548

IEEE Transactions on Power Delivery, Vol. 5, No. 2, April 1990

A MAGNETO-OPTIC CURRENT TRANSDUCER T. W. Cease Member Tennessee Valley Authority Chattanooga, Tennessee Abstract current measurement technique using the magneto-optic or Faraday effect has been developed, tested and demonstrated. The device developed is a Magneto-optic Current Transducer (MOCT). It uses the rotation of the plane of polarization by a magnetic field exhibited in certain glasses (Faraday effect). This paper deals with the comparison of test data obtained in an extensive field evaluation of the MOCT with the design qualification data gathered during the development phase of this project. The analysis indicates that operation of the MOCT based metering system under field conditions compares favorably with performance predicted during design qualification. Introduction Conventional oil-filled current transformers (CT) used on extra high voltage (EHV) and ultra high voltage (UHV) circuits by the utility industry have a number of short comings that have been designed around over the years. These include: Potential for catastrophic failure Saturation under fault current conditions Ferroresonance effects Hysteresis effects High voltage if output leads are opened Large/Heavy requiring substantial supports Accuracy limited to 0.3 percent Subject to EM1 effects PCBs This paper presents the result of work performed developing an optical current transducer that makes use of the magneto-optic effect. The benefits of an optical current transducer are the inverse of the oil-filled CT's problems. That is: * Does not contain oil, therefore can not explode * No saturation under fault current conditions * Does not contain iron, therefore no ferroresonance effects * No hysteresis effects * Output is a fiber optic cable (no danger of open circuit) Paul Johnston Westinghouse Electric Corporation Raleigh, North Carolina

* * *

Small/light weight (sensor weighs less than 2 pounds) Accuracy limited by electronics Immunity to EM1 No PCBs

Principal of Operation This new current transducer uses the magneto-optic or Faraday effect. "In 1845 Michael Faraday discovered that when a block of glass is subjected to a strong magnetic field, it becomes optically active. When plane-polarized light is sent through glass in a direction parallel to the applied magnetic field, the plane of vibration is rotated. Since Faraday's early discovery the phenomenon has been observed in many solids, liquids, and gasses. The amount of rotation observed for any given substance is found by experiment to be proportional to the field strength B and to the distance the light travels through the medium. This rotation can be expressed by the relation
J=VBL

where B is the magnetic induction in teslas, L is the length of the light path in meters, theta is the angle of rotation in minutes of arc, and V a constant to be associated with each substance. This constant, called the Verdet constant, is defined as the rotation per unit path per unit field strength." [l] For a more detailed description of the Faraday effect see reference 2. ODeration To make use of this effect, a Magneto-optic Current Transducer (MOCT) was built, as shown schematically in Figure 1.

LM

FUNCTIONAL BLOCK DIAGRAM Figure 1 Light emitted by the light emitting diode (LED) is transported by an optical fiber to the polarizer, after being polarized the light is injected into the Faraday rotator material, and exits through an analyzer. The analyzer is a polarizer rotated 4 5 degrees from the first polarizer. The net result is that an intensity modulated light beam exits

8 9 SM 732-9 PWRD A p a p e r recommended a n d a p p r o v e d by t h e IEEE Power S y s t e m R e l a y i n g Committee of t h e IEEE Power E n g i n e e r i n g S o c i e t y f o r p r e s e n t a t i o n a t t h e IEEE /PES 1989 Summer M e e t i n g , Long Beach, C a l i f o r n i a , July 9 1 4 , 1989. M a n u s c r i p t s u b m i t t e d December 5, 1 9 8 8 ; made a v a i l a b l e f o r p r i n t i n g May 9 , 1989.

