High power microwave switching utilizing a waveguide spark gap

High power microwave switching utilizing a waveguide spark gap J. Foster, G. Edmiston, M. Thomas, and A. Neuber Citation: Rev. Sci. Instrum. 79, 114701 (2008); doi: 10.1063/1.3010381 View online: http://dx.doi.org/10.1063/1.3010381 View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v79/i11 Published by the American Institute of Physics.

Related Articles Total transmission and total reflection of electromagnetic waves by anisotropic epsilon-near-zero metamaterials embedded with dielectric defects J. Appl. Phys. 113, 084908 (2013) Superconducting coplanar waveguide resonators for low temperature pulsed electron spin resonance spectroscopy Rev. Sci. Instrum. 84, 025116 (2013) Electrically small, complementary electric-field-coupled resonator antennas J. Appl. Phys. 113, 084903 (2013) High power terahertz radiation source based on electron beam wakefields Rev. Sci. Instrum. 84, 022706 (2013) High-power microwave filters and frequency selective surfaces exploiting electromagnetic wave tunneling through -negative layers J. Appl. Phys. 113, 064909 (2013)

Additional information on Rev. Sci. Instrum. Journal Homepage: http://rsi.aip.org Journal Information: http://rsi.aip.org/about/about_the_journal Top downloads: http://rsi.aip.org/features/most_downloaded Information for Authors: http://rsi.aip.org/authors

Downloaded 01 Mar 2013 to 129.118.19.71. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

REVIEW OF SCIENTIFIC INSTRUMENTS 79, 114701 共2008兲

High power microwave switching utilizing a waveguide spark gap J. Foster, G. Edmiston, M. Thomas, and A. Neuber Center for Pulsed Power and Power Electronics, Department of Electrical and Computer Engineering, and Department of Physics, Texas Tech University, Lubbock, Texas 79409, USA

共Received 11 June 2008; accepted 9 October 2008; published online 11 November 2008兲 A reduction in the rise time of a 2.85 GHz high power microwave 共HPM兲 pulse is achieved by implementing an overvoltaged spark gap inside a waveguide structure. The spark gap is oriented such that when triggered, the major electric field component of the dominant TE10 mode is shorted. The transition from a transmissive to a highly reflective microwave structure in a relatively short period of time 共tens of nanoseconds兲 creates a means to switch multimegawatt power levels on a much faster timescale than mechanical switches. An experimental arrangement composed of the waveguide spark gap and a high power circulator is used to reduce the effective rise time of a HPM pulse from a U.S. Air Force AW/PFS-6 radar set from 600 ns down to 50 ns. The resulting HPM pulse exhibits a much more desirable excitation profile when investigating microwave induced dielectric window flashover. Since most theoretical discussions on microwave breakdown assume an ideal step excitation, achieving a “squarelike” pulse is needed if substantial comparison between experiment and theory is sought. An overview of the experimental setup is given along with relevant performance data and comparison with computer modeling of the structure. © 2008 American Institute of Physics. 关DOI: 10.1063/1.3010381兴 I. INTRODUCTION

II. EXPERIMENTAL SETUP

Mechanical1 and solid-state2 microwave switches have been previously developed for microwave switching. However, these devices are limited to slow closing times 关⬃3.5 ␮s 共Ref. 1兲兴 or low power 关⬃300 W 共Ref. 2兲兴 handling when compared with a spark gap microwave switch. Since there are no readily available commercial microwave switches that can close on command in a matter of tens of nanoseconds and handle multimegawatt power levels, a custom switch was designed and fabricated for our purposes. Previous research by Farber3 has shown that a 5.2–5.8 GHz megawatt level guided microwave pulse can be reflected with the use of a unipolar high voltage pulse triggered spark gap built into the structure of a waveguide. The switching time of such an apparatus was shown to be on the order of 10 ns. This method is capable of handling a wide range of rf power levels and frequencies. It is desired to reduce the slow rise time of a high power microwave 共HPM兲 pulse for the purpose of comparing the delay times for microwave induced surface flashover with breakdown theory that would like the temporal shape of the excitation as close as possible to an ideal square pulse. Achieving a short rise time requires the use of a fast shorting waveguide switch that can transition from a transmissive waveguide to a highly reflective structure in tens of nanoseconds. The operating principle behind the plasma switch is straightforward—the HPM pulse from the source 共magnetron兲 is allowed to reach maximum power amplitude and is then “switched” or reflected toward the atmospheric test section. A high power circulator and a waveguide “shorting” switch are the two major components required to accomplish this process.

