R. S. Aadland*, W. A. Hoskins**, C. E. Vaughant Rocket Research Company Redmond, WA R. J. Kayt Pacific Electro Dynamics

Transcription

1 IEPC ACHIEVING RELIABLE, REPEATABLE STARTS OF A 26 KW ARCJET R. S. Aadland*, W. A. Hoskins**, C. E. Vaughant Rocket Research Company Redmond, WA 9873 R. J. Kayt Pacific Electro Dynamics Redmond, WA 9873 Abstract Background The Arcjet Advanced Technology Transition One of the primary challenges throughout Demonstration (ATTD) program has completed development of the 3 kw class ammonia arcjet design, fabrication and test of a high power Power PCU has been start up. Virtually every group that Conditioning Unit (PCU) to supply regulated has worked with high power (>2kW) arcjets has power to a 26 kw arcjet( 1 ). Rocket Research experienced difficulty achieving reliable and Company (RRC) and Pacific Electro Dynamics repeatable starts. Large electrode radii combined (PED) under the sponsorship of TRW and the Air with large electrode gaps result in high breakdown Force Phillips Laboratory (AFPL) have conducted voltages. Laboratory tests utilizing industrial four test sequences to determine the best method to plasma cutting or welding type power supplies achieve reliable, repeatable starts of a 26 kw arcjet. combined with a DC capacitive start circuit Several organizations have attempted high power typically result in severe electrode damage due to starts in the past with mixed results. severe current overshoot during transition to steady state( 2 ). Breakdown voltages between 1 to 2 kvdc The fundamental problem in the case of the ATTD were typically measured. However, implementation PCU is that the buck regulator topology of the main into a flight design would be impractical since the power output circuit inherently has low output high voltage, high current blocking diode required in inductance. Early trade studies indicated that a the output would generate unacceptable thermal shorting switch approach to starting offered the losses resulting in low power conversion efficiency. lowest complexity, the greatest heritage, and a Since the objective of the Arcjet ATTD program was reasonable probability of success. However, shorting to perform a space flight demonstration of a 26 kw circuits delivering up to 25 volts for up to 8 arcjet an enabling subsidiary objective was to nsec. were consistently unable to achieve arcjet produce a start circuit that would be reliable, starts. repeatable, and implemented and packaged in a form appropriate for a flight experiment. This paper discusses the design solution that led to reliable starts for high power arcjets. This includes Early designs of high power PCUs, developed by design and fabrication of a pulser circuit to replace Space Power Industries (SPI), implemented a buck the output shorting circuit, and design and regulator power conversion approach driven by the fabrication of two-stage saturable blocking high input voltages inherent in nuclear power inductors. Studies of dv/dt, duration of the pulse, sources (i.e. SP-1)( 3 ). Two start circuit designs amplitude of the pulse, cathode surface condition, were originally considered; a "shorting" switch and cathode geometries all contributed to an circuit and an auxiliary winding circuit. Both understanding of the start phenomenon. The result is circuits, shown in simplified schematics as Figures 1 a robust circuit design capable of achieving the goal and 2, utilize a fly-back inductor to generate a of successfully starting the arcjet during the narrow width high voltage pulse. A disadvantage upcoming Electric Propulsion Space Experiment inherent in both approaches is relatively high (ESEX) on the Air Force P91-1 Atmospheric switching currents caused by the low output Research and Global Observation Satellite inductance of the high power PCU. The shorting (ARGOS). switch approach was favored due to its simplicity over the auxiliary winding approach( 4 ). Several * RRC Sr. Development Engineer working groups had opportunities to test one of the ** RRC Sr. Development Engineer t RRC Project Engineer three SPI PCUs that were built under Air Force PED Sr. Staff Engineer