0885-8977/90/0100-0j4s$o1.oo0 1990 IEEE

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

549

the analyzer. This intensity modulated light is then transported through another optical fiber to a PIN diode for demodulation and results in an output of the amplifier of a voltage signal that is proportion to the current flowing in the conductor. This output is a reproduction of the current flowing in the "primary" circuit of the MOCT, limited in the practical sense only by the ability of the electronics to reproduce amplitude and frequency information contained in the optical signals from the rotator. An example of the signal output for a primary current of 2000 amperes is shown in Figure 2.

evaluate the performance of the integrated system of current sensor and electronic meter, verify the insulation system through high voltage testing, and demonstrate the performance and stability through a field test of the MOCT based meter for a period of one year. The verification of the insulation system was carried out in accordance with the test sequence contained in ANSI C57 for instrument current transformers of 750 kV BIL rating. Since the insulator system for the metering system is a standard design, the results of

MOCT OUTPUT VOLTAGE AT 2000 AMPERES Figure 2 The Faraday rotator portion of the MOCT was built, using optical quality quartz glass as the rotator It consists of the rotator itself with the associated optical components: collimating lens, polarizer and analyzer. Dimensionally, the rotator was sized so that a coDDer bar, rated for 2000 -amperes would pass through the opening. Figure 3 shows the rotator assembly mounted on the copper bar with a polycarbonate cover and the terminals for line connections. This assembly was designed for a short term mechanical withstand of 8 0 k A .

Direct measurement of these variables were made; however, it was recognized at the beginning Of the project that measurements like these were difficult when high precision over a wide dynamic range were required. As a result, an approach to testing the device was taken which permitted ratio measurements between the unit under test and secondary standards with known accuracy. Testing was performed on the MOCT to determine the characteristics of the rotator. For these tests four precision current transformers were used with ranges of 20A (1% of full scale), 200A (10% of full scale), 2000A (100% of full scale), and 4000A (200% of full scale) to provide points on the calibration curve. The MOCT was compared against these CTs to determine the accuracy of the rotator and associated electronics. This arrangement also permitted a direct comparison of the phase response of the MOCT and the known phase characteristics of the precision CT's. This comparison indicated that the phase response is constant over the current range of interest; but the precision of the measurement was poor. To overcome this limitation, the MOCT was configured as a current sensor in a watthour meter and the characteristics of the system compared to the performance of a watthour standard. While this does not permit direct measurement of the output of the current sensor, the resulting measurement system provides + 0.01%, with excellent accuracies of repeatability and Tong term stability of the reference. In addition, this measurement approach was identical to the technique that was applied to verify operation in the field, thereby improving the ability to correlate results.

MOCT CURRENT SENSOR ASSEMBLY Figure 3 Testing The rotator assembly was integrated with the electronics necessary to produce the desired output and testing performed to determine the performance characteristics of the MOCT. This testing program was divided into four phases which served to characterize the optical system and signal processing circuits,

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

550

The configuration of the test system is shown in block diagram form in Figure 4. The equipment configuration is a current source, adjustable from zero to 4000 amperes, whose output is passed through a precision CT and the MOCT rotator. The output of these two sensors are measured by an instrument package which computes the ratio of the MOCT output with respect to the ct output. This method permits comparison of ratio stability of the MOCT against a known standard. In addition, the ct output is applied to a watthour standard while the MOCT output operates the current section of an electronic meter. Application of the current to both the meter under test and the reference standard along with parallel application of a voltage source provides an accurate method of comparing the two systems in a ratiometric manner. TYPICAL TEST TEMPERATURE PROFILE Figure 5

MOCT METERING SYSTEM

"LA
LAB TEST EOUIPMENT

A typical temperature profile for a test run is shown in Figure 5. This is considered to be an extreme worst case test condition as it stresses the MOCT system by changing the temperature by 225 degrees F in a 2 4 hour period, approximately 10 degrees per hour. Examination of the temperature profiles taken during the field test show this to be a good assumption for the Tennessee area, where changes in temperature were typically 20 to 30 degrees F in a 24 hour period. Figure 6 is the variation in ratio performance of the MOCT system under the above test condition. Under this extreme test, the device will exhibit no more than 0.1% error from -3OOC (-220F) to +60oC (+1400F) when compared to the reference CT operating at room temperature.
MOCT RATIO STABILITY
MOCT PRIMARY CURRENT
2000 AMPS