The overall experimental setup is designed for the purpose of investigating HPM induced surface flashover in a simulated high altitude, low pressure environment.4 The experimental setup shown in Figs. 1共a兲 and 1共b兲 is configured with an EM-Design model 102574-2.85 high power circulator which is used for redirecting the microwave pulse after it is reflected by the plasma switch. The HPM pulse source produces up to ⬃5 MW of 2.85 GHz radiation with a rise time of 600 ns and a pulse duration of roughly 3.5 ␮s. With the plasma switch inactive, most of the HPM radiation will be absorbed by the load directly attached to the plasma switch assembly. Upon actively triggering the switch with a high voltage pulse, the plasma formation in the switch increases the microwave reflectivity of the device in a relatively short period of time. The high power circulator is configured such that this reflected HPM pulse will be directed toward the atmospheric test section 关see Fig. 1共a兲兴. Note that previously, this atmospheric test section was directly connected to the HPM pulse source.4 The rise time of the microwave power pulse at the test section was determined to be insufficient when compared with an ideal step excitation and is motivation for the present implementation that produces a roughly ⬃50 ns rise time in the test section. The plasma switch is gas isolated such that control of the gas pressure and type is possible. To avoid unintentional breakdown in the waveguide structure elsewhere, the waveguide is filled with the electronegative gas SF6. The switch is quartz tube enclosure and is filled with 1 bar of argon gas, observing several flush cycles to ensure that contamination with air is minimized. Argon is used primarily due to its relatively low ionization potential 关15.8 eV 共Ref. 5兲兴, lack of electronegativity, and lack of vibrational and rotational energy dissipation channels.

0034-6748/2008/79共11兲/114701/5/$23.00

79, 114701-1

© 2008 American Institute of Physics

Downloaded 01 Mar 2013 to 129.118.19.71. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

114701-2

Foster et al.

Rev. Sci. Instrum. 79, 114701 共2008兲

FIG. 3. LC inversion generator schematic.

FIG. 1. 共Color online兲 共a兲 Overall experimental setup. Red 共blue兲 arrows indicate power flow with plasma switch turned off 共on兲. 共b兲 Circuit equivalent of experimental setup.

At high enough densities 共⬃1011 cm−3兲, the charged particles in the plasma will reflect the 2.85 GHz electromagnetic radiation. The electrodes are vertically positioned in the high field region of the TE10 microwave mode in order to provide for maximum reflection of the HPM pulse when the switch is triggered. A cross-section drawing of the plasma switch is shown in Fig. 2. When the incident HPM pulse reaches its maximum amplitude, the switch is triggered with a high voltage pulse generator 共see Fig. 3兲. The plasma produced between the electrodes shorts the major electric field component of the

microwave, resulting in a reflected pulse that has a sharply reduced 10%–90% rise time of approximately 50 ns. The reflected fast rising pulse is then redirected through the fourport circulator onto a dielectric window for the purpose of studying microwave induced surface flashover. The circuit shown in Fig. 3 configuration is typically referred to as an “inversion generator.”6 The upper and lower electrodes of the plasma switch 共see Fig. 2兲 are connected to the output of the inversion generator at the two nodes on the far in Fig. 3. The primary energy storage components of the generator are the two 2.5 nF capacitors charged via a 30 kV dc power supply. The 2 kV input pulse depicted on the right side of Fig. 3 is required to “trigger” the inversion generator. The 2 kV pulse is generated by a prototype solid-state pulse generator properly delayed with respect to the HPM pulse. The purpose of the transformer on the right side of Fig. 3 is to step up the 2 kV trigger pulse to levels sufficient to breakdown the triggered spark gap pressurized to 3.4 bar N2. The combination of the ringing RLC discharge circuit created by closing the trigger spark gap 共C2, R4 + R5, and L1兲 in the bottom electrical mesh of the circuit and conservation of charge in the top half of the circuit 共the constant potential across C1兲 can produce up to 2Vdc, or twice the charging voltage, in the ideal case where R4 + R5 = 0 ⍀. The experimentally measured output of the inversion generator into an open load 共no breakdown in the plasma switch兲 peaks near −37.5 kV and is shown in Fig. 4 compared with simulation. The inductor, L1, is a prototype created by winding a ten American wire gauge around a 1 43 in. pipe template. The value of L1 is determined by matching simulation to experimentally gathered data and is approximately 1.8 ␮H. The resistor values R4 to R7 were chosen to reduce the noise produced by the LC inversion generator by dampening the oscillations within a few hundred nanosecond without deteriorating the LC output voltage significantly 共Fig. 6兲. III. EXPERIMENTAL RESULTS

