The High Current, Fast, 100 ns, Linear Transformer Driver (LTD) Developmental Project at Sandia National Laboratories 1

Size: px

Start display at page:

Download "The High Current, Fast, 100 ns, Linear Transformer Driver (LTD) Developmental Project at Sandia National Laboratories 1"

Dwayne Bradford
5 years ago
Views:

1 Oral Session The High Current, Fast, 100 ns, Linear Transformer Driver (LTD) Developmental Project at Sandia National Laboratories 1 M.G. Mazarakis *, A.A. Kim **, K.R. LeChien *, W.E. Fowler *, V.A. Sinebryukhov **, W. Long *, M.K. Matzen *, D.H. McDaniel *, R.McKee *, J.L. Porter *, K.W. Struve *, W.A. Stygar *, P.E. Wakeland ***, K.S. Ward ***, and J.R. Woodworth * * Sandia National Laboratories Albuquerque, N.M USA Phone: (505) , (505) , mgmazar@sandia.gov ** Institute of High Current Electronics SB RAS, 2/3, Academichesky ave., Tomsk, , Russia *** Ktech Corporation, Albuquerque, NM, USA Abstract Sandia National Laboratories, Albuquerque, N.M., USA, in collaboration with the High Current Electronic Institute (HCEI), Tomsk, Russia, is developing a new paradigm in pulsed power technology: the Linear Transformer Driver (LTD) technology. This technological approach can provide very compact devices that can deliver very fast high current and high voltage pulses straight out of the cavity with out any complicated pulse forming and pulse compression network. Through multistage inductively insulated voltage adders, the output pulse, increased in voltage amplitude, can be applied directly to the load. The load may be a vacuum electron diode, a z-pinch wire array, a gas puff, a liner, an isentropic compression load (ICE) to study material behavior under very high magnetic fields, or a fusion energy (IFE) target. This is because the output pulse rise time and width can be easily tailored to the specific application needs. In this paper we briefly summarize the developmental work done in Sandia and HCEI during the last few years, and describe our new MYKONOS Sandia High Current LTD Laboratory. 1. Introduction The fast, 100 ns, LTDs are a new paradigm in pulsed power technology. They were invented in the HCEI through the Sandia suggestion, encouragement and vigorous support. The salient feature of the approach is switching and inductively adding the pulses at low voltage straight out of the capacitors through low inductance transfer and soft iron core isolation [1, 2]. The pulse forming capacitors and switches are enclosed inside the accelerating cavity. An LTD cavity can produce large current outputs by feeding each inductive cavity core with many capacitors connected in parallel in circular arrays. In principle currents of up to 2 MA can be obtained. The limitation is the size of the cavities which, with the present capacitor and switch dimensions, cannot in practice exceed the 3 4 m in diameter. Otherwise a cavity becomes cumbersome and difficult to handle. If larger currents and higher voltages are required, voltage adder modules of many LTD cavities can be connected in parallel to the load. Utilizing the presently available capacitors and switches, we can envision building the next generation of fast pulsed power drivers without large Marx generators and voluminous oil-water tanks and pulse forming and pulse compression networks as is the case with the present technology drivers. The LTD devices can be multi-pulsed with a repetition rate up to 0.1 Hz or higher as required for the IFE driver power plants. This important capability opens the path for a pulsed power inertial confinement energy-producing facility. In this paper we briefly summarize the developmental work done during the last few years on high current, fast LTDs. To date we have completed the experimental evaluation of two (LTD I, LTD II) 500-kA cavities in both single and rep-rated modes, we have studied the performance of five 1-MA, LTD cavities individually and in voltage adder configuration with both resistive and vacuum electron diode loads, and we are currently starting the tests of a two cavity voltage adder in the MYKONOS laboratory. The five-cavity voltage adder coaxial transmission line was overmatched to the cavities impedance and was vacuum insulated, while the MYKONOS voltage adder is deionized water insulated. The ultimate goal of the present work is, following the individual testing of each cavity type, to further evaluate their performance combined in voltage adder configured modules. If the tests are successful, those modules could be considered as building blocks for larger, high current and high voltage accelerators. In section II we describe the first two types of fast, high current LTD cavities constructed and operated. In section III we present experimental results of each type of cavities tested individually with resistive loads and SF6 insulation. Section IV describes the experimental work done with a 5-1-MA cavity voltage adder connected both to a vacuum electron diode and a resistive load, and in section V we describe the new 1 MV, 1 MA, MYKONOS laboratory. Finally, in section VI 1 Sandia is a multi-program laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy s National Nuclear Security Administration under Contract DE-AC04-94AL

conventional induction accelerators like HERHMES III, ATA etc. The LTD cavities enclose the entire pulse-forming network that generates the output pulse.

