|
|
Monitoring the Health of Your Electric Power Grid Precision PC-Based 60-Hz Frequency Counter Note: a version of this article was published by Circuit Cellar magazine. The published version is on Circuit Cellar's website. By Doug Criner Would you like to know in advance if the electric power grid is on the verge of collapse? An important indicator of the health of the power grid is the system frequency–nominally 60 Hz in the U.S., but seldom exactly at 60 Hz. In the U.S. there are three separate grids: one for the West, one for the areas east of the Rocky Mountains, and one for State of Texas (or the Republic of Texas or the State of Mind of Texas). Within each of these areas, all the generators are synchronized and tied together in parallel. System operators try to keep the frequency at 60 Hz by increasing generation if the frequency drops, and vice versa. Frequency and Time Errors There are two system errors that need to be minimized: the frequency error and the time error. The frequency error is just the instantaneous deviation from 60 Hz. The time error, the integral of the frequency error, is how late or early electric clocks are running relative to true time. System operators (and power system computers) are continually jiggering the system frequency to minimize these two errors. To monitor the system frequency, we must be able to measure it within about 0.01 Hz. If, for example, the system frequency suddenly drops and then recovers over several minutes, it often means that a large power plant has tripped off-line and the load is being picked up by the other generators. The trip may have occurred in New York, but I might detect the frequency change in Chicago. A Struggling Grid If the frequency is running slow and the time is running late, then it may indicate that the system is struggling. A sudden increase in frequency can occur if a large area is blacked out, either intentionally or by accident. Although the entire grid runs at the same frequency, there is a phase angle between areas that are exchanging power. For example, if net power is being sent to Chicago from the East Coast, then the voltage phase of Chicago will be lagging slightly. A sudden change in power exchange between areas will show up as a change in the phase angle, which will be accompanied by an apparent frequency spike or dip. None of these events can be detected by simply monitoring the system voltage–which can be affected by any number of factors. Measuring system frequency to within 0.01 Hz is not trivial and we must count a significant number of cycles to get meaningful results. Most commercial frequency counters lose precision at such low frequencies. Luckily, we are interested in frequency changes over a minute or so. Project Takes Only Three Parts This project requires only three parts (five, if you count the PC and a serial cable). The main component is a pulse counter/timer module available for about $75 from Weeder Technologies. This little board, which uses a PIC microcontroller, measures frequency by counting cycles over a longer interval for low-frequency signals, and thus can measure a 60-Hz signal to 0.001 Hz, which is well into the jitter band for our application. The other parts for the project are two wall warts: an AC-AC transformer that outputs about 5-15 volts for the 60-Hz signal to measure and an AC-DC wart that outputs about 10-12 volts DC to power the Weeder board. (The Weeder manual indicates that the board’s signal inputs are "protected" up to 160 VAC, but please do not use 120 VAC for the signal input! Use only a low-voltage, UL-listed wall-wart transformer.) Basic Program The Weeder board has a standard DB9 socket to connect to the serial port of a PC. The easiest way to communicate with the Weeder board is to use Microsoft QuickBasic. This program, called QBASIC.EXE, runs from DOS, but unfortunately doesn’t come loaded with recent versions of Windows. If necessary, transfer it from an older machine. Listing 1 shows a demo program that will display the current frequency and time error. It also will beep if frequency drops to 59.95 Hz. You may want to write your own program to continuously plot frequency and time errors, store data to disk if unusual events occur, and perhaps import data into a spreadsheet for analysis. The Weeder board uses a garden-variety 20-MHz crystal oscillator for counting. The accuracy may be off a little, but if the board is at a reasonably constant room temperature, the crystal should be sufficiently consistent, particularly since we are most interested in short-term frequency changes. You can derive the necessary "bugger factor" that will achieve a zero average time error over several days. Parts List Pulse counter/timer module, p/n WPTCT-M, Weeder Technologies, phone (850) 863-5723, www.weedtech.com. Wall power supply, 12 VDC, p/n 163491, Jameco Electronics, phone (800) 831-4242, www.jameco.com. (Any 9-18 VDC, minimum 50 mA, UL-listed wall wart may be substituted. Check polarity before connecting to the Weeder board.) Wall transformer, 9 VAC, p/n 184751, Jameco Electronics. (Any 5-15 VAC, UL-listed wall wart may be substituted.) DB9 serial cable, 6 ft, male-female, p/n 25700, Jameco Electronics. (Or purchase locally.)
Program Listing 1 - FREQ.BAS
July 2001
|