IL96060A - Optical fiber current sensor - Google Patents

Description

>03 IN I'D ¾ DDinan Ο Τ ^VP >n -|ΊΟ>ΪΙ An improved optical fiber current sensor 1. THE STATE OF ISRAEL, ,η»η·>ΌΝ Π' ATJN^ jvryin , κνο* m' a ATOMIC ENERGY COMMISSION, *nw-^m j ynA npnat? onan SOREQ NUCLEAR RESEARCH CENTER 2. RAMOT UNIVERSITY AUTHORITY FOR npnal? η κο»οη2» JIN ηΐϊη Dim APPLIED RESEARCH AND INDUSTRIAL a"ya >n*»¾>vn nimai »¾nn»« DEVELOPMENT LTD.

The inventors: :o>N> onn 1. Amit Ben-Kish >p- n>oy .1 2. Ehud Shafir 3. Prof. Moshe Tur n\!)Q *31-|D .3 C: 81296 FIELD OF THE INVENTION The present invention relates to a sensor for measuring an electric current flowing through a current carrying conductor, based on optical sensing using the Faraday rotation effect.

BACKGROUND OF THE INVENTION It is known to use optical fibers for measuring the magnitude of an electrical current flowing through a conductor. Typically, optical fibers are employed for measuring the current flowing through very high voltage cables and busbars and offer the particular advantage that, since they are electrically insulating they can withstand high electric fields and also do not influence the measurement.

One known form of optical fiber current sensor utilizes the Faraday rotation effect whereby a longitudinal magnetic field in the direction of propagation of a beam of linearly polarized light causes rotation of the polarization plane. Fig. 1 shows a typical system for measuring electrical current using the Faraday rotation effect. Such a system includes a laser 1 for producing a beam 2 of monochromatic light and a first polarizer 3 intercepting the beam 2 so that light passing therethrough is linearly polarized in a predetermined plane. The beam emerging from the first polarizer 3 is passed through a first end of a single mode optical fiber 4 which encircles an electrical conductor 5 having a current passing therethrough whose magnitude I is required to be measured.

A second polarizer 6 is disposed at an opposite end of the optical fiber 4 and functions as a polarization analyzer for detecting a rotation of the plane of polarization between the light beam entering the optical fiber 4 and the light beam emerging therefrom after encircling the electrical conductor 5. A detector 8 is responsively coupled to the second polarizer 6 for producing an electrical signal related to the rotation of the plane of polarization at the end of the optical fiber 4.

In accordance with Faraday's Law, it may be shown that the magnitude of the polarization rotation of the beam passing through the optical fiber 4 is given by: where: Θ is the angular rotation in radians V is the Verdet constant H is the magnetic field strength, and dl is an incremental path length of the optical fiber By applying Ampere's Law for a closed fiber loop around a current carrying conductor, equation ( 1 ) , showing the relationship between the polarization rotation and the magnetic field strength, may also be expressed in terms of the current I as follows: Θ = NVI (2) where: I is the current encircled by the fiber loop, N is the number of optical fiber loops, and Θ and V are as defined above.

Thus, it follows from equation (2) that ideally, the sensitivity of the arrangement shown in Fig. 1 may be increased by increasing the number of loops with which the optical fiber 4 encircles the electrical conductor 5.

It can also be shown that when the polarization plane of the second polarizer 6 is perpendicular to that of the first polarizer 3, the output power POU£ of the beam incident on the detector 8 is proportional to sin26. Since, from equation ( 2 ) , Θ is directly proportional to the current I, it follows that the output power POU£ likewise varies sinusoidally with the current I. Such a relationship is shown graphically by the solid line 9 in Fig. 2 of the drawings.

Under these circumstances, when the magnitude of the current I is zero, the beam entering the optical fiber 4 is subjected to no Faraday rotation and so will emerge from the optical fiber 4 with no net rotation. It will therefore be blocked by the second polarizer 6 and POU£ will equal zero. As the magnitude I of the current flowing through the electrical conductor 5 increases, the beam entering the optical fiber 4 will be subjected to Faraday rotation and the rotated component will pass through the second polarizer 6 so as to be detected by the detector 8.

Such an arrangement is not completely satisfactory for several reasons. Thus, for small values of current I, the sensor is neither sensitive nor linear and furthermore the net direction of the current is unclear owing to the fact that the transfer function is symmetric.