TEST SYSTEM FUNCTIONAL BLOCK DIAGRAM Figure 4 Temperature Characteristics The first test was to characterize the rotator output with variation in temperature. The rotator and interface electronics were placed in an environmental chamber that allowed temperature profiles to be run on the system while it was operating, and the effect of temperature measured. Two temperature effects were studied. First, the change in the MOCT output that could be related to shifts in the steady state operation due to temperature excursions possible in field conditions. This set the extremes of the temperature range. Second, it was known that temperature induced stress on the optical system could result in short term changes in the output levels: therefore, a study of the effect of rate of change in temperature was important to characterize the system.

*)

@ J

1015
l o 2 1

1011005

1 -

.
,

0 0995-

z
0 k a a

0990925

0984

These test were run for each of the precision current transformer ranges with similar results for all ranges. Each test in each range was repeated several time with only slight variations in results from test to test.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

551

System Results In order to perform long term field tests of the MOCT, a solid state metering system under development, was adapted to interface with the output of the MOCT electronics. By driving a metering system the current signal of the MOCT could be integrated over time to allow easy determination of the long term accuracy and stability of the MOCT in a field test against a reference system and the MOCT as well as the reference system could be rechecked to establish if there were any drifts developing during the field test period. Testing was performed on the system (rotator and solid state meterins electronics) using a watt-hour standard. Load curves were run for a 1.0 power factor (PF), Figure 7,

MOCT METERING SYSTEM

0.5 POWER F A C T O R
4

-3

- 2 / , -4

,
20

,
40

,
60

,
80

C U R R E N T (PERCENT OF C L A S S ) C L A S S .5 L I M I T S 0 D A T A POINTS

,I

100

Figure
MOCT METERING S Y S T E M
1 .O POWER F A C T O R

1
-3
~

1
,
100

,
20

,
40

,
60

,
80

- C L A S S .5 L I M I T S

C U R R E N T ( P E R C E N T OF C L A S S ) 0 D A T A POINTS

The load curve at 0.5 lagging power factor provides an indirect, but sensitive, measure of the phase angle reproduction capabilities of the system. The sensitivity arises from the error term that is present if there is a phase difference between the current signal seen by the meter under test and the signal applied to the reference standard. As an example, at 0.5 power factor, a phase error of 0.1 degree would result in a registration error of approximately 0.3%. Examination of the load curve shows that any error is negligible, when compared to the reference performance of the system at unity power factor. Using the load curve approach to measuring phase angle reproduction also illustrates that the phase error is constant over the range of currents to be measured by the system. Testing was also performed holding the current constant and varying the temperature to +6OoC. The results, shown in from -30% Figure 9, show that the MOCT and metering system had an accuracy better than 0.5 percent over the entire range. This particular test was conducted at the light load (10% of rated current) current of 200A, making it a severe test of the response of the meter system due to the poorer signal to noise situation under this test condition.
M O C T Light Load Temp. Performance
to51387 1.009
1.008

Figure 7 and a 0 . 5 PF (lagging), Figure 8 . These data points are plotted for measurements taken on the system for currents ranging from 20 A to 4000 A , covering a dynamic range of 1% of rated current to 200% of rated current. The specification called for the measurement accuracy of the current transducer and the meter to fall within an error bound of & 0.5% of reading over the above current range. This specification error bound is superimposed on the load curves. The figure for the error bound in the system specification was developed by combining an error budget for the meter and an error budget for the current transducer into a total allowed error for the system. Metering class CT's permit by specification, an allowance for 0.3% ratio error at 100% of rated current and 0.6% ratio error at 10% of rated current. Taking this specification and making it more stringent by requiring the dynamic range to be extended from 1% to 200%, while requiring the ratio error to be constrained to be no more than 0 . 3 % over this current range was considered to be an aggressive objective for a new design. Assigning this figure for the current transducer error budget left an allowed maximum error of 0.2% for the meter portion of the system.