FIG. 2. Cross-section view of plasma switch with the switch section enclosed by a thin-walled quartz tube. For switching, a negative high voltage pulse is applied to the top electrode.

The waveforms from a typically observed operation of the plasma switch are shown in Figs. 5共a兲 and 5共b兲, which indicate that the plasma switch significantly reduces the rise time of the microwave pulse. The measured reflected power from the plasma switch is approximately 70% of the peak power amplitude. For example, a reflected pulse with a magnitude of 2.5 MW would require 3.5 MW of incident power.

Downloaded 01 Mar 2013 to 129.118.19.71. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

114701-3

Microwave switching using a spark gap

FIG. 4. 共Color online兲 LC inversion generator output voltage for 22.5 kV charging.

FIG. 5. 共Color online兲 共a兲 Representative waveforms for reflection 共light gray: incident power; dark gray: transmitted power; black: reflected power; gray: absorbed power; red: ICCD camera gate兲. Note that the transmitted power signal was adjusted by 0.2 dB in postprocessing in order to bring the absorbed power level to zero for preswitching times. 共b兲 Waveform of 共a兲 on a smaller time frame 共light gray: incident power; black: reflected power兲.

Rev. Sci. Instrum. 79, 114701 共2008兲

FIG. 6. Observed voltage 共gray兲 and current 共black兲 waveforms across the gap of the plasma switch. Charging voltage on C1 and C2 is 26 kV.

The remaining power of the microwave pulse is transferred to the free charge carriers in the plasma, causing absorption primarily via electron collisions, leading to electronic and ionizing transitions of the argon gas. In the case of a diatomic switching gas, rotational and vibrational excitations are pathways for power absorption as well. The representative power waveforms shown in Figs. 5共a兲 and 5共b兲 indicate that the electrodes of the plasma switch are passively reflecting ⬃0.25 MW of the 3.75 MW incident HPM pulse before the plasma switch is discharged. After the plasma switch discharges, the transmitted power drops to approximately zero and the reflected power sharply increases to ⬃2.5 MW in a relatively short amount of time 共⬃50 ns兲. This reflected pulse is then redirected through the four-port circulator toward the atmospheric test section 关Fig. 1共b兲兴. Discharging the plasma switch produces the oscillating voltage and current waveforms shown in Fig. 6. A measured peak current of roughly 450 A flows through the plasma switch when a discharge is formed. This indicates that a conductive column of plasma is being produced between the electrodes. The LC inversion generator is designed to output a peak voltage close to twice the initial charging voltage of 30 kV. However, it can be seen from the waveforms in Fig. 6 that the output voltage of the circuit in Fig. 3 collapses before it can reach the peak output amplitude. This occurs because the electric field strength between the electrodes reaches the breakdown threshold for 1 bar of argon before the peak voltage output is produced from the inversion generator. The plasma is formed when electric field accelerated electrons ionize gas molecules, producing more charge carriers, resulting in an increased conductivity of the plasma. Eventually, the highly conductive plasma causes the switch impedance to collapse and a high current to flow through the gap. The voltage fall time is primarily limited by the few hundred nanohenry parasitic inductance of the switch assembly including high voltage leads. To compare the plasma geometry between air and argon, an intensified charge-coupleed device 共ICCD兲 camera was used to take images of the plasma development inside the gas isolated tube at the moment the plasma between the elec-

Downloaded 01 Mar 2013 to 129.118.19.71. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

114701-4

Foster et al.

