PREPARATION OF THE NATIONAL MAGNETIC FIELD STANDARD IN CROATIA

Transcription

1 n IMEKO TC 11 International Symposium METROLOGICAL INFRASTRUCTURE June 15-17, 11, Cavtat, Dubrovni Riviera, Croatia PREPARATION OF THE NATIONAL MAGNETIC FIELD STANDARD IN CROATIA A. Pavić 1, L.Ferović, I.Leniče 3 University of Zagreb, Faculty for Electrical Engineering an Computing, Zagreb, Croatia 1 : armin.pavic@fer.hr Abstract The rationale for, an the way of preparation of the croatian national stanar for low frequency magnetic fiel are presente. The technical characteristics of the system for the reference magnetic fiel generation are escribe. The calculation of the reference fiel nonuniformity an the epression of the reference fiel generation uncertainty are presente, an the eperience with the preliminary use of the system for the magnetic fiel-meters calibration is escribe. Keywors: magnetic fiel, stanar, calibration, uncertainty. 1. INTRODUCTION As a consequence of growing public concern regaring the possible averse health effects of low frequency magnetic fiels, the national an international protective legislation prescribes measurement an control of fiel levels in human environment. The magnetic fiel-meters use for such legal metrology have to be calibrate regularly. Calibration of magnetic fiel meters requires magnetic fiel stanar that generate reference magnetic fiels of high uniformity (in continuation of the paper this stanar will be calle a reference). Such reference magnetic fiels are generate by specially constructe reference coils. Magnetic fiels generate by the reference coils are not completely uniform, an the fiel nonuniformity contributes to the overall calibration uncertainty. The maimal permitte level of the reference fiel nonuniformity is recommene by international stanars [1, ], where some reference coils constructions are recommene as well. For realization of our reference magnetic fiel, the stanar square loop reference coil has been use. The coil has been carefuly constructe an buil, bearing in min that coil construction parameters influence, not only the reference fiel uniformity, but also the coil electrical characteristics. Reference coil electrical characteristics, as resistance an inuctance, are important for sizing an construction of the power supply neee for the reference system realization. For realization of the national magnetic fiel reference, the calculation of the reference fiel uncertainty is of utmost importance. For that purpose the reference fiel nonuniformity has been calculate, using the numerical fiel calculation [3]. Finally, the constructe magnetic fiel reference system has been teste in calibration of 5 ifferent fiel-meters. The proceure an results of this calibration are escribe at the en of the paper.. REFERENCE SYSTEM CONSTRUCTION Calibration of a magnetic fiel-meter is normally one by introucing the fiel-meter probe into a nearly uniform reference magnetic fiel of nown magnitue, irection an waveform. While DC magnetic fiel-meters, with small Hall-effect probes, can be calibrate in the narrow space of a reference fiel generate by permanent magnet, for calibration of AC fiel-meters, with larger, inuctive (coil) probes, the reference fiels with larger area of uniformity are neee. For generation of such reference fiels, systems with large reference coils are use. Such in of reference system has been constructe at the Faculty for Electrical Engineering an Computing the University of Zagreb. The magnetic fiel reference system is generally functioning so that electric current of nown strength is let to flow through the winings of the reference coil, generating the magnetic fiel of the nown strength at the centre of the coil, where the probe is put for calibration. A basic schematic view of a circuit for reference magnetic fiel generation (taen from [1]) is presente in Fig 1. Fig. 1. Schematic of circuit for reference fiel generation During calibration, the probe is place in the central area of the reference coil where the fiel has highest uniformity. At the centre of the reference coil, the magnetic flu ensity is proportional to the current strength I. Factor of proportionality epens upon the shape an imensions of the reference coil, an can be calculate. y means of this coil factor, for any current strength I, magnetic flu ensity at the centre of the coil can be etermine. The main components of such a system for reference magnetic fiel generation are: power supply, an instrument for current measurement, an a reference coil. While a stabile power-supply, an precise ampermeter are stanar components that can be generally foun, the reference coil is a special component that contributes the most to reference system uncertainty. Therefore, construction of the reference coil is of special importance. 95