It is therefore preferable to rotate the second polarizer 6 so that its polarization plane is at an angle of 45° to that of the first polarizer 3, whereupon the relationship between the output power POU£ and the current I is given by the dotted line 9' in FIG. 2. In such an arrangement the sensitivity of the current sensor is increased. Furthermore, as will be seen from the dotted line 9' in FIG. 2, the graph of POU£ against I is approximately linear for low values of the current I and since there is a measured power output even for zero current, any change in current will produce a corresponding change in the measured power pout ·-η sucn manner that the direction of Faraday rotation may be determined.

The Faraday rotation effect relates to any optical medium which is subjected to a magnetic field. In high voltage electrical installations for which optical current sensors may typically be used, several different high voltage busbars are commonly arranged in relatively close proximity to one another. This is the case, for example, in multi-phase electrical transmission networks. In all such systems wherein two or more current carrying conductors are arranged in close proximity, it is clearly important to ensure that, when employing the Faraday rotation effect in order to measure the current flowing through one of the conductors, the magnetic fields associated with adjacent conductors do not influence the measured polarization rotation and thus detract from the measuring accuracy.

Thus, it will be seen from the equation (1) above that if the optical fiber completely encircles the electrical conductor whose current is to be measured, the loop integral along the optical fiber of Η·<31 measures only those currents flowing within the electrical conductor and the effect of external magnetic fields outside of the optical fiber loop is zero. From this follows the requirement that, in order to neutralize the effect of external magnetic fields, the optical fiber sensor must completely encircle the electrical conductor whose current is to be measured.

In an ideal current sensor employing the Faraday rotation effect, the polarization plane would be rotated as a result of the circular birefringence generated by the magnetic field along the optical path and there would exist no other factors tending to produce such rotation. However, in practice, other phenomena can affect the state of polarization such as, for example, linear birefringence caused by intrinsic manufacturing imperfections, by transverse pressure applied to the optical fiber or by bending the fiber. The first two causes of linear birefringence can, to a large extent, be corrected but since the optical fiber 4 must encircle the electrical conductor 5, bending -the optical fiber 4 is inevitable. There will thus always exist some linear birefringence which superimposes a further polarization distortion on that rotation which is due genuinely to the Faraday effect, thereby complicating the relationship between the detector output and the measured current I. can be shown that the amount of retardation, i.e. the phase delay between the linear components of the optical guided wave travelling in the fiber, caused by bending the fiber (as distinct from Faraday rotation) is given by: 0.85 P p 2nR'N and δ = — (3) λ N where: p - is the phase retardation in radians caused by bending (i.e. the linear birefringence), R - is the bending radius in meters N - is the number of fiber loops λ - the vacuum wavelength of the light source in meters r - is the radius in meters of the optical fiber, and 6 - is the retardation per loop in rad/m.

For the current sensor of Fig. 1 where the first polarizer 3 is parallel to the plane of the loop and the second polarizer 6 is at angle of 45° thereto, it may be shown that: out 2θείηφ) (4) where: out is the intensity of the light beam emerging from the second polarizer, in is the intensity of the light entering the first polarizer, Θ is the Faraday rotation in radians, and Φ is given by: Φ2 = p2 + (2Θ)2 (5) It will be seen from equation ( 5 ) that when the linear birefringence p and the circular birefringence 2Θ are comparable, the system transfer function is complicated since it now depends also on p.

In the low current region, the bending-induced linear birefringence is very much greater than the Faraday induced circular birefringence. In this case p » Θ, and equation (5) simplifies approximately to: Φ ~ P (6) and substituting equation ( 6 ) in equation ( 4 ) gives : out = %PfnfI + 2Θ si ip) = ½P 1i1n1 + S-I (7) where S is the system scale factor and all the other variables have the same meaning as defined previously. The scale factor S is given by: S = N'V'Sin 'P^ = V'S±n(N6) «Ρ^ P δ (8) According to equation ( 7 ) , the output power of the light reaching the detector is proportional to the current passing through the electrical conductor. For the measurement of low currents, it is necessary to increase the scale factor S. According to equation (2), increasing the number of turns N with which the optical fiber encircles the electrical conductor improves the sensitivity of the measurement. However, the magnitude of linear birefringence p when present, as determined by equation (3), is also proportional to the number of turns N and it may therefore be seen from equation (8) that once the product IV·6 is chosen such that sln(N6) = 1, the scale factor S cannot be increased further by increasing N.

