During a “PQ Primer” seminar I recently presented, we discussed how photovoltaic inverters synthesize the fundamental power frequency when there is no voltage reference from the grid. For years, time was told by counting the zero crossings of the 60 Hertz (Hz) waveform. The grid’s massive rotating generators’ inertia kept a fairly constant time. Operators would have to occasionally change the generators’ frequency in response to when they were altered by large load swings to get time back on track. But 60 Hz was a generally reliable basis for keeping time.
Plenty of equipment used this signal to synchronize processes, though they may use dividers or multipliers for values other than
60 Hz. When distortion or transients caused multiple zero crossings of the signal, things could run fast if proper filtering techniques weren’t used. When equipment was turned on, the actual time needed to be re-entered in the system, which can be inaccurate. Battery-backed real-time clock (RTC) integrated circuits then became the source of time, even when the equipment was turned off. In particular, a 32.768 kHz crystal was used in an oscillator circuit for an RTC, which typically had an accuracy of 2 seconds a week, depending on temperature.
Power quality monitors were among the equipment using the power-line frequency and the RTC for back-up time. As the requirements to correlate disturbances and steady-state data over fixed intervals across a larger geographic area to be consistent between multiple units, other means to synchronize the time became necessary. Utility substations often had IRIG-B time sources. Once local and through the internet equipment networking became common, a technique called network time protocol (NTP) was the method used where possible. This brought the accuracy for the synchronization of clocks between multiple units down to one cycle or better. But the latency from the propagation delay between the time source and the end-user had to be considered for system-wide accuracy over a wide geographic area to hold.
Over time, this accuracy was not enough for some systems, especially as a new tool was being deployed across substations. The synchrophasor or phasor measurement unit went from a research project to thousands deployed in the past decade. These instruments need the time stamps across the system to be within 1 microsecond. Phasors are a vectoral representation of the magnitude of a sinusoidal signal (such as the voltage) and the phase angle relationship between the phases (typically 120 degrees in our 3-phase power grid). The phasors rotate 60 times a cycle, but they can vary that rate during system disturbances. In fact, changes in the phasor data are often seen tens of minutes before large-scale system instability that can end up in a cascading blackout. Since the rate of change of phase is frequency, monitoring frequency can give a warning a minute or less before it is seen in the voltage magnitude. By the time the root mean square voltage is changing, things are generally going downhill rapidly without quick intervention.
A search in YouTube for “Florida 2008 Blackout” turns up a presentation showing an amazing graphical display of the changing phase angles across the entire Eastern United States due to the blackout in Miami.
A GPS time source from satellites is used to get this high-accuracy time values in geographically dispersed locations. Typically, three GPS satellites are needed to ensure accuracy. The clocks in the satellites are accurate to 3 nanoseconds. The systems need to compensate for propagation delays based on the distance from the satellites, but it is constant and much more predictable compared to NTP because of the signals’ fixed route.
PQ instruments that are Class A compliant with IEC 61000-4-30 and tested using IEC 62586-2 methodologies are required to have an accuracy of 1 cycle or better “regardless of the total time interval.” Most instruments today use GPS to synchronize their clocks; however, the standard allows for “a synchronization procedure applied periodically during a measurement campaign, through a GPS receiver, reception of transmitted radio timing signals or by using network timing signals.” In addition, the 10-minute intervals used to aggregate data must have those precise time stamps. Since the power frequency can deviate during that time window, it is not an exact number of power frequency cycles. For example, if perfect, 60 cycles/second × 60 sec/min × 10 min intervals = 36,000 cycles. But if frequency is 59.995 Hz over the interval, there will be three fewer cycles in that 10-minute interval. The instruments need to adjust for this by adding or subtracting cycles as needed.
System-based PQ monitors need to know the time precisely across the entire system to correlate data and disturbance events to determine the cause and source and help prevent re-occurrences.