2 n IMEKO TC 11 International Symposium METROLOGICAL INFRASTRUCTURE June 15-17, 11, Cavtat, Dubrovni Riviera, Croatia.1. Reference coil construction For reference fiel generation, a square form of reference coil has been use, as it is propose in international stanars [1, ]. To achieve high fiel uniformity in the coil central volume, where the probe is put for calibration, the 1 m wie reference coil configuration has been use. The shape an size of the reference coil configuration is presente in Fig.. This configuration has a square coil with N= turns, arrange in 5 layers, each with 4 turns of thicness =1,1 mm. Accoring to [], the (inner) sie imension of the inner layer turn is a=1 m. This coil (wining) length is l=4,4 cm, an the wining with (thicness) is w=,55 cm, as presente in Fig.. Therefore, our multilayer coil requires more accurate fiel calculation than the one enable by epression (1). For an accurate calculation of the magnetic flu ensity at any point in the fiel generate by the reference coil with wining of N turns, the wining spatial arrangement, i.e., the eact position of each turn shoul be consiere. General geometry for such fiel calculation is presente in Fig. 3. Current loop in Fig. 3 represents one rectangle turn, with the point P for fiel calculation islocate from the turn centre. l a o y a o z w m=5 ~ n=4 l w a Fig.. Shape an imensions of the reference coil In Fig. the plan view, sie view, an the cross-section of the reference coil are rawn. To fit together, three coil rawings are presente with ifferent scaling. Knowing the presente shape an imensions of the reference coil enables us to calculate the reference fiel for the given current strength... Reference fiel calculation For calculation of the magnetic flu ensity at the centre of a square current loop, we can use a simple formula (base upon the iot-savart law) IN = m p a. (1) where I is current strength, N is number of turns, an a is imension of the square loop sie. From (1) we can see that reference fiel level is inversely proportional to the coil sie imension a. Thus, recommenation for larger coil imensions means the requirement for stronger current source, what complicates the calibration system realization. However, epression (1) can be use only for approimate fiel calculation, because it approimates current loop with a line without imension, i.e. it oes not tae in account the thicness of the current loop. In such, iealise case, all the current woul be in one layer, an without any thicness. In case of a real coil, where each turn represents separate current loop, the situation is more complicate. As we can see from the Fig., our coil turns are not in the same layer, an also they have ifferent istances from the coil aes. All turns o not have the common centre point, an the fiel level at each point shoul generally be calculate as a vector sum of contributions of all N turns. Fig. 3. Geometry for calculation of rectangular coil fiel ase upon the iot-savart law, the fiel z-component z (aial fiel) prouce by a rectangular current loop I at the point P(,y,z), as shown in Fig. 3, can be epresse as: 4 µ IN ( 1) C z = + 1. () 4π = 1 r [ r + ( 1) C ] r ( r + ) where: C 1 = C 4 =a+; C = C 3 =a ; an 1 = =b+y; 3 = 4 =y b. This is a general epression for a rectangular loop fiel, which in case of a square loop is simplifie by b=a. Accoring to [1, ], only the fiel z-component z is of our interest, because only this (aial) fiel component, which is parallel to the coil aes, effects the meter reaing uring calibration. y aition of partial contributions of each of N turns, calculate by means of (), an accurate calculation of aial fiel at any point can be performe. That can enable a etaile analysis of magnetic fiel prouce by the reference coil. This analysis is necessary to etermine the fiel nonuniformity, which is the main source of the reference system uncertainty. 3. REFERENCE SYSTEM UNCERTAINY The reference fiel nonuniformity strongly influences the uncertainty of reference fiel generation, which is the main component of the calibration system uncertainty. To eep the calibration uncertainty low, the reference fiel nonuniformity in the probe area, accoring to [], shoul not ecee 1 %. y means of () fiel calculation aroun the coil centre was use to quantify our reference fiel nonuniformity. 96