Various solutions to the problems introduced by the presence of linear birefringence have been proposed as follows: [1] Kazuo Shiraishi, Satoshi Sugaya, and Shojiro Kawakami, "Fiber Faraday rotator", Appln. Opt., vol.23, no. 7, pp.1103-1106, April 1984. [2] G. . Day and S.M. Etzel, "Annealing of bend-induced birefringence in fiber current sensors", in Proc. IOOC- ECOC, 1985, pp. 871-874. [3] R. Ulrich and A. Simon, "Polarization optics of twisted single-mode fibers", Appln. Opt., vol. 18, no. 13, pp. 2241-2251, July 1979. [4] A.J. Barlow and D.N. Payne, "Polarization maintenance in circulary birefringent fibers", Electron. Lett., vol. 17, no. 11, pp. 388-389, May 1981.

Reference [1] proposes raising the circular birefringence, caused by the Faraday effect, by fabricating fibers with a higher Verdet constant. It will be seen from equation (1) above that the Faraday rotation is directly proportional to the Verdet constant and thus will be increased by increasing the magnitude of the Verdet constant. However, fabricating optical fibers with sufficiently high Verdet constants is difficult and militates against such a solution.

Reference [2] suggests removing the influence of the bend-induced linear birefringence by annealing the coiled fiber so as to release the transverse pressure therein. However, removal of the fiber coating and subsequent high temperature annealing tends to weaken the fiber, making it difficult to produce a compact multiturn, high-sensitivity device.

Several proposals have been suggested for manufacturing optical fibers whose intrinsic circular birefringence is increased relative to their linear birefringence.

Thus, for example, Reference [3] proposes a solution whereby sufficient torsional stress is induced by twisting the fiber in order to swamp the bend-induced linear birefringence p. The Faraday induced rotation 6f then augments the natural circular birefringence caused by twisting θ^. Thus, in such an optical fiber, the circular components are dominant and (6f + θ^) » p. In this case Φ * 6f + 0£ and equation (4) reduces to: where: ef = VNI, et - ge, g = a property of the material, and E = twist per unit length.

It can be seen that acts as a bias and the output power Pout is proportional to the current I. However, the number of turns per meter that a commonly used fiber can withstand before breakage (approximately 70) imposes a relatively low upper limit on the degree of circular birefringence which can thus be produced. Consequently, such fibers remain sensitive to bend-induced birefringence. Furthermore, it is difficult to produce a uniform twist along the fiber, resulting in non-uniform intrinsic circular birefringence.

As described in Reference [4] , a great effort is being invested in developing new optical fibers with intrinsic circular birefringence, induced by twisting the fiber during the extrusion process. In this case, as in Reference [3], the sum of the circular birefringence caused by twisting and by the Faraday rotation is much greater than the linear birefringence caused by bending.

A major drawback with all of the above solutions is that manufacturing such fibers is difficult, the extrusion process is expensive and the resulting optical fiber is highly sensitive to thermal drifts.

SUMMARY OF THE INVENTION It is an object of the invention to provide an optical current sensor employing a regular single-mode fiber and whose sensitivity may be increased by increasing the number of turns.

According to the invention there is provided in a Faraday sensor for measuring an electric current flowing through a current carrying conductor, the sensor comprising: a light source for producing a beam of polarized light, an optical fiber for surrounding the conductor and propagating the light beam from a first end of the fiber to a second end, and detecting means at the second end for detecting a change in polarization of the light beam; the improvement wherein: the optical fiber is bent into a plurality of first and second sections which in combination encircle the conductor an integral number of times, and wherein the first sections measure Faraday rotation with negligible linear birefringence whilst the second sections circumvent the effect of linear birefringence without influencing the measurement.

In a preferred embodiment of the invention, the optical fiber is bent into a plurality of substantially linear segments, the fiber being looped at the transition between adjacent segments. The linear segments measure Faraday rotation with negligible linear birefringence, and the loops between adjacent linear segments circumvent the effect of linear birefringence whilst playing no part in the measurement.

Thus, according to the invention, the effect of linear birefringence is substantially circumvented and the resulting sensor combines compact configuration permitting the sensitivity to be increased by increasing the number of turns.

BRIEF DESCRIPTION OF THE DRAWINGS For a better understanding of the invention and to understand how it may be employed in practice, a preferred embodiment will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which: FIG- 1 is a pictorial representation showing schematically a prior art optical current sensor; FIG. 2 is a graph useful in explaining how the sensitivity of the sensor shown in FIG. 1 may be improved for low currents; FIG. 3 is a pictorial representation showing an optical current sensor according to the invention; and FIG. 4 is a graph showing the dependence on the number of turns of scale factors for a prior art optical current sensor and for an optical current sensor according to the invention.

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT Referring to FIG. 3 of the drawings there is shown an electrical current sensor shown generally as 10 for measuring the magnitude I of a current flowing through an electrical conductor 11.

