend, and that person would go into the service closet and manually trip the breaker for several minutes. The reboot cycle of the PBX was long. Rather than putting a hidden camera in to catch the PQ culprit, they just locked the PBX closet door with a key that only the manager had, and the problem ceased.
When the voltage sags, the current response depends on the type of load that is powered from the source. If monitoring at the service entrance or at a distribution point with multiple types of loads, it can be difficult to understand the cause and effect. When monitored at the load, the response should be one of these:
The current drops in proportion if the load is a linear one, such as a resistive-type used in electric heaters.
The current will increase if the load is a constant-power device, such as most applications of electric motors with constant mechanical loads (and not used in adjustable-speed drive configurations).
The current will decrease to nearly zero, then increase slowly and eventually exceed the original levels, if it is a typical rectified input, switching power supply used in computers and other IT equipment.
If you are monitoring the power at the point of common coupling in a building with mixed loads, and the voltage decreases to zero for a half or even several cycles without any change in the current, you need to recheck where the voltage leads and current probes are connected. It is very unlikely that everything is correct. Similarly, if the load current goes up significantly, such as 6–10 times from a large motor starting up, and there is no change in the voltage, something is very likely askew.
The same rule applies to voltage transients. An example can be found in the figure above (from a 25-year-old recording), which shows the voltage waveform. The current waveform had no disturbances. The reason that the instrument recorded such waveforms was that the alligator clip lead for the voltage recording was placed on a heavily oxidized bolt. When a large truck would rumble past the building, vibrations would cause the clip lead not to make electrical contact, sort of like an arcing signal in reverse.
If you do record data that wants to dispute the indisputable rules, be sure that everything is set up and working properly. While it may be possible to expand the laws of physics and become a Nobel Prize recipient, the odds are extremely slim and the cause is likely something else.