1.007
1.005

5
e

1.004 1.003 1.002

1.001

1 0.999 0.998 0.997 0.996

0.993 0.992 0.991 0.99 -40

1
-20 20
40

60

T E M P (aC)

Figure 9

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

552

Location In order to be able to install the MOCT in an operational switchyard, a porcelain insulator (originally designed for a capacitive coupled voltage transformer) was used to support the rotator and provide an insulation medium. The electronic were located in the base housing, as shown in Figure 10. The fibers to transport the light were passed through the center of the hollow insulator, and the insulator filled with epoxy. The assembled MOCT was tested in a high voltage laboratory for the ANSI/IEEE C57.13 specified tests for current transformers, including a 750 kV BIL test. The MOCT system was then connected in one phase of a 161 kV air-insulated transmission line. It was connected to monitor the current flow in the "A" phase of the transmission line.

The MOCT sends data over an optical link to an equipment panel in the switch house. The data that is sent consist of a KYZ pulse train (each pulse represents a specific number of watt-hours or energy) which is input to a solid state register for storage and display. Also, the register calculates demand (power) from the pulse train. Three other quantities are also sent optically to the equipment panel and they are: (1) the bus voltage (from the VT signal) , (2) the current in the line being monitored (from the MOCT), and (3) a voltage proportion to the optical losses in the MOCT light path (includes the rotator). These optical signals are converted to voltage signals and are available at terminals on the equipment panel. These three signals were included for diagnostic purposes in case problems arose. To collect the data a computer has been configured to communicate with the MOCT register, the MOCT equipment panel, and the reference meter. The computer gathers data every 15 minutes. All data from these devices are collected and stored locally on disk. Also, collected is the ambient temperature in the switchyard and the temperature of the MOCT electronics. Up to one month of data can be stored locally on a hard disk. The computer can also communicate with another computer in the downtown offices and transfer data for long term storage. The computer in the downtown offices can store up to one year of data on disk and can save data on to magnetic tape for permanent storage. By having one year of data available, long term trends can be analyzed as well as daily fluctuations. Data Collected After one year of field operation the MOCT has accumulated a total of 113 ,456.0 megawatthours (MWh) and the reference system has accumulated 113,366.6 MWh. The difference is 89.4 M W h for a percentage difference between the two systems of 0.08%.

Figure 10 Oil Filled CT and Reference Meter A conventional oil-filled metering accuracy CT was connected in the same phase of the transmission line. This CT supplies it's signal to a commercially available, high accuracy, solid-state revenue meter that is equipped with an RS-232 interface. The same VT signal used for the MOCT metering circuit is used as an input to the conventional meter, thus eliminating the VT as a variable in determining relative difference between the two devices. Data Collection System A data acquisition system was installed in the control house and, used to collect data for evaluating the performance of the MOCT. Data is collected from the MOCT and the reference system along with temperature data. The data collected is KWH Total, KW Present Demand, KW Cumulative Demand, KW Peak Demand, and Time of Peak Demand. This data allows a direct comparison between the MOCT metering system and a conventional metering system.