2 461 IEPC sponsorship and to gain experience with the shorting to achieve reliable starts. The original PCU switch start circuit. Limited successes or outright specification increased the start pulse voltage failure to achieve reliable breakdown was uniformly requirement to 2 kv in an attempt to provide reported. Early testing at JPL encountered margin. After early test failures the start voltage difficulty using the shorting switch start circuit. goal was increased to 2.5 kv. In retrospect it is clear Argon glow discharges transitioning to low flow that major additional increases were needed. NH 3 starts were used as a work around to the problem. Since reliable breakdown could never be Circuit Development achieved with the shorting switch start circuit, JPL developed a start circuit that could be used with a The chronological history of start circuit Linde plasma cutting power supply( 5 ). Start development for the ATTD 26 kw PCU can be difficulties diminished, but continued throughout divided into 4 primary headings: this very successful endurance test. NASA Lewis. sho 1. " st The SPI breadboard c t. "shorting" start circuit. Research Center attempted to use one of the three ( b. (The baseline design at start of program.) units for 2 kw H 2 arcjet testing but only achieved one successful start and resorted to argon-to-h 2 2. The Engineering Model "shorting" start circuit transitions as a work around( 6 ). Texas Tech. with modifications. (The baseline design with University developed a high frequency pulse circuit improvements, fabricated on a flight type from a modified TIG welder for starting their N 2 circuit card assembly and integrated within the high power arcjet with a switch over to the high PCU chassis.) power DC power supply once stable arc was power DC power supply once a stable arc was 3. The "pulser" start circuit. (The new start obtained( 7 ). The Air Force Phillips Laboratory and. circuit Rocket approach) Research both have used DC capacitor discharge circuits as a work around for laboratory 4. The "flight design" start circuit. (The new start testing. Using the shorting switch start circuit, circuit approach adapted to a flight design Rocket Research obtained starts only after circuit card assembly and integrated within the preconditioning the electrodes by sustained PCU chassis. operation at 2 kw with a laboratory low power PCU. None of these approaches are applicable to The following sections provide a description, theory flight systems. The early work on flight type of operation and test results for each of the above inductor shorting circuits was not encouraging. headings. With this background, the Arcjet ATTD program had to solve the problem of high power arcjet starting in a way that could be implemented in SPI Breadboard "Shorting" Start Circuit flight hardware. The SPI breadboard start circuit, shown in a simplified schematic form in Fig. 2, utilizes a direct A new approach described herein utilizes a fly-back "shorting switch" approach to generate the high inductor combined with a magnetic switch (blocking voltage start pulse. The "shorting" switch is closed inductor) in place of a blocking diode. In addition, for approximately 5 is which initiates a current the magnetic switch is incorporated into the PCU flow through the three parallel output inductors. output inductor. As a result, very little degradation The switch is then opened thereby releasing the of the PCU's power conversion efficiency occurs energy stored in the inductors causing the inductor from the blocking inductor, voltage to "fly-back" creating a high voltage narrow width sinusoidal pulse across the open circuit of the Objective of Effort arcjet electrodes resulting in arc breakdown. Since the main power components are all active due to the For the Arcjet ATTD program a reliable starting temporary short circuit condition, the power circuit method was a necessity for the flight experiment, immediately supplies the required current to sustain The starting point was the SPI shorting switch a steady state arc discharge. The breadboard start inductor fly-back circuit. At the outset, it was circuit used a 3X3 matrix of shorting transistors in a believed that peak voltages generated from the series/parallel arrangement yielding a nominal peak inductive fly-back circuit of 1 to 2 kv would be output voltage of 2.5 kv. adequate for arc breakdown. It was further believed that combining the switches in a series/parallel During a test series at RRC, the SPI breadboard start arrangement would improve the circuit adequately circuit was tested with a modified D-lE version of a 2

3 IEPC kw class arcjet. The constrictor and nozzle The primary concern regarding start reliability dimensions of this arcjet were identical to those of following the test series above was the breakdown the 26 kw ATTD arcjet. Initially, the start circuit event. The results indicated that a higher peak was unsuccessful in achieving breakdown at a peak voltage would be required to increase the success voltage level of 2.5 kv throughout an NH 3 mass rate in achieving breakdown. Therefore the shorting flow rate range of 25 to 25 mg/s. The primary test circuit was modified during implementation into the results are summarized below: ATTD engineering model PCU provided by PED. However, based on the test related issues described * Breakdown was obtained only after a above, it was still generally believed that a 2.5 to 3. substantial decrease in electrode gap was kv peak voltage would be adequate for ensuring combined with a reduced mass flow rate and reliable breakdowns. the use of a higher voltage start circuit from a laboratory low power PCU. Engineering Model (EM) "Shorting" Start Circuit * The difficulty in obtaining a breakdown with & Modifications the breadboard start e st b d circuit c was w initially i Based on the test results described above, the attributed to an oxide layer build up on the shorting circuit was enhanced to produce the EM arcjet arcjt electrodes. ectre. Breakdowns readons with the bread- shorting start circuit. Fig. 5 shows a photograph of board start circuit were achieved only after the ATTD engineering model PCU indicating the the ATTD engineering model PCU indicating steady-state operation of the the electrodes at location of the shorting start circuit. The EM location of the shorting start circuit. The EM several kw. shorting start circuit utilized a 4 X 2 matrix of * The breadboard start circuit exhibited sec- transistors in a parallel/series arrangement yielding ondary stray start pulses, attributed to bounce a nominal peak output voltage of 2.7 kv. During in the command switch, following the initial initial tests at the RRC facility, the start circuit was high voltage pulse. On numerous occasions the unsuccessful in achieving breakdown of the D-IE secondary start pulse caused a large enough per- arcjet. As a result of the failed breakdown attempts, turbation which resulted in a failed transition a series of design modifications to the start circuit to steady state as shown in the oscilloscope voltage and current plot of Fig. 3. was initiated with the objective of increasing the voltage-time profile of the pulse. The results of this investigation are summarized in Table 1 below. An oscilloscope voltage and current plot showing a successful start transition of the SPI breadboard A graphical representation of the waveform shaping PCU is shown in Fig. 4. The "start" current shown is results obtained from the capacitive and inductive actually the "shorting" current which was obtained circuit modifications is shown in Fig. 6. In addition simultaneously by routing the associated conductor to the circuit modifications listed in Table 1, through the current probe along with the main methods such as decreasing mass flow rate, output conductor. repetitive pulsing at 3 Hz and "off-pulsing" the Table 1. Test Results of Design Modifications to the Baseline Start Circuit. Modification Objective Results Higher voltage zener diodes in Increase peak voltage. Shorting switches failed. shorting switch. Output capacitive waveform Increase pulse width. Broadened pulse width but shaping. failed to achieve breakdown. Increase start inductor (shorting) Increase peak voltage by adding Increased peak voltage but current. stored energy. failed to achieve breakdown. Increase output inductance Increase peak voltage and pulse Increased peak voltage but Configuration # 1 width by adding stored energy in failed to achieve breakdown. (3) 8 gh step inductors inductance. Configuration # 2 (1) 5 lh inductor plus (2) 3 mh shorting inductors Add FETs to modified inductor Increase peak voltage. Occasional breakdowns at circuits above (1 X 2 matrix) 4.1 kv peak. 3

4 463 IEPC propellant valve were attempted. Although these waveform necessary to achieve reliable arc methods generated occasional breakdowns, the need breakdowns. The waveform of the high voltage pulse for a significantly higher voltage pulse was could be varied by changes in both the charge recognized as the solution for obtaining reliable and duration and output snubber capacitance. The results repeatable breakdowns, showed that arc breakdowns were achieved between 4.5 and 7. kv depending upon the rise time or dv/dt "Pulser" Start Circuit of the pulse. The breakdown voltage increased as the dv/dt increased. Waveforms generating peak From the above results it was clear that a departure voltages below 5.2 kv were less than 1% reliable. from the shorting start circuit offered the best This data suggested that a minimum peak voltage of chance of achieving reliable breakdown. A "pulser" 5.2 kv was required from the start circuit to achieve circuit approach was suggested by NASA LeRC( 8 ). reliable arc breakdowns. A comparison between the The "pulser box" was initially built as a stand alone oscilloscope voltage waveforms of the breadboard unit designed to be used as a development tool in and pulser start circuit is shown in Fig. 8. determining the required pulse waveform needed for achieving reliable breakdowns of the 3 kw class The "pulser box" was then connected in parallel ammonia arcjet. The circuit, shown in a simplified across the output of the PCU and a set of three 9 mh schematic form in Fig. 7, utilized a fly-back "blocking" inductors were connected in series with inductor, energy storage capacitor, timer/driver the existing 5 H output inductors. The individual circuit, high voltage switch (FETs), snubber blocking inductors were located outside of the PCU capacitor and steering diodes. The timer/driver chassis due to their large physical size. The blocking circuit controls the switching of the high voltage inductors essentially act as a magnetic switch which FET array. Closing the switch allows current to is "open" during the high dv/dt of the fly-back increase, over a nominal period of 15 Its, to a level inductor thereby allowing a high voltage output of 2 amps in the primary winding of the fly-back pulse. If breakdown does not occur prior to the inductor. Opening the switch transfers energy from saturation limit (voltage-time integral) of the the primary to the secondary creating an initial blocking inductors, the magnetic switch will current of roughly 16.7 amps (2 amps/12:1 ratio) prematurely "close" and the output voltage will in the secondary winding. The voltage generated collapse resulting in a failed start. However, under across the primary winding is limited to 5 volts normal conditions when breakdown occurs prior to due to the clamping semiconductor diodes within the the saturation limit, current from the secondary high voltage switch (FET). The fly-back voltage winding of the fly-back inductor temporarily impressed across the secondary winding is therefore sustains the arc. During this period the magnetic 6 volts (5 volts X 12:1 ratio). However, peak switch remains open since the voltage across the voltages in excess of 6 kv were measured due to electrodes (5 to 1 Vdc) fall below that of the leakage inductance of the circuit. The snubber PCU input voltage (15 to 225 Vdc). A necessary capacitor across the output controlled the rise rate saturation of the blocking inductor occurs 15 its of the high voltage pulse between.25 and 2 kv/is. following breakdown due to this 5 to 175 volt The steering diodes located on the output of the differential. The magnetic switch then "closes" and circuit ensure that the polarity of the output pulse primary current from the buck regulator power remains negative. Without these diodes, sinusoidal converter sustains the arc discharge. It should also be ringing of the output following a failed breakdown noted that higher levels of sustaining current create event at negative high voltage may cause a reverse lower arc voltages due to the negative arc impedance polarity breakdown between the arcjet electrodes and thereby result in a higher voltage differential due to a high positive voltage of the cathode, across the blocking inductors. The increased differential accelerates saturation of the inductor Testing of the pulser box was conducted both at the thereby decreasing the time until primary converter Rocket Research Company facility and at the Air current can sustain the arc. Force Phillips Labs facilities on the D-IE arcjet. Implementing this new approach described above "Pulser Box" Testing at RRC resulted in the first successful breakdown and transition to steady state operation of the arcjet During initial testing at RRC, the pulser circuit was with the pulser box circuit and PCU. Figure 9 shows connected solely to the arcjet independent from the a representative oscilloscope voltage and current PCU to determine the required high voltage pulse plot of the transition from start to steady state 4

5 IEPC operation. It was recognized, however, that in spite duced since lower rise rates also generally result in of this successful achievement, a more thorough higher voltage-time integrals. While it is desirable investigation into the breakdown characteristics of to minimize breakdown voltage in efforts to the 3 kw class ammonia arcjet with a narrow width minimize complicated high voltage design high voltage pulse was needed to ensure adequate requirements, the voltage-time integral must stay margin in defining the flight start circuit design. The below the inductor saturation limit for breakdown investigation included a series of breakdown tests to occur. Therefore, the rise rate must be optimized which were conducted with the pulser box and arcjet to allow the lowest breakdown voltage without at the AFPL facility, requiring a voltage-time integral larger than the inductor saturation value. "Pulser Box" Testing at AFPL Details of the arcjet, pulser test set up and test "Flight "Fliht Design" Start Circuit procedures are described by Tilley et al.( 9 ) In the Based on the results of the pulser circuit testing, a first series of tests, the effects of cathode tip breadboard version of the complete flight design geometry, cathode gap, and propellant flow rate on start circuit was fabricated and installed in the the breakdown voltage were investigated and it was ATTD engineering model PCU for testing with an demonstrated that breakdown voltage was not engineering model arcjet. Figure 11 shows a strongly affected by cathode tip geometry or cathode simplified schematic of the flight design start gap. However, breakdown voltage could be circuit integrated to the PCU main power circuit. significantly reduced by reducing the propellant Due to physical size limitations, the.7 Farad flow rate. energy storage capacitor used in the pulser box was lowered to.14 Farad in the flight design circuit. In a second series of tests, the effect of voltage rise However, the decrease in capacitance was partially rate on breakdown voltage was investigated. All compensated for by increasing the charge voltage measurements were made with a cathode geometry, from 4 to 15 Vdc. cathode gap and propellant flow rate identical to those to be used in flight. Measurements were made During initial test series at RRC, the PCU was at four different voltage rise rates obtained by mounted outside the vacuum chamber and connected adding combinations of capacitance to the pulser to the engineering model arcjet with large welding output. cables. This configuration permitted easy access to the PCU and allowed thrust measurements to be The results of this test series are summarized on made of the arcjet during the hot firings. Prior to any Fig. 1. Because the voltage ramp up is nearly linear, hot firings, breakdown voltage was measured over a the voltage-time integral can be approximated by: matrix of flow rates and voltage rise rates. Variations in output snubber capacitance allowed S_ Vt =Vdt =f = Vb 2V testing at varying rise rates. A temporary set of development composite blocking inductors with a where, saturation limit of 6 kv-gs were used. The results where, are plotted on Fig. 12 and are consistent with Sv-t = Voltage-time integral (kv-ts) measurements of the effects of propellant flow rate V b and voltage rise rate on breakdown voltage as = Breakdown Voltage (kv) measured earlier with the stand-alone pulser. V = Rise Rate (kv/.ls) As indicated by Fig. 12 and earlier results by Tilley The individual data points for each test are plotted in et. al.( 9 ), significant reductions in required addition to the averages to convey the variability in breakdown voltage were achieved at reduced flow the breakdown voltages required for a fixed set of rates. This effect allows the possibility of further operating conditions. This data demonstrates the ensuring start reliability by decreasing flow rate potential use of multiple pulses to ensure that during the start-up sequence. During start-up, the breakdown occurs in situations where peak output ATTD arcjet will operate at approximately 1 kw voltage is constrained. for several seconds before power is ramped up to the steady-state level of 26 kw. Therefore, it would be As the rise rate decreases, the breakdown voltage possible to reduce the flow as low as 1 mg/s generally decreases. However, a tradeoff is intro- during start, and still maintain a power-to-flow-rate 5

6 465 IEPC ratio as that of steady-state operation. The flow rate The pulser circuit concept, suggested by NASAcould then be ramped up with the power. LeRC, was developed into a flight circuit board and coupled with a magnetic switch. Development Following the breadboard model tests, the start testing of the pulser circuit verified the feasibility circuit was incorporated onto a printed wiring board of the approach through consistent breakdowns and and flight configuration blocking inductors were start transition at full mass flow rate. Integration installed. Although the capacitive output snubber of a "flight design" pulser circuit into the PCU was eliminated, the maximum voltage rise rate was chassis decreased start reliability due to reduced measured at 5 kv/ls due to parasitic capacitance in component capabilities. However, test results have the compact flight design. The 9 mh blocking demonstrated substantial increase in start reliability inductors were combined with the 5 gih steady- of the pulser circuit is possible when combined with state output inductors by physically integrating methods such as multiple start pulses and/or reduced both inductors into a single composite inductor. The mass flow rate starts. There have been no failed start saturation limit of this inductor in terms of a attempts with the PCU connected to the arcjet with representative voltage-time profile is shown in the flight configuration power cable. With reliable, Fig. 13. Again due to physical size limitations repeatable starts of the arcjet now being achieved, within the PCU chassis, the voltage-time integral of fabrication of the flight start circuit is in progress. each individual composite inductor was limited to 3 kv-ts for the flight design circuit. The fly-back inductor utilizes a multiple core, tape wound, low References profile configuration. 1. Vaughan, C. E., Aadland, R. S., Cassady, R. J. and The added capacitance of the long welding cables Kay, R.J., "Integrated Mission Simulation of a 26 connecting the PCU to the arcjet, in addition to the kw Flight Arcjet Propulsion System", AIAAlimited rise rate and blocking inductor saturation , June, value, limited the available peak voltage to 5.5 kv. 2. Cassady, R. J., Lichon, P. G., King, D. Q., "Arcjet As a result, less than 1% start reliability was Endurance Test Program, Final Report", AL-TRobtained at the full mass flow rate of 24 mg/s 9-69, March, during this phase of testing. 3. Wong, S. P., "Testing of 3 kw arcjet PCU with However, in a second phase of testing, the PCU was Arcjet Thrusters", IEPC-91-7, October, installed inside the vacuum chamber with the flight 4. Wong, S. P., "Testing of the 3 kw Arcjet PCU configuration cable connecting it to the arcjet. There Starter using the Shorting Switch Approach", were no failed starts during the 15 subsequent IEPC-91-3, October, firings at the 24 mg/s flow rate. It was concluded that the reduced capacitance of the flight cable 5. Polk, J. E., Goodfellow, K. D., Pless, L. C., sufficiently increased the peak voltage of the pulse resulting in reliable breakdowns. "Ammonia Arcjet Engine Behavior in a Cyclic Endurance Test at 1 kw", IAF , September, Summary and Conclusions 6. Hamley, J., NASA-LeRC, Personal Communi- Upon initiation the Arcjet ATTD program was not cation, planning to develop a start circuit. The original 7. O'Hair, E., TTU, Personal Communication, intent was to adapt existing circuit concepts into H amel flight designs. The arc breakdown phenomenon for y,. NASA-LeRC, Personal Communi- 3 kw class ammonia arcjets had been poorly characterized and as more data was produced, it was 9. Tilley, D. L., McFall, K.A., Castillo, S., Andrews, clear the concepts developed to date could not J. C. and Bromaghim, D. R., "An investigation of produce the start energy required. A departure from the Breakdown Voltage Characteristics of a 3 the baseline shorting circuit was required to deliver kw Class Ammonia Arcjet", AIAA , adequate start energy to the arcjet. June,

7 IEPC INPUT POWER OURCE S1SH S2 ARCJET <-/ o l b-- i- >- S3 7--tC< J? SSTART Fig. 1. Simplified Schematic of the Auxiliary Winding Start Circuit _- S1 "SHORTING' + SWITCH NPUT POWER C< 5 H SOURCE I c S a S 5 gh "- ARCJET S3 /5 gh Fig. 2. Simplified Schematic of the "Shorting" Switch Start Circuit 7

