ISSN 00204412, Instruments and Experimental Techniques, 2012, Vol. 55, No. 2, pp. 268–273. © Pleiades Publishing, Ltd., 2012. Original Russian Text © A.V. Akimov, P.V. Logachov, A.A. Korepanov, F.V. Averin, O.V. Savinova, G.L. Mamaev, S.L. Mamaev, 2012, published in Pribory i Tekhnika Eksperimenta, 2012, No. 2, pp. 129–134.
GENERAL EXPERIMENTAL TECHNIQUES
Magnetic Cores Made of an Amorphous Tape for the Induction Accelerator A. V. Akimova, P. V. Logachova, A. A. Korepanova, F. V. Averinb, O. V. Savinovab, G. L. Mamaevc, and S. L. Mamaevc a
Budker Institute of Nuclear Physics, Siberian Branch, Russian Academy of Sciences, pr. Akademika Lavrent’eva 11, Novosibirsk, 630090 Russia email:
[email protected] b JSC “Ashinskiy Metallurgical Works” (AMW), ul. Mira 9, Asha, Chelyabinsk oblast, 456010 Russia email:
[email protected] c Moscow Radiotechnical Institute (MRTI), Russian Academy of Sciences, Varshavskoe sh. 132, Moscow, 117519 Russia email: mamaev_g@mtunet.ru Received August 31, 2011
Abstract—It is proposed to use induction cells with cores wound by a thin tape made of an ironbased amor phous alloy in the linear induction accelerator for flash radiography. The comparison of pulse magnetization characteristics of the toroidal cores made with layer insulation and without it is given. The results of switching on of the injector based on the manufactured cores are presented. DOI: 10.1134/S0020441212010253
INTRODUCTION The linear induction accelerator (LIA) intended for flash radiography has been designed at the Budker Institute of Nuclear Physics. It is assumed that its design parameters will be as follows. The electron energy is 20 MeV, the amplitude of the beam current is 2 kA, the duration of the pulse top is 200–300 ns, the energy stability is ±1%, the pulse repetition rate is ≤0.1 Hz, and the interval between pulses in the dou blepulse mode is 2–10 μs. At the first stage, the electron beam injector with a 2MeV energy and 2kA current has been designed. The 21kV voltage inductors are used to form accelerat ing voltage pulses at accelerating tubes of the injector. The overall dimensions and cost of the entire accel erator are mostly determined by the inductor volume. At the same time, the core losses are the additional load of the pulse power system of the accelerator. Therefore, the main requirements for the cores of the induction system of the accelerator are the maximal induction drop and small demagnetization losses. At present, the most accessible material featuring low losses in the pulse mode is a thin tape (20–30 μm) made of an ironbased amorphous alloy or nanocrys talline alloys [1–3]. It is known that, if the adjacent tape layers are poorly insulated, interlayer eddy cur rents exert a strong influence on magnetic characteris tics of the cores in the pulse remagnetization mode. The voltage between the adjacent layers of the core
operating with a peaktopeak induction of ~2–3 T and pulse duration of <0.5 μs may vary from a few volts to tens of volts. The complexity of applying insulation is aggravated by a need for annealing at a temperature of 400– 550°С. The insulation must not only withstand this temperature but also produce no mechanical stresses in the tape when the magnetic core gets cold. In prac tice, several methods for applying insulation are used [4], e.g., winding of the tape together with a thin insu lating film (Mylar, polyamide, and micaloaded paper). However, this method does not allow one to obtain a high tapefilling factor of the core because of a comparatively large insulation thickness (several micrometers). The other method consists in drawing the tape through a solution with an insulating material (magnesium methylate or liquid glass), drying, and further winding of the core. A similar technology is used by the Moscow Radio technical Institute (MRTI) [5]. It showed good results in manufacturing magnetic cores operating in nano second and microsecond ranges. Among drawbacks of this method, we note a relatively low winding rate (several centimeters of the tape per second) due to a need for drying the coating. As an alternative cheaper version, the manufacturing method of the magnetic cores without insulation by the “Ashinskiy Metallurgi cal Works” (AMW) technology was considered. This work presents the measured magnetic characteristics obtained in the pulse mode for the magnetic cores
268
MAGNETIC CORES MADE OF AN AMORPHOUS TAPE
269
have halved losses as compared with the best amor phous alloys. The new alloy (АМЕТ 1СР) has been designed at the AMW for the project period, and its characteristics are close to those of the 2НСР alloy, but differ in a higher saturation induction (1.56 T).