This statistic is certainly meaningful; however, it does not fully indicate the details of the system performance. Another way of describing the operation of the system is illustrated in Figure 11, which details the performance as the difference in percent registration of the MOCT metering system with respect to the reference meter. This allows an examination of the performance with significantly higher resolution. This examination of these data provides insight into the operation of the system and also raises questions that will require further investigation.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

5 . 5 3

MOCT P e r f o r m a n c e
1987.1OB8

n -03-

T I M E (MONTHS

FIELD TEST PERFORMANCE Figure 11 It should be noted that without the data collection system discussed in the previous section, it would be difficult or impossible to reconstruct the conditions affecting the operation of the system. The ability to examine and analyze these real time data has contributed significantly to the advancement of MOCT technology. Returning to the examination of Figure 11, there are several characteristics which should be described for the purpose of clarifying the presentation of the data. First, the percent registration curve describes the percent difference in registration between the reference meter and the MOCT based metering system,
dpercent
=

Figure 11 provides a second piece of performance information when the percent registration curve is displayed with the second curve showing the ambient air temperature at the time of each reading. It is relatively obvious from a comparison of the two curves that there is a correlation between the change in registration and the seasonal changes in temperature. Although the agreement is exceptionally good between the systems, there is a small divergence in the readings as the temperature drops and this is reduced as the higher temperatures return. This indicates that one or more components in the system are responding to a change in the temperature. It does not indicate which is changing, but this can be examined after the test is terminated and the individual components can be measured for their temperature characteristics, as the MOCT based system was examined prior to the start of the test. Conclusions As can be seen the MOCT system compares favorably with the conventional system. In terms of comparison of performance against objectives for the project, the field test has demonstrated the following: * Optical sensors can be designed to perform reliably in operating system environments * Stable and accurate performance in the substation environment is obtainable with proper design * Effects of temperature, humidity, vibration, and electromagnetic interference are negligible or nonexistent when the system is compared to a high accuracy conventional system This test represents a significant program in the development of optical sensor technology for power system application. In the authors' opinion it is the most highly instrumented and closely monitored installation to be reported in the literature. It is also believed to be the first high accuracy metering application using optical sensing to be successfully demonstrated under actual operating conditions in a high voltage environment. Although this testing was performed with a revenue metering system, there is nothing in the laboratory testing and field data that indicates the MOCT approach can not be interfaced with solid state/digital protective relaying systems. When this step is taken, the capabilities and features of optical current sensing will open the possibility of a high reliability, wide dynamic range sensor that will provide an optimum interface between high voltage systems and electronic equipment expected to monitor and control those systems. References 1. Jenkins, F. A., and H. E. White (1957): "Fundamentals of Optics,'I McGraw-Hill, New York 2. S. J. Weikel and P. M. Johnston, "The Application Of Magneto Optic Sensors To Current Sensing", presented at the PEA Fall Relay Committee Meeting, September 1987.

REFERENCE- MOCT REFERENCE

and allows the performance to be examined in detail. The percent registration calculation was performed by the data logging system, based on register readings taken from each meter system on a 15 minute basis. The reference meter provided this information with a resolution of & 0.01 MWh, while the register associated with the MOCT metering system supplied a reading with a resolution of & 1.0 MWh. Since both metering systems were started from an initial reading of zero, this difference in reading resolution causes the computation of percent difference to be especially sensitive when the register readings are small. This sensitivity is responsible for the short term variations in the curve. As the registrations of each metering system increased, the short term variation becomes a smaller contribution and eventually is negligible. This particular installation was a difficult system to meter due to the reversal in load flow which occurred on a daily basis and caused the meters to stall when the power flow reversed. In addition, the current levels were such that the MOCT never operated above 25% of its rated current, and for the majority of the test, the unit was operating under 10% of rated current. In order to insure more accurate registration of the reference, the CT selected to operate that unit was a 1500:5 ratio, permitting its operation closer to nominal current.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

554

Johnson received a BA in Mathematics from St. Vincent College in 1969, a BSEE from Massachusetts Institute of Technology in 1970, a MSEE from Carnegie Mellon University in 1972, and a MBA from Duke University in 1985. Joined Westinghouse in 1970 as a design engineer at the R&D Center. Transferred to the Meter Division in 1978. Presently responsible for optical transducers, solid state metering, and advanced manufacturing process development. Has been granted 11 Patents for design work and has published papers on subjects metering systems, reliability of electronics in the power system environment, and manufacturing systems for electronics used in severe environment applications. T. W. Cease (Member) received his BSEE from the University of Arkansas in 1972. Joined TVA in 1978 as an Instrument Engineer at Bellefonte Nuclear Plant. Subsequently, in 1985, became a Project Engineer in Research and Development.

P. M.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.

DISCUSSION
P.G. Mclaren. University of Manitcha. Winnipeg, Manitoba, Canada R3T 2N2.

It is encouraging to see optical transducers take another step nearer to applicaiion. Previous Japanese work [Al. [Bl was aimed at switchgear applications, in panicular SF, substations and concentrated on bandwidth rather than accuracy. Reference [A] used an optical system similar to that used by the authors whereas [B] used a shorter optical path housed in the air gap of an iron core surrounding the measured c u m n t Could the authors comment on why they used the longer optical path technique rather than the iron core technique.
For relaying purposes, parricularly high speed relaying and high speed data recording, it is essential that the transducerlelectronics is free from saturation and reproduces the primary current waveform accurately. Could the authors give some more details of the bandwidth of their device or what they expect to achieve in this respect in future devices? I appreciate that the present device was intended for metering purposes which does not require high bandwidth performance.

M. Kanoi, et al, "Optical voltage and current measuring system for elechic power systems", IEEE Transactions on Power Delivery, Vol. PWRD-1. No. 1. Jan. 1986, p. 91.
T. Mitsui, et al, "Development of Fiber-optic Voltage Sensors and Magnetic Field Sensors", E E E Transactions on Power Delivery, Vol. PWRD-2, No. 1, Jan. 1987, p. 87.

Manuscript received J u l y 25, 1989.

P. M. JOHNSTON f T. W. CEASE: The authors thank Dr. McLaren for his comments and his recognition of the differences between earlier work and our development efforts aimed at both high accuracy and wide dynamic range. Use of the flux concentrator approach of placing an optical element in the gap of a magnetic core is a technique that can be used to increase the amount of modulation of the light for a given current. Applying this technique reduces the difficulty of recovering small currents with the optical system but the advantage comes at a compromise to the dynamic range of the transducer as well as complexities due to temperature effects and non-linearities because of the magnetic core material. The optical design approach taken for this sensor was believed to be a better one since it eliminated the non-linear effects and temperature problems while it also made the

sensor element virtually immune to interference from external fields. In the gapped core approach, one assumes that external fields can be ignored and a straight piece of glass could be used. Using a prism to guide the light around the conductor can be shown make the sensor immune to currents which do not flow through the area enclosed by the light path, as long as the light is propagating in a direction essentially parallel to the magnetic field. Since the optical path length is longer, and the amount of modulation obtained is directly proportional to path length, the longer optical path somewhat offsets the signal to noise gain obtained from the higher permeability of the gapped core approach. At the same time, the system does not have to deal with the non-linearities of the magnetic core at both low and high currents. The selected optical design had a further advantage of being able to be built from a single piece of glass. A monolithic sensor element has shown superior performance over earlier prototypes that were fabricated from multiple pieces of glass and then joined at the corners to make the closed optical path structure. Construction of the sensor is also less complicated and more reliable since all parts of the structure are contained in the glass element, with no air gaps that require special protection from contamination, such as dirt or moisture. Regarding the bandwidth of the sensor, this is limited by the transit time of light through the sensing element. The design described in our paper has a transit time on the order of 1 nanosecond, indicating that frequency response is more than adequate for power system measurements (including very specialized ones). The bandwidth, limiting factor for the system 1s the signal processing described electronics. A trade-off exists between minimum sensitivity to primary current and the desired bandwidth. For the system described in the paper , the bandwidth 1QkHz. This selected was approximately provided ample frequency response for high speed systems and permitted recovery of small currents for metering purposes.
Manuscript received October 11, 1989.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on October 23, 2009 at 02:16 from IEEE Xplore. Restrictions apply.