2 we give an up to date summary of the work done with high current fast LTDs. a b Fig. 1. The 0.5-MA (a) and 1-MA (b) LTD cavities 2. High current fast LTD cavity designs The LTD cavities constitute a new type of induction accelerator cavity where the high voltage pulses are not fed from an outside pulse-forming network as with the conventional induction accelerators like HERHMES III, ATA etc. The LTD cavities enclose the entire pulse-forming network that generates the output pulse. This pulse is applied across the insulator that separates the anode and cathode output electrodes (A-K gap) of the cavity. The cavity shape is similar to a flattened doughnut with the axial A-K gap at the center of the inside cylindrical surface. At all times the walls of the cavities are at ground potential. In Fig. 1 we present two different sizes LTD cavities [3, 4]. The top metal cover and the plastic insulator that insulates the charged parts from the cavity top wall are removed in order to show the cavity interior network of switches and capacitors. The cavity of Fig. 1, a can deliver a 0.5-MA, 100-kV pulse to a matched 0.2 load. It contains two circular arrays of 40-nF, 25-nH single-ended capacitors. In Fig. 1 only the top array is seen. The bottom array is separated from the top by a ~ 1 cm plastic insulator plate. The top capacitors can be charged up to kV maximum charge and the bottom ones up to 100kV. Each pair of negatively and positively Pulsed Power Technology 256 charged capacitors is connected in series with a separate switch positioned vertically and capable of holding 200-kV DC potential difference. This basic unit, named brick, composed of two capacitors and one switch connected in series, defines the rise-time, current and period of the cavity output pulse. The capacitors and switches are the only two active elements, which are repeated many times in a cavity interior depending on the required output current pulse amplitude. The capacitors are charged with DC power supplies. The charging currents do not exceed 30 ma. In Fig. 1, a the cavity is fitted with a liquid resistive load enclosed in two concentric plastic cylinders located in the central hole of the cavity toroid. In testing the cavities we utilized KBr water solutions. The 0.5-MA and 1-MA cavities (Fig. 1, a and b) contain 40-nF capacitors. The switches are pressurized with dry air. Depending on the capacitor charging voltage, the switch operating pressure can vary between a few atmospheres and up to 6 atmospheres. We have developed a repetition system that can recycle the cavities very fast [5, 6]. A rep-rate close to 0.1 Hz was achieved. Both types of constructed cavities have the same axial length of ~ 22 cm. The 1-MA cavity is the largest (3 m in diameter), while the two 0.5-MA LTD I and LTD II cavities have a smaller diameter and approximately equal to 2 m. Cavities LTD I and LTD II have 20 bricks while the 1-MA cavity contains 40 bricks. Each brick provides a maximum of approximately 25 ka current to a matched load. The cavities, in addition to the solid plastic insulating plates, are also filled with LUMINOL oil for additional insulation from the cavity walls. 3. Experimental evaluation of each cavity performance with resistive loads Both LTD I and LTD II cavities were operated at single as well as rep-rated modes. All cavity support systems such as capacitor charging, switch pressurization, switch vacuum purging, pre-magnetization of the ferromagnetic cores and triggering systems were made repetition rate capable [5, 6]. In all the experiments the cavity output load was a liquid resistor installed at the center of the cavity (Fig. 1, a). The research results with LTD I cavity are summarized as follows: 1) We fired 13,000 shots at ± kv charging and at rep-rate up to 5 shots per minute without experiencing any component overheating or problems with the switches, the capacitors, and the supporting automated systems. 2) The measured cavity output jitter was quite small and less than 2 ns (1 ). 3) We came very close to the IFE required repetition rate. Towards the end of this campaign we were able to fire every 10.3 s, which is equivalent to Hz. 4) The pulse-to-pulse reproducibility was excellent with variations less than 1% (1 ). 5) Finally during the 13,000 shots, following the determination of the switch optimum operating pressure, not switch pre-fires were recorded [3].

3 Oral Session The experimental work with the LTD II cavity had different goals. In addition to firing a large number of shots for component longevity verification at ± 100 kv charging, the cavity performance for different switch pressures was studied. Namely the switch pressure was varied from 53 psia (= 366 kpa) to 67 psia (= 463 kpa) in steps of 2 psia (~ 14 kpa). For each pressure setting 1,000 shots were fired at ± 100- kv charging. Fig. 2 compares a typical output voltage trace, at ± 100 kv charging, with Pspice [7] circuit code calculations. The circuit utilized here in addition to the cavity LCR components it also included the core losses and load inductance (Fig. 3). Load voltage, kv experiment pspice t (ns) Fig. 2. LTD II load voltage trace (dot trace) compared with Pspice simulation (solid trace) For every shot the following voltage pulse output parameters were recorded and statistically analyzed: switch closure time (the time interval between the arrival of the trigger pulse on the switches and the start of the output pulse), switch closure jitter, voltage pulse rise time, output voltage shape, fall time, FWHM, and amplitude; similar parameters were recorded for the output current pulses. As the switch pressure slightly increased above 65 psia (= 450 kpa), the trigger jitter, pulse rise time, FWHM, and standard deviations increased. Our data suggest that the switches perform equally well in the kPa pressure range studied for a ± 100-kV charging operation. At this pressure range the pulse rise and fall times, FWHM, trigger jitter, and amplitude standard deviations were approximately the same with a very slight increase going from lower to higher pressures. However, the switch closure times (Fig. 4) increased from lower to higher pressures by approximately 10 ns. Following conditioning of the switches at the beginning of the experimentation with approximately 500 shots, no pre-fires were observed in the entire pressure range studied. It is interesting to note that although the studied pressure range varied from 5 to 32% above the self break pressure for a 200 kv voltage across the switches (Fig. 5), the output pulse characteristics remained practically unchanged [3]. Switch closure time, ns Switch pressure, kpa Fig. 4. Switch closure time as a function of the switch pressure Switch self break voltage, kv Switch pressure, kpa Fig. 3. LTD II and 1-MA LTD equivalent circuit including core losses and load inductance. For the case of LTD II, input circuit parameters were the following: C = 400 nf, L 1 = 12 nh, R 1 = Ohm, R C = 1.3 Ohm, L 2 = 0.7 nh and R = 0.2 Ohm and the capacitor charge voltage 200 kv = ( 100) kv 257 Fig. 5. Self-break curve of the 20 LTD II cavity switches The 1-MA cavities were tested at HCEI [4]. Fig. 6 shows the output voltage and current on a match 0.1 I load, MA I load, MA V load, kv t, ns Fig. 6. Current and voltage output of the 1-MA LTD cavity with 0.1 resistive load V load, kv

Pulsed Power Technology resistive load. The load was placed inside the cylindrical opening of the cavity in a similar way as for the LTD I and LTD II cavities.

4 Pulsed Power Technology resistive load. The load was placed inside the cylindrical opening of the cavity in a similar way as for the LTD I and LTD II cavities. In order to avoid flashovers and tracking of the axial A-K gap insulator the volume between the A-K gap and the resistor outside wall was filled with SF6. Experiments were conducted with different resistive loads at 100 kv charging. The optimum output cavity impedance [4], which maximizes the output power to ~ 96 GW, was found to be ~ Five 1-MA LTD vacuum voltage adder experiments We built ten 1-MA LTD cavities at the HCEI [4]. The first five were assembled in a voltage adder configuration before being shipped to Sandia. The coaxial transmission line of the voltage adder was vacuum insulated and connected first to an electron vacuum diode (Fig. 7) and later to a resistive load. Fig. 7. Five 1-MA LTD cavity voltage adder At ± 90 kv charging the voltage reached a maximum value of ~ 400 kv at ~ 100 ns. At the same time the current peaked at ~ 800 ka, while the electron beam power became ~ 320 GW. Ud_exp Ud_psp Id_exp Id_psp Fig. 8. Five 1-MA LTD cavity voltage adder results compared with numerical simulations. The diode voltage for clarity is inverted. The simulations were done for U CH ± 90 kv Since all the cavities were triggered simultaneously, the Pspice model included the inductance (18 nh) of the voltage adder coaxial line. Because of that the output voltage was lower and equal to ~ 400 kv. Figure 8 shows the diode voltage U D and current I D traces for an LTD cavity charge U CH = 90 kv and a diode A-K gap of 1.4 cm. The experimental results are compared with Pspice [7] simulations [4]. With the electron diode load, the voltage adder was operated at charge voltage between 80 to 90 kv and for a number of AK gap settings varying from 1.0 cm to 1.7 cm. During the resistive load experiments we fired a total of 500 shots; 300 of them were at 100 kv charging. The MYKONOS LTD Laboratory experiments The ten 1-MA cavities were originally designed and built to run in a vacuum or Magnetic Insulated Transmission Line (MITL) voltage adder configuration. However, by the time we received them from Tomsk in Sandia, we decided to use de-ionized water as voltage adder insulator. Our motivation was to test the advantages of water insulation as compared to MITL transmission approach. It was hoped that the vacuum sheath electron current losses and pulse front erosion would be avoided without any new difficulties caused by the de-ionized water. Special care has been taken to eliminate air bubbles in the voltage adder. To that effect, a number of modifications were implemented in the cavity structure to make them capable of operating in a de-ionized water environment. The MYKKONOS LTD module is the first voltage adder ever built to be de-ionized water insulated. The water insulation may have substantial advantages as compared to MITL, especially in long 5 6 MV accelerators [8]; as it is well known in the scientific community, a pulse propagating in an MITL suffers frond erosion resulting in pulse width decrease, which in long lines could reach up to 20%. In addition, the sheath electron current, which could be about 1/3 of the total pulse-current, unless re-trapped by very low impedance or inductance loads can be lost on the transmission line anode walls, further reducing the energy transport efficiency and causing severe hardware damages. On the other hand, the major concerns of utilizing water in an enclosed system such as that of a voltage adder are air bubble formation and metal wall erosion. In order to make the cavity water compatible we had to implement the following modifications: the slant angle of the plastic A-K gap insulator was reduced from 45 to 4 and cut in the opposite (= 94 ) than the 45 direction. The stainless steel surfaces of the cavity exposed to the de-ionized water and the cathode stalk were passivated to avoid corrosion and water resistance degradation. A 6.35-mm diameter channel was drilled on the top side of the A-K gap cathode electrode connecting to the outside of each cavity in order to facilitate the removal of bubbles possibly formed at the A-K gap triple points. The channel is fitted with the same diameter stainless steel tube that extends all the way to the outside top of the cavity. The cavities are hanged vertically from the rails as shown in Fig. 9. This tube passes through 258

The MYKONS two LTD cavity voltage adder Oil reservoir Water reservoir rep-rated tests of the two cavity labview operating system achieved a rep-rate of 8.57 shots a minute.