The current sensor 10 comprises a laser 13 for producing a beam 14 of quasi-monochromatic light and a first polarizer 15 so that, on emerging from the first polarizer 15, the laser beam 14 is polarized in a first polarization plane parallel to the plane of an optical fiber loop shown generally as 16 surrounding the electrical conductor 11. The polarized light beam emerging from the first polarizer 15 is propagated from a first end of the fiber 16 to a second end thereof.

A second polarizer 17, having a polarization plane which is at an angle of 45° to that of the first polarizer 15, is provided at the second end of optical fiber 16 so as to analyze the polarization state of the beam emerging from the optical fiber 16.

A beam 18 emerging from the second polarizer 17 is detected by a detector 19 such that the power of the beam 18 detected by the detector 19 is a function of the magnitude I of the current flowing through the electrical conductor 11.

The optical fiber 16 is bent around the electrical conductor 11 an integral number of turns N, so as to form a substantially square formation having sides 21, 22, 23 and 24, respectively (constituting respective adjacent segments). At the transition between adjacent segments 21 and 22, the optical fiber 16 is bent into a loop 26 and corresponding loops 27, 28 and 29 are likewise formed within the optical fiber 16 at the transition points between the segments 22,23; 23,24 and 24,21. Each of the loops 26, 27, 28 and 29 comprises three turns (m=3) whose radius is equal to Re.

The net result of such an arrangement is that the optical fiber 16 surrounds the electrical conductor 11 such that the path described by the optical fiber 16 is substantially linear except at the transitions between adjacent linear segments, where loops are formed within the optical fiber 16.

It will be clear that in the linear segments 21, 22, 23, 24 of the optical fiber 16 the linear birefringence is negligible since there is no bending of the optical fiber 16. Consequently, in these parts of the fiber 16 encircling the current carrying conductor 11, only the Faraday effect acts on the polarization plane producing a Faraday rotation whose magnitude is given by equation (1) above.

On the other hand, at the four transitions 26, 27, 28 and 29 there is large linear birefringence on account of the bending, but there is no net Faraday rotation since, as will be clear from equation (1) above, the loop integral around the closed loops 26, 27, 28 and 29 is zero. From equation (3) above, it is clear that the magnitude of the linear birefringence p is proportional to 271" m/Re for a given optical fiber having a radius r and a given wavelength of light λ. Consequently, by adjusting the number of loops m and the bending radius Re it is possible to so adjust the linear birefringence p that it is an integral multiple of 2TT. Consequently, the beam emerging from each of the respective loops 26, 27, 28 and 29 has the same polarization state as the beam entering.

Thus, the net rotation of the beam entering the optical fiber 16 is due only to the circular birefringence caused by the Faraday effect at the linear segments 21, 22 23 and 24 and is given by: Θ = VJH-dl + V JH-dl + VJH-dl + VJH-dl 21 22 23 24 = V H '•.dl = NVI (10) It can be shown that the system transfer function is given by: pout = % pin I1 + s±n(2B)) (11) where the symbols have the same meaning as defined above. It will be seen from equation (10) that the system transfer function is not influenced by linear birefringence, contrary to hitherto proposed optical current sensors.

Consequently, the geometrical arrangement according to the invention permits the linear dimensions of the optical fiber 16 to be reduced since there is no linear birefringence in the linear segments 21, 22, 23 and 24 of the optical fiber 16, thereby permitting their dimensions to be minimized. Additionally, the sensitivity of the current sensor 10 by increasing the number of loops N with which the optical fiber 16 surrounds the electrical conductor 11.

When it is desired to measure small currents producing a correspondingly small Faraday rotation θ, Θ « 1 radian in equation (10) and the system transfer function is then given by: pout - ¼ pin + 2Θ> = ½ p±n + Ssg'1 <12> where SSg is the system scale factor for the square loop 16 and the remaining symbols have the same meaning as defined above.

Substituting equation (10) into equation (12) gives: ssq - p±n'N'v <13> Equation ( 13 ) indicates that the system scale factor is proportional to the number of loops N with which the optical fiber 16 encircles the electrical conductor 11, permitting small currents to be measured simply by winding the optical fiber 16 many times around the electrical conductor 11.

FIGv 4 shows in the same graph the influence of the number of loops on the resulting scale factor normalized by the input power and the Verdet constant V for typical prior art optical current sensors (as shown in FIG. 1) and of the optical current sensor according to the invention (FIG. 3). The linear dimensions of the prior art current sensor were selected similar to those of the invention but were adjusted so as to optimize the sensitivity of the prior art current sensor for each value of N. The lower curve 35 relating to a prior art optical current sensor, shows that there is substantially no increase in the sensitivity of the current sensor as the number of loops is increased.