3 n IMEKO TC 11 International Symposium METROLOGICAL INFRASTRUCTURE June 15-17, 11, Cavtat, Dubrovni Riviera, Croatia.1. Reference fiel nonuniformity The fiel nonuniformity at some point in the fiel is a measure of the eparture of the aial fiel level z at that point from the fiel level at the centre of the coil. Epresse as a percentage, it can be calculate as: z n = 1. (3) The fiel level changes while moving from the centre of the coil system along the coil aes (z-aes). The magnitue of that change characterizes the longituinal uniformity of the reference fiel. Fiel calculation is showing that fiel level ecreases with the longituinal istance from the centre of the coil (in longituinal irection) as presente in FIg. 4. z / z (m) Fig. 4. Longituinal uniformity of the reference fiel At the other sie, fiel calculation is showing that fiel level ecreases with the transversal istance from the centre of the coil (in transversal irection) as presente in FIg. 5. z / (m) Fig. 5. Transversal uniformity of the reference fiel A view on Fig. 4 an Fig. 5 inicates important fact that, concerning the effect of the fiel nonuniformity on the fielmeter reaing uring calibration, errors with probe positioning in longituinal an transversal irection mutually compensate. We can see also, that transversal nonuniformity is slightly lower than the longituinal. This means that a higher error margin is allowe in transversal positioning of the probe. That is convenient; because the longituinal positioning is easier (one has only to align the planes of the probe an the reference coil). To get some quantitative inicators, the eact values of the fiel nonuniformity have been calculate for several points isplace from the coil centre in longituinal, as well as in the transversal irection, Results are presente in tables I an II. TALE I. Calculate fiel nonuniformity n along the z-ais Distance from the centre (;;) cm Displacement 1 (;;1) 3 (;;3) 5 (;;5) n % -,49 -,444-1,6 TALE II. Calculate fiel nonuniformity n along the -ais Distance from the centre (;;) cm Displacement 5 (5;;) 7 (7;;) 1 (1;;) n %,633 1,54,6 The actual fiel-meter probes have usually the shape of roun coil with iameter to,1 m. During calibration, the probe is place in the reference coil centre, so that planes an aes of these two coils coincie. Results of fiel calculation, presente in tables I an II, perfectly match the results from [] where the largest eparture of magnetic fiel from the central value, for the,1 m iameter probe, is eclare to be 63%. The ifference of,3% comes from the fact that in [] the central fiel value was calculate by means of (1). It is noteworthy that a fiel-meter with a coil probe inicates a magnetic fiel value that is an average of the cross-sectional area of the probe, an the ifference between this average an the central value will be less than the maimum percentage eparture from the central value. In case of the,1 m coil probe, the average fiel is only,31 % more than the central (calibration fiel) value. Moreover, if the shape an imensions of the probe are nown, presente accurate fiel calculation gives a possibility to etermine the actual average fiel value in the probe area. That enables the full compensation for the reference fiel nonuniformity, thus eliminating the contribution of fiel nonuniformity to the uncertainty of the reference fiel generation... Uncertainty of the reference fiel generation esie the uncertainty component ue to the fiel nonuniformity u n, the uncertainty of the reference fiel generation u is influence by the uncertainty of the current strength measurement u c, as well as the uncertainty of the reference coil imensioning u (which reflects the possible influence of imprecision in coil imensioning). Accoring to [], the uncertainty of the reference fiel generation u can be calculate as a combine uncertainty of these three components: n c u = u + u + u (4) 97

4 n IMEKO TC 11 International Symposium METROLOGICAL INFRASTRUCTURE June 15-17, 11, Cavtat, Dubrovni Riviera, Croatia Taing the worst case of,633 fiel nonuniformity as an error margin in a rectangular istribution (so that it inclues eventual errors in probe positioning) gives an uncertainty ue to fiel nonuniformity of the value u n =,37 (coverage factor 1). The uncertainty of current measurement u c epens upon the stability of power supply an the precision of the instrument use for current measurement. In our reference system we use a stabilize power source an precise measuring instrument, so that error margins of current measurement are within ±,1 %. That results in the uncertainty value u c =,58 (coverage factor 1). Measurement of the realize reference coil imensions have shown a high precision of coil manufacturing, so that the uncertainty of the reference coil imensioning can be estimate to be u =,1 % (with coverage factor 1). Accoring to (4), we can calculate the uncertainty of our reference system (with a coverage factor 1) being u =,37 +,58 +,1 = ±,39 % Etension. of the confience level to respectable 95 % still leaves us a relatively low reference system uncertainty u = ±,78 % 4. REFERENCE SYSTEM VERIFICATION Calculate low uncertainty of the reference system inicates it's capability to became a national stanar for low frequency magnetic fiel, an be use for magnetic fiel-meters calibration within the legal control of electromagnetic fiels in human environment. To verify the calculate accuracy of the reference fiel generation, a small coil probe has been imensione an carefully manufacture, so that it's inuctance is precisely nown. The coil was place at the centre of the reference coil, which was powere by a stabilize source of the sine waveform current. y means of probe voltage, the fiel value at the centre was measure an compare to the fiel value etermine. from the coil factor. The ifference i not ecee,45 %, proving the high accuracy of reference fiel generation. To test the reference system capabilities in practice, the reaing of 5 ifferent fiel-meters has been teste in the reference fiel, the same way as it woul be one uring calibration. Each of 5 teste meters has a 3D probe, which contains 3 coils, assemble together facing 3 orthogonal space irections. Therefore, each meter is teste in 3 ifferent (orthogonal) positions, mare with, y, an z. Meters are mare as M1 to M5, an results of fielmeter testing are presente in Table III, at the en of the paper (net page). In the first colon of the table the referent fiel values, calculate by means of coil factor, are state (bol) while other colons contain fiel-meters reaings in three ifferent positions (, y, an z). The average values, an meter errors are calculate an state in the table as well. Meters have been teste in power frequency fiel of 5 Hz, which is common aroun the most sources of low frequency magnetic fiel. The only potential problem, encountere uring this test, were magnetic isturbances coming from environmental fiels in the faculty builing, which were evient at the low values of the reference fiel. Calibration results have been compare to the meters technical specifications, an ehibite errors showe perfect concorance with the specifie fiel-meters accuracy. 3. CONCLUSION A system for generation of low frequency reference magnetic fiel has been constructe an teste. The accurate fiel calculations, as well as the system practical verification, prove the system ability to generate highly uniform reference magnetic fiel with high accuracy. That ability maes this reference system appropriate for realisation of the national magnetic fiel stanar, which coul be use for calibration of fiel-meters use in legal environmental measurements. To preserve a low calibration uncertainty at the lower reference fiel values, some problems still have to be resolve. As the most important appears the elimination of magnetic isturbances uring calibration in wea reference fiels, which can be achieve by construction of an aequate Faraay's cage. REFERENCES [1] ANSI/IEEE: IEEE Stanar Proceures For Measurement of Power Frequency Electric an Magnetic Fiels from AC Power Lines ; ANSI/IEEE St , [] IEC: Measurement of low frequency magnetic an electric fiels with regar to eposure of human beings, International Stanar IEC 61786: [3] A. Pavić an Ž. Štih: The influence of reference coil wining arrangement on the uniformity of the reference fiel for magnetic fiel meters calibration; Proceeings of th International Metrology Symposium an 1st Regional Metrology Organisations Symposium RMO 8, Novemb. 1-15, 8, Cavtat-Dubrovni, Croatia, pp Author (s): Prof. r.sc. Armin Pavić, armin.pavic@fer.hr; As.prof.r.sc. Lua Ferović, lua.ferovic@fer.hr ; As.prof.r.sc. Ivan Lenice, ivan.lenice@fer.hr ; University of Zagreb, Faculty of Electrical Engineering an Computing, Unsa 3, 1 Zagreb, Croatia:. 98

5 n IMEKO TC 11 International Symposium METROLOGICAL INFRASTRUCTURE June 15-17, 11, Cavtat, Dubrovni Riviera, Croatia TALE III. Results of 5 fiel-meters calibration /µt Meters M1 M M3 M4 M5,997,988,989,97,991,966 y,966,987,965,987,97 z,983,984,993,986,97 av.,979,987,976,988,969 Error /% -1,8-1,33 -,4-1, -3,13 4,9998 4,991 4,996 4,878 4,965 4,9 y 4,949 4,951 4,813 4,956 4,88 z 4,958 4,914 4,878 4,957 4,83 av. 4,966 4,954 4,856 4,959 4,877 Error /% -,68 -,9 -,87 -,81 -,46 5,8 49,41 49,1 48,5 49,43 49,6 y 49,35 49,35 47,96 49,6 47,4 z 49,39 49,19 48,55 49,57 48,1 av. 49,38 49, 48,34 49,54 48,37 Error /% -1,9-1,61-3,37 -,97-3,3 1,19 1, 99,5 97, 99,16 99, y 98,57 99,88 96,48 99,8 95, z 98,5 99,51 97,86 99,6 96, av. 99,9 99,64 97,11 99,3 96,8 Error /% -1,1 -,56-3,7 -,96-3,39 494,5 54, 489, 48, 489,4 493, y 5, 496, 486, 491,7 471, z 511, 485, 49, 49,4 477, av. 54,3 49, 486,7 49,5 48,3 Error /% 6,13 -,8-1,49 -,7 -, SM.pf 99