5 Oral Session a gap left between two adjacent cavities when they are stacked together into the voltage adder. A small water reservoir is connected at the upper end of the tube which can be pumped if necessary to further force the bubbles to exit the A-K gap region (Fig. 10). Fig. 9. The MYKONS two LTD cavity voltage adder Oil reservoir Water reservoir rep-rated tests of the two cavity labview operating system achieved a rep-rate of 8.57 shots a minute. The operation and data acquisition system is similar to that of LTD I and LTD II. However the complexity, the sophistication, and the number of operations are increased to handle the greater number of cavities (10), plus the de-ionized water, liquid resistor, and oil recirculation systems. The experimental plan with the two cavity voltage adder is as follows: First we will fire approximately ~ 500 shots to condition the switches, then will fire 1000 shots at rep-rate frequency of up to 0.1 Hz. If no problems are encountered, we will continue with pulse shaping experiments, staggering the two cavity triggering and also by time varying the firing of a number of brick groups. Finally we will proceed with the assembly of the10-cavity, 1-TW voltage adder. 6. Summary An extensive evaluation of the LTD technology is being performed at SNL and the High Current Electronic Institute (HCEI) in Tomsk Russia. Two types of High Current LTD cavities (LTD I II, and 1-MA LTD) were constructed and tested individually and in a voltage adder configuration (1-MA cavity only). All cavities performed remarkably well and the experimental results are in full agreement with analytical and numerical calculation predictions. A two-cavity voltage adder is been assembled and currently undergoes evaluation. This is the first step towards the completion of the 10-cavity, 1-TW module. This MYKONOS voltage adder will be the first ever IVA built with a transmission line insulated with deionized water. The LTD II cavity renamed LTD III will serve as a test bed for evaluating a number of different types of switches, resistors, alternative capacitor configurations, cores and other cavity components. Experimental results will be presented at the Conference and in future publications. Bubble removal pipe Fig. 10. Schematic diagram of an 1-LTD cavity modified in order to operate in a water insulated voltage adder configuration. Only a small section of the cavity upper part is presented in order to show the bubble evacuation pipe and the water reservoir. The larger reservoir is for the oil Presently we have assembled and are testing a two-cavity water insulated voltage adder. This is the first step towards the completion of the 10-cavity, 1-TW module. Figure 9 shows the rails and the two cavities in place. The current through the load flows radially from the anode cylinder to the cathode stalk. The radial geometry was selected in order to minimize the inductance and capacitance of the load. The estimated load resistance is 0.22, the capacitance 3 nf and the inductance 1.1 nh. A new LabVIEW [9] software program was written capable of operating up to 10 cavity voltage adders. Presently the rep-rate frequency is set to 6 shots per minute. The first simulated 259 References [1] A.A. Kim, High power driver for Z-loads (unpublished, 1998). [2] M.G. Mazarakis and R.B. Spielman, A Compact, High-Voltage E-Beam Pulser, in Proc. 12th IEEE International Pulsed Power Conference, Monterey, California, July 1999, IEEE N 99CH36358, 1999, p [3] M.G. Mazarakis, W.E. Fowler, A.A. Kim, V.A. Sinebryukhov, S.T. Rogowski, R.A. Sharpe, D.H. McDaniel, C.L. Olson, J.L. Porter, K.W. Struve, W.A. Stygar, and Joseph R. Woodworth, Phys. Rev. ST Accel. Beams 12, (2009). [4] A.A. Kim, M.G. Mazarakis, V.A. Sinebryukhov, B.M. Kovalchuk, V.A. Visir, S.N. Volkov, F. Bayol, A.N. Bastrikov, V.G. Durakov, S.V. Frolov, V.M. Alexeenko, D.H. McDaniel, W.E. Fowler, K. LeChien, C. Olson, W.A. Stygar, K.W. Struve,

6 Pulsed Power Technology J. Porter, and R.M. Gilgenbach, Phys. Rev. ST Accel. Beams 12, (2009). [5] S.T. Rogowski, W.E. Fowler, M.G. Mazarakis, D.H. McDaniel, C.L. Olson, R.A. Sharpe, and K.W. Struve, presented at the 15th IEEE Pulsed Power Conference, Monterey, CA, July [6] M.G. Mazarakis, W.E. Fowler, D.H. McDaniel, C.L. Olson, S.T. Rogowski, R.A. Sharpe, and K.W. Struve, High Current Fast 100-ns Driver Development In Sandia Laboratory, presented at the 15th IEEE Pulsed Power Conference, Monterey, CA, July [7] MicroSim Schematics, Version 6.1b November 1994, MicroSim Corporation, 20 Fairbanks, Irvine, California 92718, USA. [8] W.A. Stygar, M.E. Cuneo, D.I. Headley, H.C. Ives, R.J. Leeper, M.G. Mazarakis, C.L. Olson, J.L. Porter, T.C. Wagoner, and J.R. Woodworth, Phys. Rev. ST Accel. Beams 10, (2007). [9] LabView Manual (National Instruments), 260

TRANSMISSION LINE AND ELECTROMAGNETIC MODELS OF THE MYKONOS-2 ACCELERATOR*

TRANSMISSION LINE AND ELECTROMAGNETIC MODELS OF THE MYKONOS-2 ACCELERATOR* E. A. Madrid ξ, C. L. Miller, D. V. Rose, D. R. Welch, R. E. Clark, C. B. Mostrom Voss Scientific W. A. Stygar, M. E. Savage Sandia