Rev. Sci. Instrum. 79, 114701 共2008兲

FIG. 7. Images of the plasma switch firing in 1 bar of argon and air; 50 ns exposure time.

trodes reached the critical density required to reflect the 2.85 GHz HPM pulse. The ICCD camera used a gate 共exposure兲 time of 50 ns 关see Fig. 5共a兲兴 to capture the images of the plasma. The exposure shown in Fig. 7 is a gray scale image of the volume breakdown occurring inside the argon and air filled gas isolated tube, respectively. Specifically, the image of the plasma forming in argon on the left side of Fig. 7 is that of the breakdown occurring in the waveforms shown Figs. 5共a兲 and 5共b兲. Possibly due to the lower ionization potential of air 关O2: 13.6 eV; Ar: 15.8 eV 共Ref. 5兲兴, unintentional breakdown in the extremely nonuniform field solely due to microwave excitation occurred more readily in air 共inversion generator turned off兲. This unintentional breakdown would throw off the desired timing and cause HPM reflection to occur before it was desired. Hence, argon was chosen over air as switching medium. It should also be noted that the more diffused discharge in argon produced typically a better rise time of the reflected microwave signal.

FIG. 8. 共Color online兲 Scattering parameters 共%兲 vs plasma conductivity: reflected 共S11兲, light gray; transmitted 共S21兲, gray; absorbed, black.

switch structure, as measured from the waveguide ports, is the relevant figure of merit for this measurement and can be calculated via simulation in Ansoft HFSS™. For 2.85 GHz excitation, the calculated S11 is −8.51 dB, meaning about 14% of the incoming power is reflected back out the incident port. Likewise, the S21 parameter is −0.669 dB, which means that the remaining 86% of incoming power passes through the structure. The field enhancement from the electrodes protruding into the waveguide cross section is clearly seen in Figs. 9 and 10 for microwave and HV excitation, respectively. The field magnitude is quantified, see Fig. 9, for

IV. COMPUTER MODELING AND SIMULATION

The operation of the plasma switch was simulated using the high frequency structure simulation Ansoft software HFSS™. Simulations of the structure were conducted using various plasma geometries over a range of varying plasma conductivities. Scattering parameters were observed in the HFSS™ simulations for a simple WR-284 waveguide geometry with electrodes oriented in the center of the major electric field component of the microwave. Waveport excitations were used at the incident and the transmitted ports of the structure and a solution frequency of 2.85 GHz was used for the analysis setup. Power reflection, transmission, and absorption were observed as a function of plasma conductivity for a cylindrical plasma geometry equal in size to the gas isolated portion of the plasma switch. The HFSS™ simulation results shown in Fig. 8 are the scattering parameters of the simulated waveguide as a function of the plasma conductivity. Given that experimentally the plasma switch reflects approximately 70% of the incident while absorbing close to 30% 共virtually no power is transmitted after the plasma switch is fired兲, a matching conductivity of about 264 S/m is identified from Fig. 8. With this particular simulated geometry the overall impedance of the plasma is approximately equal to 0.5 ⍀. In order to properly function as a switch element, the waveguide spark gap must exhibit a low passive microwave reflection when inactive. The S11 scattering parameter of the

FIG. 9. 共Color online兲 Electric field magnitude in plasma switch due to 2.85 GHz waveport excitation.

FIG. 10. 共Color online兲 Electric field magnitude in plasma switch due to high voltage excitation; −37.5 kV applied voltage.

Downloaded 01 Mar 2013 to 129.118.19.71. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

114701-5

Rev. Sci. Instrum. 79, 114701 共2008兲

Microwave switching using a spark gap

FIG. 12. 共Color online兲 Delay time comparison of fast rise time experimental data 共gray兲 with slow rise time experimental data 共black兲.