8 467 IEPC j o- "Start" Current Arc Current PCU Output Voltage. -5 S, Arc Voltage a -1- Initial Pulse -15. Secondary Pulse Time (Micro-Seconds) Fig. 3. Failed Start Transition of the Breadboard SPI PCU Due to Perturbation of Secondary Start Pulse. 2 : "Start" Current 15 E - 1 Overshoot I 5- o S 1 Ca - * I > -5: Initial Start Pulse Arc Current I -5 1 ^ h, 1 / A Arc Voltage Secondary Start Pulse Time (Micro-Seconds) Fig. 4. Successful Start Transition of the Breadboard SPI PCU During a Secondary Pulse. 8

9 IEPC OUTPUT POWER CONNECTORS OUTPUT INDUCTORS START CIRCUIT CARD ASSEMBLY INPUT POWER CONNECTORS Fig. 5. Photograph of the ATTD Engineering Model PCU 9

10 469 IEPC pf ADDED OUTPUT CAPACITANCE CD, 2 B A S E U N E 3 1x2 FET MATRIX 1x 2 FET MATRIX _" \WITH THREE 8 ph STEP INDUCTORS 4 1 x 2 FET MATRIX 1 ACHIEVED SEVERAL SUCCESSFUL WITH (2) 3 mh BREAKDOWNS AT FULL MASS FLOW SHORTING INDUCTORS RATE (-4.1 kv) 5 I I I I I I I I TIME (ps) Fig. 6. Representative Waveforms of the Shorting Start Circuit Modifications STEERING + DIODES.7 FARAD o '._ SOURCE STORAGE ) CAPACITOR SNUBBER CAPACITOR OUTPUT Q1 N2/N1= 12/1 COMMAND TIMER/ START DRIVER 5V 1= MULTIPLE, HIGH CURRENT HIGH VOLTAGE FET TRANSISTOR Fig. 7. Simplified Schematic of the "Pulser Box" Circuit 1

11 IEPC S -1. > o SHORTING' START CIRCUIT S -4. WAVEFORM FAILED TO SBREAKDOWN ARCJET TYPICAL PULSER CIRCUIT -7. BREAKDOWNS Fig. 8. TIME (ps) Comparison Between Pulser and Shorting Start Circut Waveforms OV 1 V/DIV ARC VOLTAGE Vi, = 225 Vdc LOW MODE 24 mg/sec 24 mg/sec LO MO DE FOR -1 ms OPEN CIRCUIT - /,RSHTA VOLTAGE OF 225 Vdc AMPS 5 AMPS/DIV +1 AMP PCU O PCU OUTPUT ^rrun CURRENT SCURRENT PULSER CURRENT BREAKDOWN EVENT 5 ps/division Fig. 9. Pulser Current to PCU Current Transition to Steady State 11

12 471 IEPC _19 9- kv/us R 8- S 7 - Rise Rate = 1.6 kv/us 6 /.86 kv/us > kv/us m Voltage-time Integral (kv-us) r Individual Data * Averages Fig. 1. Breakdown Voltage vs Vt Integral Pulser Circuit Tests - 24 mg/s ENERGY STORAGE CAPACITOR CAPACITOR 5 VOLT FET STEERING DIODES + VOLTAGE REGULATOR * INPUT 15 s S r i POWER - VDC SOURCE S vdc 9 mh 5 gh FLYBACK INDUCTOR 2 12:1 ARCJET /S 9 mh 5 gh 12:1 S3 2p /9 mh 5 gh Fig. 11. COMPOSITE BLOCKING INDUCTORS Simplified Schematic of the Flight Design Start Circuit 12

13 IEPC ) 7- P 6 = Rise Rate = 7.1 kv/us 5. kv/us kv/us s Fig. 12. Breakdown Voltage vs Flow Rate Breadboard Model Flight Design Circuit OV U -2V O -4V O -6V z VOLTAGE-TIME SATURATION LIMIT TIME (Is) Fig. 13. Saturation Limit of the Composite Inductor in the Flight Design Start Circuit 13