200 µm Fig. 1. Appearance of the surface of the tape made of the 9KCP alloy.
made with insulation by the MRTI technology and without it. 1. SELECTION OF THE MATERIAL: MAGNETIC CORES WITH THE INSULATION OF THE TAPE The results of investigations of different magnetic materials for cores made with different tape insulation types were presented in [6]. In addition to the tape mark, many process parameters (type of the tape insu lation, tape winding method, thermal treatment mode, and core casing) affect enduse properties of the magnetic cores. Therefore, the most correct method is to compare properties of the magnetic cores manufactured by the same technology. The main sorts of amorphous and nanocrystalline alloys manufactured in Russia (AMW) and amor phous tape produced in the United States (AlliedSig nal) were compared in [7]. In addition, all the samples were manufactured by the MRTI technology. In brief, it includes the following stages: (i) covering of the tape with a liquid glass, (ii) preliminary drying, (iii) quality control of the insulation coating, and (iv) winding around a holder with the controlled tension. The mag netic core is annealed in the longitudinal magnetic field to obtain a high squareness ratio of the hysteresis loop. As a result, it is possible to conclude that (i) of the amorphous alloys, 2605SC (AlliedSignal) and 2HCP (AMW) alloys possess the minimal remagnetization losses; and (ii) when the peaktopeak induction is smaller (~1 T), 5БДСР nanocrystalline alloy (AMW) and its foreign (Finemet) and (Vitroperm) analogs INSTRUMENTS AND EXPERIMENTAL TECHNIQUES
It is necessary to note that quality of the surface and geometry of the tape produced at the AMW is inferior to that of the AlliedSignal tape. The nonsquareness of the tape crosssection and increased roughness, which makes the manufacturers use thicker insulating coat ings, leads to a decrease in the metalfilling factor of the magnetic core. In spite of this drawback and taking into account the fact that the inductor system is the key system in the LIA, we decided to use the tape of Russian origin. For the injector of the induction accelerator, the 1000 × 630 × 25 mm (outer diameter × inner diameter × height) magnetic cores were manufactured from a ~30μmthick 2HCP tape at the MRTI. To reach a good mechanical strength, upon the thermomagnetic treatment, end surfaces of the magnetic cores were covered with a thin epoxide compound layer, and the bandage made of thin glass fiber laminate bands was applied to the outer side surface. The tapefilling factor of the magnetic cores varied from 0.69 to 0.74, when the average value of eight manufactured samples kfl = 0.71. The coercive force measured at a 50Hz frequency and magnetic field amplitude of 85 A/m was ≈7 A/m. The squareness ratio of the hysteresis loop varied from 0.76 to 0.81. This is about a 10% understated value for the 2НСР material, which is heattreated in the longitudinal magnetic field. This can be attributed to the influence of the earlier saturation of inner layers of the magnetic core due to a relatively high outertoinner diameter ratio (~1.6). 2. MAGNETIC CORES WITHOUT INSULATION BETWEEN LAYERS The manufacturing method of the magnetic cores without insulation between layers was proposed at the AMW. The surface of the amorphous tape fails to be ideally planar and contains sufficiently large microas perities; therefore, in spite of the absence of insula tion, adjacent layers are short not over the whole sur face, but only in places of microprojections (the appearance of the 9КСР tape surface is shown in Fig. 1). It can significantly increase the effective resis tance of the tape surface and decrease losses due to the interlayer eddy currents. The weak influence of the amorphous tape insulation on the losses was noted in the earlier investigations of amorphous alloys [8]. This Vol. 55
No. 2
2012
270
AKIMOV et al. FL
Uch
L
Udmg
ТПИ1 10к/50
U(t)
2 Rld RL TS I(t)
1
Fig. 3. Layout of measurements of the pulse characteristic B(H) of the magnetic cores in the LIA injector: (FL) form ing line; (RL) Rogowski loop, (TS) tested sample; (U(t)) voltage across the measuring turn, and (I(t)) current of the Rogowski loop.
of the longitudinal magnetic field application during their cooling down. Fig. 2. Compound magnetic core: (1) cores without layer insulation and (2) insulating rings.
effect was explained by the strong surface roughness of the tape manufactured by the old technology and, as a result, relatively high contact resistance between tape layers. It is possible to reduce interlayer eddy currents in the core without covering the tape by sectionalizing the magnetic core into separate parts wound by a nar rower tape (Fig. 2). As the tape width decreases, the area of separate loops, in which the eddy currents are induced, decreases, and, therefore, the influence of the eddy currents on the magnetic characteristics of the magnetic cores decreases. This approach was used by the AMW for producing magnetic cores from the 2НСР tape with a 5mm width. Upon the winding and thermal treatment of 1000 × 630 × 5 mm cores, each core was impregnated in two stages to acquire mechanical strength. At first, it was impregnated with a weak bakelite lacquer solu tion and then covered by a thin epoxide compound layer. However, due to the absence of insulations between tape layers, the lacquer and compound well penetrate into the core and, upon drying, produce mechanical stresses on the tape surface. This leads to a decrease in the residual induction and inclination of the hysteresis loop; i.e., strengthening the core deteri orates its magnetic characteristics. The impregnation and covering mode, which insig nificantly influenced the static loop of the cores, was selected. Taking into account that the demagnetizing field in the LIA inductors is 100–200 A/m, the square ness ratio of the loop affects the peaktopeak induc tion only slightly. Therefore, when the cores were ther mally treated, we decided to exclude a complex stage
As a result, 570 5mmhigh cores were manufac tured by the AMW. About 9% of them were rejected after measuring their losses in the pulse mode. The remaining cores were grouped by 5 pieces into 104 compound magnetic cores to average the magnetic characteristics. As insulation between adjacent cores, two layers of a 0.15mmthick varnished cloth were used. The average tapefilling factor of the obtained magnetic cores was 0.82. This fact can be explained both by the absence of the insulation and by the decreased influence of the nonuniformity of the tape thickness on the tape layer package in an individual core. 3. PULSE CHARACTERISTICS OF THE MAGNETIC CORES The pulse magnetization curves (dependence of induction B on the field intensity H during the pulse magnetization) of the manufactured magnetic cores were measured when they operated in the rated mode in the LIA injector. The pulse generator designed for powering the induction system of the injector was used for this purpose. Figure 3 shows the layout of measure ments of the pulse magnetization curves. The genera tor, which is based on a single forming line (FL) and is charged from the source Uch, produces a rectangular pulse with a 20kV amplitude, 100ns rise time, and flat top with duration of 200–220 ns across the equiv alent load. Induction B and field intensity H are deter mined by the ratios B(t ) =
∫
t 0
U (t )dt Skch and H(t) =
I(t)/lav, where S is the cross sectional area of the tested sample, and lav is the length of the middle magnetic field line of the magnetic core.
INSTRUMENTS AND EXPERIMENTAL TECHNIQUES
Vol. 55
No. 2
2012
MAGNETIC CORES MADE OF AN AMORPHOUS TAPE U, kV 25
I, kA 2.0
20 15
Number of samples 4
60
t = t1.5
U(t)
Number of samples
271
I(t)
1.5
50
1.0
40
AMW
MRTI 3
2
30 10
0.5 20
t=0 5
0
Idmg
0 400
600
800 t, ns
1000
–0.5 1200
1
10 0 620
660
700
740 780 W1.5, J/m3
820
0 860
Fig. 4. Current and voltage waveforms in the measurement scheme of the pulse characteristics of the magnetic cores.
Fig. 5. Distribution of the specific losses W1.5 of magnetic cores of the LIA injector inductors.
The pulse demagnetizing field Hdmg with duration of ~300 μs, which is formed from the voltage source Udmg through the decoupling choke is as great as 200 A/m. This large value is caused by the double pulse operation mode of the generators of the acceler ator. It is necessary to remagnetize the core in an inter val between pulses to form a train of two pulses with an interval from 2 μs across the inductor. The high remag netization current is ensured by the current in the demagnetization circuit [9].
induction rise rate dB dt av = B(t1.5) t1.5 was kept unchanged (≈5 T/μs).
The peaktopeak induction with allowance for a fall is 2.1 T. The small peaktopeak working induction selected for the 2НСР alloy is caused by several rea sons: (i) need for the fast remagnetization of the inductors (this process is delayed if the magnetic core starts to saturate when the pulse falls down); (ii) mar ginal induction allows a future increase in the pulse duration of the accelerator up to 250–300 ns; and (iii) when the peaktopeak induction is small, the addi tional supply of the generator with the magnetization current (including nonlinear) is not high. Therefore, the FL adjustment is simplified, and the requirements for the switch and insulating element of the double pulse generator (magnetic “valve”) are weakened [9]. The value of the additional load created by the inductor for the generator can be estimated by calcu lating the energy consumed by the inductor for the pulse time: W1.5 =
∫
t1.5 0
U (t)(I (t) − I dmg )dt Vkch , Idmg is
the demagnetization current at the measurement moment (Fig. 4), and V is the volume of the magnetic core. The integration time is counted from the moment when the voltage U(t) starts rising to moment t1.5 , when the peaktopeak induction in the core is 1.5 T (Fig. 4). During each measurement, the average INSTRUMENTS AND EXPERIMENTAL TECHNIQUES
Figure 5 shows the distribution of specific loss val ues W1.5 of the manufactured magnetic core samples. It is possible to select two lots among MRTI magnetic cores. The four samples of the first lot have larger losses and they scatter more than the magnetic cores of the second lot. This can be attributed to finalizing of the technology for largesize product winding and annealing. In calculations of the induction and specific losses of the AMW compound magnetic cores, the ironfill ing factor was assumed to be 0.7. This allows one to compare absolute losses of the magnetic cores by the calculated values W1.5. Thus, from the viewpoint of the load of the generator, the magnetic cores made with out insulation do not differ on average from the mag netic cores with insulation (average energy loss value is approximately the same). However, since the parame ters scatter from delivery to delivery, the loss scatter of the compound magnetic cores was ~25%. As it was expected, the nonlinearity of the load is higher due to the low squareness ratio. Figure 6 shows curves B(H) obtained for the two best magnetic core samples (with isolation and without it). The need for introducing a stronger nonuniformity in the FL of the generator for compensating the top fall leads to deterioration of the load matching and to pulse tailing. To determine the possibility of increasing the oper ating induction for forming pulses with durations up to 300–350 ns, the magnetic cores produced by the MRTI and AMW were tested in the mode with induc tion drop up to 2.5 T, when dB dt av was up to 6.7 T/μs (Fig. 7). It is apparent that the magnetic cores of both Vol. 55
No. 2
2012
272
AKIMOV et al.
B, T
B, T
2.0
2.8 2.4
1.6
2.0 1.2
1.6 AMW 645 J/m3
0.8
2.1 T 5.9 T/µs
1.2 MRTI 701 J/m3
0.4 0 –100
2.5 T 6.7 T/µs
0.8
1.8 T 5.1 T/µs
0.4 0
100
200
300
400
500 600 H, А/m
Fig. 6. Pulse magnetization curves of the best magnetic core samples made by the MRTI and AMW.
types have a oneandahalf marginal induction and can be used for forming longer or highvoltage pulses. One of the compound’s magnetic cores has passed longterm parameter stability tests. During the tests, the magnetic core operation mode was close to the rated operation mode in the injector (ΔB = 1.6 T, dB/dtav = 5.4 T/μs, and Hdmg = 200 A/m). The mea surements of the W1.6 value (loss energy in the mag netic core at ΔB = 1.6 T) showed that the change of its value did not exceed the measurement error (±3%) for 2.5 × 105 pulses. The tests with about oneandahalf overvoltage also did not lead to increasing losses. However, during the tests of the cores for the mag netic valves, which had been made by the same tech nology, an irreversible increase in losses in the mag netic core was observed at an induction rise rate of ~10 T/μs, amounting to the double excess of the rated value. Most likely, the parameter degradation is related to electrical breakdowns between layers and further decrease in resistances between tape layers. This phe nomenon was discovered for samples with dimensions of 214 × 76 × 5 mm. For this diameter ratio (almost 3), the total voltage is redistributed between the outer lay ers of the core in the process of saturation of its inner layers. In this case, the voltage between layers can be increased multiple times. For the inductor magnetic cores, taking into account the small operating induction drop and small diameter ratio, this effect does not become apparent, if the demagnetization is normal. The estimate of the maximal admissible turn voltage of the inductor mag netic cores can be approximately the doubled rated value.
0
0
200
400
600
800
1000 1200 H, А/m
Fig. 7. Pulse magnetization curves of the magnetic cores of the LIA injector in the mode with increased peaktopeak inductions at demagnetizing field Hdmg = 100 A/m; the grey lines is the magnetic core of the MRTI, and the black lines are the compound magnetic core of the AMW.
The inductor system of the LIA injector was assem bled of eight inductors on the MRTI magnetic cores and 88 inductors with the AMW magnetic cores. Dur ing commissioning, the rated voltage was obtained at the first half of the injector (1 MV) (Fig. 8) and approximately 80% from the rated value on the second half of the injector (this limitation is related to the insufficient vacuum electric strength of the accelerat ing tube and not to the inductor parameters). CONCLUSIONS The magnetic cores of the inductor system were manufactured for the injector of the linear induction U, kV
I, kA 0
0 –200
–0.4
–400
–0.8 I
–600 –800
–1.2 –1.6
U
–2.0
–1000 0
100
200
300 t, ns
400
500
Fig. 8. Voltage and current oscillograms of the vacuum diode of the LIA injector.
INSTRUMENTS AND EXPERIMENTAL TECHNIQUES
Vol. 55
No. 2
2012
MAGNETIC CORES MADE OF AN AMORPHOUS TAPE
accelerator by the AMW and MRTI. The technology of manufacturing pulse magnetic cores with tape insu lation completely, which was tested earlier, meets the requirements of both the existing accelerator and pro jected LIA with increased pulse durations. The sim pler technology for manufacturing compound tape magnetic cores without layer insulation was also approved. The average losses for their remagnetization are comparable with losses in the magnetic cores with insulation. The discovered drawback related with the turn voltage limitation for this technology does not allow one to use it for the magnetic cores operating at induction rise rates of >10 T/μs. As a whole, the inductor system on the magnetic cores of both types ensured the design parameters and was used in the commissioning of the LIA injector. ACKNOWLEDGMENTS This work was supported in part by the Ministry of Education and Science of the Russian Federation, project nos. 14.740.11.0160 and P2493.
INSTRUMENTS AND EXPERIMENTAL TECHNIQUES
273
REFERENCES 1. http://www.amet.ru 2. http://www.metglas.com 3. http://www.vacuumschmelze.de 4. Smith, C.H., Turman, B.N., and Harjes, H.C., IEEE Trans. Electron Devices, 1991, vol. 38, no. 4, p. 750. 5. Mamaev, G., Bolotin I., Ctcherbakov A., et al., Proc. 1997 Particle Accelerator Conf., Vancouver, Canada, 1997, p. 1313. 6. Molvik, A.W. and Faltens, A., Physical Review Special Topics, Accelerators and Beams, 2002, vol. 5, no. 8, p. 080401. 7. Bolotin, I., Mamaev, G., Mamaev, S., et al., Proc. 1999 Particle Accelerator Conf., New York, 1999, p. 1482. 8. Nathasingh, D.M., Smith, C.H., and Datta, A., IEEE Trans. Magnetics, 1984, vol. MAG20, no. 5, p. 1332. 9. Akimov, A.V., Bak, P.A., Korepanov, A.A., et al., Vest nik Nov. Gos. Univ., Ser. Fiz., 2008, vol. 3, no. 4, p. 68.
Vol. 55
No. 2
2012