The curve designated 36 is a theoretical plot of the normalized SCALE FACTOR against N (number of loops) for the optical current sensor according to the invention. It will be seen that the theoretical curve is a straight line passing through the origin, indicating that the scale factor is directly proportional to the number of loops. Consequently, the sensitivity of such a theoretical current sensor can be increased, as required, simply by increasing the number of loops.

The scale factor was measured in the laboratory for an optical sensor having the configuration shown in FIG. 3 of the drawings and including, respectively, one and three loops. The results are plotted on curve 37 whose deviation from the theoretical curve 36 is sufficiently small (and well within the bounds of experimental error) to demonstrate both the qualitative and quantitative improvement in sensitivity of the preferred embodiment over prior art optical current sensors, as the number of loops is increased.

It should be pointed out that the curve 37 was plotted for a hand-made optical current sensor for which the loops at the transition points between adjacent linear segments were also produced by hand. It will be understood that the loops must be wound accurately in order to produce a resultant linear birefringence through these loops equal to 2im. The fact that this was not fully accomplished in the laboratory test model sufficiently accounts for the relatively small deviation between the theoretical and experimental curves, 36 and 37, respectively.

It will be understood that the principle of the invention resides in employing an optical fiber loop having a geometry comprising a plurality of first and second sections, wherein the first sections measure Faraday rotation with negligible linear birefringence whilst the second sections neutralize the effect of linear birefringence and play no part in the measurement. In the preferred embodiment described above, the geometry of the optical fiber comprises a plurality of linear segments having three loops at the transitions between adjacent segments which are so dimensioned that the net retardation due to linear birefringence is a multiple of 2π. However, it will be apparent that any number of loops m may be employed so long as the radius Re of the loop is adjusted so that the net retardation is an integral multiple of 2n.

It will also be clear that, whilst the preferred embodiment has been described with particular reference to a square formation, any other suitable configuration may be employed so long as the measuring arms of the optical fiber are approximately linear. In this respect, it should be noted that even slight curvature in the "linear" sections of the optical fiber can be tolerated without adversely affecting the overall accuracy of the sensor, providing that the retardation due to linear birefringence in the "linear" sections is swamped by the Faraday rotation therein. In all such arrangements, the net sensitivity of the current sensor is increased by increasing the number of loops with which the optical fiber encircles the current-carrying conductor. -16-

Claims (10)

1. In a Faraday sensor for measuring an electric current flowing through a current carrying conductor, the sensor comprising: a light source for producing a beam of polarized light, an optical fiber for surrounding the conductor and propagating the light beam from a first end of the fiber to a second end, and detecting means at the second end for detecting a change in polarization of the light beam; the improvement wherein: the optical fiber is bent into a plurality of first and second sections which in combination encircle the conductor an integral number of times, and wherein the first sections measure Faraday rotation with negligible linear birefringence whilst the second sections circumvent the effect of linear birefringence without influencing the measurement.

2. A sensor according to Claim 1, wherein the optical fiber is bent into a plurality of substantially linear segments which in combination encircle the conductor an integral number of times, and wherein the effect of linear birefringence at a transition between adjacent segments is substantially eliminated.

3. A sensor according to Claim 1 or 2, wherein the optical fiber is bent into at least one loop at each transition between ad acent segments, each one of said loops being so dimensioned that the light beam passing therethrough is subjected to a cumulative linear retardation at each transition of 2nn where n is an integer.

4. A sensor according to any one of the preceding claims, wherein the optical fiber is bent into a polygonal loop around the conductor.

5. A sensor according to any one of the preceding claims, wherein the light source comprises: means for producing a beam of quasi-monochromatic light, and a first polarizer for polarizing said beam in a first polarization plane.

6. A sensor according to Claim 5, further comprising a second polarizer disposed between the second end of the fiber and the detector for analyzing the polarization state of the emergent beam..

7. A sensor according to Claim 5 or 6, wherein the polarization plane of the first polarizer is perpendicular or parallel to the plane of the optical fiber.

8. A sensor according to Claim 6, wherein the second polarizer has a polarization plane at an angle of 45° to that of the first polarizer, thereby extending the dynamic range of the sensor.

9. A sensor according to Claim 8, wherein the optical fiber loop has a substantially rectangular formation.

10. A sensor substantially as described herein with reference to FIG. 3 of the drawings. For the Applicants LD COHN AND PARTNERS 81296spc. JT/rg/19.8.90