Published online by Cambridge University Press: 05 December 2005
A pulsed power generator by inductive voltage adder, versatile inductive voltage adder (VIVA-I), which features a high average potential gradient (2.5 MV/m), was designed and is currently in operation,. It was designed to produce an output pulse of 4 MV/60 ns by adding 2 MV pulses in two-stages of induction cells, where amorphous cores are installed. As a pulse forming line, we used a Blumlein line with the switching reversed, where cores are automatically biased due to the presence of prepulse. Good reproducibility was obtained even in the absence of the reset pulse. Within ∼40% of full charge voltage, pulsed power characteristics of Marx generator, pulse forming line (PFL), transmission line (TL), and induction cells were tested for three types of loads; open-circuit, dummy load of liquid (CuSO4) resistor, and electron beam diode. In the open-circuit test, ∼2.0 MV of output voltage was obtained with good reproducibility. Dependences of output voltage on diode impedances were evaluated by using various dummy loads, and the results were found as expected. An electron-beam diode was operated successfully, and ∼18 kA of beam current was obtained at the diode voltage of ∼1 MV.
Recently, the development of high-energy pulsed power generator was widely required (Mesyats, 2003; Mesyats et al., 2003; Bastrikov et al., 2003; Ozur et al., 2003) in the applications of a driver for inertial confinement fusion (Hora, 2004), pulsed microwave, free electron laser (Alesini et al., 2004), medical treatment, and nuclear spallation using pulsed proton beams (Nation, 1979). An inductive voltage adder system is a promising technology to obtain a high-voltage pulsed power, particularly in the high-current region in the order of several kilo-amperes. From such a viewpoint, a lot of high voltage accelerators using the inductive voltage adder system was developed in the past (Smith et al., 1988; Ramirez et al., 1988, 1992; Hoppe et al., 1998). Highly repetitive rate operation can be expected as well. The averaged potential gradient was, however, not high enough. For example, the averaged potential gradient was limited to 1.7 MV/m on “HERMES-III” at Sandia National Laboratories (Ramirez et al., 1992). If we are able to achieve a much higher potential gradient, it will be constructed to be more compact and realistic to various industrial applications.
From such a viewpoint, we have attempted to develop and demonstrate a compact pulsed power generator, VIVA-I, which features and is characterized by a high potential gradient. We designed a high potential gradient of 2.5 MV/m by using amorphous metallic cores. Experimentally, such an induction accelerator was successfully developed, of which details, including design concept and pulsed power characteristics, will be reported in this paper (Masugata et al., 1988).
Figure 1 illustrates the basic principle of a pulsed inductive voltage adder system. Due to rapid change of the magnetic flux of the primary circuit, a high-voltage output will be induced in the secondary circuit. Experimentally, primary pulses from a pulse generator are fed to two induction cells, in which magnetic cores are installed. Figure 1b shows the equivalent circuit. Two primary pulses are added in series on the secondary circuit.
The inductive voltage adder system, VIVA-I, consists of a Marx generator, a Blumlein pulse forming line (PFL), a pulse distributor, four transmission lines (TLs), two induction cells, and a diode. The primary energy storage of VIVA-I is the Marx generator of 17-stages, which is exactly the same as that used in a pulsed-power generator, ETIGO-I (Yatsui et al., 1985). The stored energy of the Marx generator is 43 kJ at a charging voltage of ± 75 kV, which yields 2.55 MV of output voltage.
Both PFL and pulse distributors are located in de-ionized water. Figure 2 shows the cross-sectional view of this section. The PFL used is cylindrical Blumlein line (BL) with switching reversed. The intermediate conductor (IMC) of the BL is directly charged by the Marx generator in ∼500 ns. The BL switch is a self-breakdown SF6-gas switch located between IMC and the outer conductor (OC). The design parameters of the BL output are as follows: ZBL (output impedance) ∼ 12.5 Ω, τBL (pulse width) ∼ 60 ns, and VBL(max) (output voltage) ∼ −2 MV.
The output pulse of the BL is fed to the pulse distributor through four-channel pre-pulse switches (water, self-breakdown), and divided into four 50-Ω pulses. The pulses are then transported to the inductive voltage adder section through four oil-filled coaxial TLs.
Figure 3 shows the cross-sectional view of the adding section, which consists of two induction cells and a diode. The outer diameter and length of each cell are ∼1 m and ∼0.8 m, respectively. Two 50-Ω TLs are connected to each cell in parallel. Thus, the average potential gradient of ∼2.5 MV/m is achieved by the application of 2 MV. Amorphous cores are immersed in oil and the oil-vacuum interface is made of an acrylic insulator. Properties of amorphous metallic cores installed inside the cells are presented in Table 1. The amorphous tape is 21-μm thick and the insulator is 6-μm thick, yielding a filling factor of ∼70%. The receptivity of the tape is ∼130 μΩ·m
Figure 4 shows a typical B-H curve. As seen from Figure 4, the core features a large flux swing that is ΔB = Bs + Br ≥ 1.75 T, where Bs and Br are the saturated flux density and residual flux density, respectively.
The output voltage from two induction cells is added in series by a center conductor/MITL (magnetically insulated transmission line). Such an added pulse is finally delivered to the diode. In the initial load test of the system, we installed a resistor of liquid (CuSO4) solution or electron-beam diode.
Figure 5 presents the equivalent circuit of the whole system of VIVA-I. Several results of the initial tests are presented as follows.
Initial tests of the whole system were carried out at Vch (charging voltage of Marx generator) ∼30 kV, which is less than 40% of the full charge voltage. To obtain characteristics of magnetic cores in the saturated level within the limited range of Vch, we first installed two cores to each cell. As a result, the value of Vτ of each cell was reduced to ∼0.067 V × s instead of the initially designed value of ∼0.1 V × s.
At first, the open-circuit tests were carried out by removing the center feeder/MITL of the adder. Figure 6 shows typical waveforms at Vch = 30 kV. As seen from Figure 6a, the charging voltage of PFL reaches 1.1 MV at t ∼ 600 ns, and then the BL switch is closed. From Figure 6b of the applied voltage of the cell (Vind), pre-pulse voltage of ∼100 kV is seen to be present for ∼600 ns, and after that negative main pulse of Vind ∼ 1.0 MV is obtained with the pulse width of ∼75 ns. Due to the presence of the pre-pulse above, cores are automatically biased, and good reproducibility was obtained even in the absence of the reset pulse. Figure 6c shows the excitation current of the cell (Iex). We see Iex builds up at t ∼ 610 ns, and gradually increases and reaches ∼8 kA at the peak of Vind at ∼700 ns. After t ∼ 780 ns, Iex significantly increases and reaches the peak value of 25 kA at t ∼ 850 ns, indicating the saturation of the magnetic cores at t ∼ 780 ns.
The integral of Vind from t ∼ 610 ns to ∼780 ns is calculated to be ∼0.065 V × s, which reasonably agrees with the previously stated Vτ value (∼0.067 V × s). As seen in the above agreement, it is clearly evident that the inverse excitation takes place by the prepulse.
Figure 7 shows the dependence of VPFL, Vind, and Iex on Vch. We see that both VPFL and Vind increase almost proportionally to Vch. Such a linear increase in Vind on Vch suggests that the magnetic cores are not saturated before the peak of Vind.
Pulse-power characteristics were studied by using a resistance of liquid (CuSO4) solution as a dummy load. Three kinds of resistances (R[ell ] = 50, 120, and 580 Ω) were used in this test, where the dummy load was installed within the induction cell.
Figure 8 shows typical waveforms of VPFL, Vind, Vd, Icell, and Id. From Figure 8, we obtain the following peak values:
It is noted here that Vind is reduced to almost half of VPFL. This is due to the fact that the input impedance of the induction adder decreases with the connection of the dummy load.
Figure 9 shows Vind and Vd (the peak diode voltage applied to the load) as a function of Vch. We see Vind and Vd are almost proportional to Vch. The voltage difference between Vind and Vd is basically due to the inductive voltage drop in the center conductor, L[ell ](dId /dt). We have calculated L[ell ] from the dimension of the center conductor to be ∼450 nH, which clearly corresponds with that estimated from these experimental results by using the relationship,
The electron-beam generation experiment was carried out by using an annular diode (cathode; carbon, anode; Ti foil of thickness ∼20 μm), the outline of which is shown in Figure 10. The inner and outer diameters of the cathode are ID = 58 mm and OD = 76 mm, respectively, and the cathode-anode gap length dK-A = 17 mm. The axial magnetic field of BZ ∼ 0.84 T is applied to the diode and the drift region. The electron-beam current is measured by a Faraday cup placed at ∼160 mm downstream from the anode.
Figure 11 shows these typical waveforms of VPFL, Vind, Icell, Id, and Ib (beam current). Clearly, the electron-beam is generated with Vd ∼ 1 MV (estimated from Vind and Id by using Eq. (1)), Id ∼ 20 kA, and Ib ∼ 18 kA.
By moving a thermo-sensitive paper in the axial direction, we have examined the behaviors of the axial transport of the electron beam. Figure 12 shows the results of various beam patterns in the z-axis. Although the electron beam can be transported downstream in the strong B-field region with almost the same diameter as the electron-beam diode, it begins to expand with larger diameter in the downstream with the weaker B-field.
A versatile induction voltage adder, VIVA-I, was successfully constructed. It features a high potential gradient of ∼2.5 MV/m. The initial operating tests were carried out with less than 40% of full charge voltage. As a pulse forming line, we utilized a Blumlein line with the switching reversed, where the amorphous cores are automatically biased due to the presence of the prepulse, and good reproducibility was obtained even in the absence of the reset pulse. In the open-circuit test, ∼2.0 MV of output voltage was obtained with good reproducibility. Dependences of output voltage on diode impedances were evaluated by using various dummy loads, and the results were found as expected. An electron-beam diode was successfully operated. The electron beam transport was observed with the beam current of ∼18 kA.
This work was partly supported by a grant-in aid for scientific research from the Ministry of Education, Science and Culture of Japan.