FIG. 11. Electric field magnitudes along plasma switch structure: black line, centerline of waveguide; gray line, 0.25 cm from electrode tip.

4 MW microwave power level with microwave phase set identical to Fig. 11. The solid black line represents the field magnitude along the centerline of the waveguide. The location corresponding to the center of the spark gap exhibits an approximately 50% higher electric field compared to the unmodified waveguide. This field enhancement increases with closer proximity to the electrode tip, as shown by the gray line, in which the field levels 0.25 cm from the tip are plotted. At this distance, the maximum electric field is 63 kV/cm, constituting an enhancement of 370% over the 17 kV/cm field present in the unmodified portions of the waveguide. The electric field resulting from the unipolar high voltage pulse supplied by the inversion generator is depicted in Fig. 10. The highly asymmetric field profile is due to the grounding of the waveguide structure. The large overvoltage created by the field enhancement of the electrodes and aided by the pulsed field applied from inversion generator ensures rapid breakdown and ionization of the spark gap, resulting in a fast 共⬍100 ns兲 transition from a low to a high S11 device. V. APPLICATION OF MICROWAVE SWITCH

Having successfully reduced the power rise time from 600 to 50 ns with the presented switch enabled dielectric window flashover testing in the atmospheric test section 关see Fig. 1共b兲兴 with results more easily comparable to numerical simulations of microwave induced surface flashover. In brief, surface flashover is induced when the fast rising HPM pulse is applied to the dielectric window. Delay times between pulse application 共marked by the 50 ns rise兲 and window flashover 共marked by power reflections from the dielectric test window兲 can then be recorded from the observed waveforms. Figure 12 shows a comparison with previous research4 that utilized a slow rising HPM pulse.

It is obvious from the data in Fig. 12 that the average delay time for breakdown is significantly reduced when the sharply rising HPM pulse is applied to the window. This can be attributed to the several hundred nanosecond reduction in the incident microwave pulse rise time. It should finally be noted that the impact of the small prepulse amplitude of 0.25 MW versus the 2.5 MW in the main pulse, see Fig. 5共b兲, is considered insignificant for the HPM flashover experiments, primarily due to the threshold nature of the electrical breakdown process. Further, the charge carriers, i.e., electrons, are not swept out of the region during the prepulse under gigahertz microwave excitation as they would be under unipolar excitation. Rather, the electrons remain rather stationary under the added effect of elastic collisions in the pressure regime above 60 torr 共0.08 bar兲.7 VI. DISCUSSION AND CONCLUSIONS

We have presented an experimental setup that enabled reducing the rise time of a 3.5 ␮s HPM pulse from 600 to 50 ns at megawatt power levels. It was shown that a custom designed waveguide spark gap can be used as a fast switching mechanism for sharply reducing the rise time of a guided HPM. As a switching medium, argon was found to be superior to air. The faster pulse with a more ideal step excitation will be utilized for generating experimental HPM window flashover data more conducive to comparing with theoretical models. 1

Z. J. Yao, S. Chen, S. Eshelman, D. Denniston, and C. Goldsmith J. Microelectromech. Syst. 8, 129 共1999兲. 2 K. E. Mortenson, J. Borrego, P. E. Bakeman, Jr., and R. J. Gutmann, IEEE J. Solid-State Circuits 4, 413 共1969兲. 3 H. Farber, IEEE Trans. Microwave Theory Tech. 13, 127 共1965兲. 4 G. Edmiston, A. Neuber, J. Krile, L. McQuage, H. Krompholz, and J. Dickens, IEEE Trans. Dielectr. Electr. Insul. 14, 783 共2007兲. 5 A. A. Fridman and L. A. Kennedy, Plasma Physics and Engineering 共Taylor & Francis, New York, 2004兲. 6 J. E. Burke, T. G. Engel, and M. Kristiansen, Conference Record of the 18th IEEE Power Modulator Symposium, 1988 共unpublished兲, pp. 228–240. 7 J. T. Krile, A. A. Neuber, H. G. Krompholz, and T. L. Gibson, Appl. Phys. Lett. 89, 201501 共2006兲.

Downloaded 01 Mar 2013 to 129.118.19.71. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions