eoretical and Experimental Study of the Electromagnetic Environment Surrounding a Magnetic Induction Launcher

~ ~ ~ ~ ~ a c ~ E ~ e c t r o m a g n e t i c (EM() f eld measurements were conducted near a prototype system that launches metal plates via EM induction. These plates are intended to augment a vehicle's passive armor by intercepting incoming kinetic energy (KF,) projectiles some distance away from the vehicle. The subscale EM induction launcher consists of two 4.5-turn, 15 cm square spiral coils machined from 1.27 cm thick copper-beryllium plate. This type of system is designed to launch a 15 cm square aluminum plate in an edge-on orientation. As was done for an earlier design [W. Coburn, C. Le, and H. Martin, "Electromagnetic Field Measurements Near a Single Stage Recdnnection Gun," U.S. Army Research Laboratory Report ARL-MR-206, April 19953, both the short-term magnetic fields associated with the launch process and the long-term electric fields associated with the flying plate were measured. A theoretical model designed to simulate the magnetic fields surrounding the launcher has been developed and its results compare favorably with measured data.


I. INTRODUCTION
This study is part of a continuing effort to remain abreast of the nature of the EM environments surrounding emerging weapons systems so that their effects on the host vehicle, electronics, and personnel can be predicted while the system is in the design phase.It is similar in scope to a study involving an earlier prototype EM launch system [l], in which magnetic and electric fields associated with the launcher and plate were measured.The system under study here is a type of single-stage coil gun designed to launch metal plates at speeds in excess of 200 d s , with the intent of intercepting and breaking up incoming KE projectiles before they strike the host vehicle.In conjunction with systems designed to defeat other threat types, this will allow for less massive passive armor to be used on future vehicles.The current prototype differs from those studied previously [l-31 in that it will fire a plate of twice the mass at the same velocity (i.e., it has twice the efficiency) and uses different materials in the Manuscript received April 8, 1996.P.R.Beming phone 410-278-4648, email beming@arl.milcoils and support structure.The larger fields that accompany this increase in performance and the possibility of shielding effects in the support structure were part of the motivation behind this study.In addition, it allowed for a test of a computer model, originally developed to predict launcher performance, that calculates magnetic fields generated by the launcher and portions of its power supply.

A. EMInduction Launcher
The launch mechanism is based on the principle of magnetic induction: when a changing magnetic field is applied to a conducting object, such as a metal plate, electric currents are "induced" within the object.The interaction of the field and these currents can then result in a Lorentz force being applied to the object.Here the magnetic field is supplied by two 15.2 cm square spiral coils; one parallel to either face of the plate.The coils are made by taking a 15.2 cm square by 1.27 cm thick plate of a copper-beryllium alloy and milling a 0.63 cm wide slot in a square spiral pattern, leaving behind a 0.63 cm wide, square spiral coil.Only 4.5 of the 5 possible turns were machined, as a previous launcher based on round spiral coils [l] failed near their centers, and it was felt that leaving the central section of the plate intact would strengthen this area.External electrical connections are made by 1.27 cm diameter copper-beryllium rods connected to the centers of each coil, and the two coils are connected internally by a copper-beryllium busbar at a rear corner.
The coils are individually supported by two 30.48cm square, 2.54 cm thick plates of an epoxy-fiberglass composite (GlO), in which a 1.27 cm deep square spiral pattern has been milled to accept a coil.A cavity for the plate is formed with sections of 1.27 cm thick G10 plate that also separate the two coil support plates, so that the overall coil-to-coil separation is 3.81 cm.The top and bottom of the stack are capped with 5.00 cm thick 610 plates.Eight lengths of 1.90 cm thick high-strength threaded steel rod hold the stack together.In addition, two 38.1 cm long, 15.24 cm wide sections of steel U-channel are attached to the top and bottom of the stack, and an additional four 1.90 cm rods connect the overlapping parts of these alongside the stack.

B. Test Stand
The launcher is powered by connecting it to a high voltage capacitor bank, through a high voltage switch.A diagram of the test stand used in this study can be found in Fig. 1.It shows a two-capacitor bank, which has a nominal total capacitance of 1600 p. F and a maximum voltage of 10 kV.The high-voltage terminals of the capacitors are connected to an ignitron switch through an aluminum plate.A feed cable connects the switch to the top terminal of the launcher.The launcher rests on an aluminum plate that also forms its "ground" connection, back to the capacitor cases.Another aluminum ground plane was added to the front face of the stand to shield the sensors from the electronics associated with the ignitron switch.A Pearson coil and a Rogowski coil were used to measure the total current and the time rate of change of the current, respectively, during all shots.
Also shown in Fig. 1 is the coordinate system used in determining the positions of the magnetic and electric field sensors: the system's origin is at the center of the launcher's volume, the x-axis corresponds to the shotline, and the z-axis is in the vertical direction.The overall precision to which position measurements were made is on the order of 15 cm.The magnetic field sensors consisted of 10 cm diameter, multiturn H-dot coils, whose signals were integrated in order to obtain the H-field components.The electric field sensors consisted of 10 cm long dipole antennas that were directly connected to conditioning electronics (e.g., amplifiers).The response of these sensors is essentially flat between 50 and 500,000 Hz.All sensors were calibrated in a TEM cell prior to this study.All signals were recorded using LeCroy 6810 waveform digitizers, which have a 12-bit resolution and channel widths as low as 0.2 p. AC coupling was used when electric fields were being measured.
The launcher was fired many times for the purposes of this study.During any one launch, three sensors were used, each determining one component of a field at three different locations.To obtain the other components of the field, the As the system for modelling the acceleration of the plate has been described in detail in previous publications [2,3], only an overview is presented.The launcher's electrical circuit is essentially an RLC circuit, differing from the simplest type in that the system inductance L varies with the position of the plate x.The launch process is modelled by coupling the RLC circuit differential equations with the equation of motion of the plate: in which F is the force on the plate and I is the current in the coil.Thus, if one knows the resistance R, the capacitance C, and the function L(x) (and thus the inductance gradient function dwdx), one can predict useful quantities such as the final velocity of the plate, the current trace, etc.The function U& can easily be measured with a simple inductance meter if the coil exists or modelled in the case of a hypothetical design.

B. Quasistatic Model
The primary function of this model is to calculate magnetic fields.It first assumes that the H field outside of a length of conductor of finite cross-section resembles the field due to an infinitesimally thin filament of the same length.It further assumes that the current is constant.A program ("BFIELD10") was written to simulate coil structures with any point around the structure.system treats the coil assembly as if it were a t " simple DC electromagnet.This makes it difficult to model the plate because all of the currents present in the plate dur-Lenz's Law 141 implies that the eddy currents induced plate will take forms that attempt to cancel the field ing launch are a direct Of *e changes in the magnetic Fig. 4. Launcher current as a function oftime, compared to the x-component of the magnetic field measured at a single point.
within the plate itself and will constantly adjust to follow the external field.While calculations based on Maxwell's equations can describe this process, they prove quite difficult to perform.As such, the complex electrodynamic process is in no way simulated in this model, but rather the effect is.The scheme is as follows: 1) a coil design and current are chosen, and its magnetic field is simulated as above; 2) a strucry loops is developed to simulate the plate; nts within these "plate" filaments are adjusted in such a way as to minimize the magnetic field within the "plate".The coarseness of any structure based on a finite number of filamentary loops precludes an exact cancellation, so some sort of minimization scheme must be resorted to.A diagram of the filamentary structures chosen for the present case is given in Fig. 2, where it is seen that the plate is modelled by 200 loops located at the upper and lower surfaces of the plate.The current in each upperAower loop pair is adjusted so that the H-field at the central point o€ their volume is zero.
With the overall field attributable to both the coils and the plate now approximated, the secondary function of the model ented: calculating the Lorentz forces on coil and plate structures.This not only gives useful information about likely modes of coil failure, but also allows us to calculate the total force on the plate when it is at a particular position x.Using (l), the value of dwbc at that point can then be inferred.Fig. 3 compares the results of a calculation for a 15.2 cm square, 5-turn square coil and 15.2 cm square plate with measured values.It can be seen that the predicted curve is slightly offset from and slightly higher than the measured curve.These defects appear to be artifacts associated with this particular plate model; it has been found that they can be corrected for if a slightly smaller plate is assumed.Nonetheless, the level of agreement indicates that the model is reasonable.

A. H-Field Results
H, , Hy , and HZ were measured at 10 pointssome near the side of the launcher, some alongside the shotline.In each case the shape of the trace closely follows that of the coil current, as illustrated in Fig. 4, where H,(t) €or the point (0 m,0.46 m,0.69 m) is shown alongside I($J (averaged over all shots).No signals were observed beyond the duration of the Iauncher's current pulse (>3 ms).
A log-log plot of peak H-field magnitudes H versus distance from the launcher Y can be found in be seen that the field has a complex structure.This complexity is not seen, however, when the fields produced by the coils, internal connections, and plate are simulated, as indicated by the solid line in FigS.A plate position of ~4 .6 5 cm was assumed in these simulations, as the plate acceleration model indicates that that is the position of the plate when the current pulse reaches its maximum.Much of the field's structure is recreated, however, when a filamentary model of the power feed cable is included in the simulation, as indicated by the crosses in Fig. 5.With this addition, the simulated fields are in reasonable agreement with the measured fields.Note that the model does not take into account fields attributable to other parts of the test stand circuit or of the effects of shielding by the coils or support structures (only shielding by the plate itself).The slope of the line in Fig. 5 is -2.45, not -3 as would be the case for a pure dipole field.Simulations indicate that while the field attributable to the coils themselves is dipolelike, the presence of the vertical internal connectors and the plate modifies this, particularly in and around the xy-plane, where most measurements were made.Characterizing the simulated peak launcher fields with the equation: we find that, along the x-or y-axes (using MKS units), a 4 .6 8 , nm2 for 6 1 m, and nm2.5 for r>l m.Along the zaxis a 4 .6 7 and nw3 for all r.If this were a pure dipole field, then n would equal 3 and the z-axis a would be twice that of the x-and y-axes cases.

B. E-Field Data
E-field data were taken at 23 points.Results for three points, (0.56 m, 0.23 m, 0 m), (1.12 m ,0.23 m, 0 m), and (1.68 m,0.23 m,O m), are given in Fig. 6.The response at these points is typical: at the start there is a small sinusoidal component that one can associate with the launch process, and then later there is a large bipolar signal one can associate with the plate during free flight.These two aspects of the data will be discussed separately.
Like the H-field signals, the launch phase signal resembles a decaying sinusoid, except that here it follows the voltage on the capacitor bank, not the current.Figure 7 contains a log-log plot of the peak fields in this phase plotted against distance from the center of its most likely source: the large plate that connects the switch to the HV terminals of the capacitors.Attempts were made to eliminate biases caused by the late-term signal.The large scatter in the data might be attributed to residual biases and shielding effects.Data points associated with locations that should clearly be shielded from the "HV plate" by the front aluminum shield are highlighted in Fig. 7, and they are indeed low.The line in Fig. 7 represents a fit to all the other points; its slope is -3, as would be expected from the dipole-like "HV plate"/ground plane combination.The large, late-term bipolar signals are clearly associated with the plate in flight, as they are not present when the launcher is triggered without a plate.Furthermore, if one assumes that their source's x-position corresponds to the sensors x-position at the time of crossover, then a source velocity of 50*2 m/s can be derived, in good agreement with the 49 m/s expected velocity of the plate.The signal amplitudes also decrease as the distance from the shotline increases.
There are several possible ways the plate could generate these electric field signals.For instance, if there were large residual eddy currents in the plate during flight, then the moving/decaying magnetic dipole field would generate an electric field.Calculations show, however, that fields such as this would be on the order of a few Vlm, even if hundreds of kA were involved, and would not explain the pulse shapes.Also, no magnetic field associated with the plate in flight was observed.An alternate source of this field would be surface charges on the plate, perhaps induced by friction during launch.Charges of only about lO-' C would be required to generate the kVlm fields observed.Unfortunately, it is acult to imagine a plausible charge distribution that can explain the observed pulse shapes.A moving point charge, for instance, would yield the following for sensors in the plane of the plate (as in Fig. 6): a bipolar x-component pulse, a monopolar y-component pulse, and zero z-component.The final, and perhaps most likely, explanation might be the distortion of static background electric fields caused by the presence of the conducting plate.These fields would not register on our AC coupled sensors, but changes in them caused by the moving plate would.Again, only modest charges would have to be present on nearby objects in order to generate fields of this magnitude.Furthermore, this would explain the poor sensor-to-sensor and/or shot-to-shot repeatability observed: fields would depend on local conditions and would fluctuate over time due to changes in humidity, etc.
V. CONCLUSION Magnetic and electric fields have been measured at many points around an EM induction launcher.A simple quasistatic model developed to simulate the magnetic field shows reasonable agreement w i t h measured values, and considerable insight has been obtained concerning the various contributors to this field.Short-term electric fields have been measured and their source identified as part of the power supply.Long-term E-fields have been tied to the plate, and possible explanations for them discussed.No evidence of residual eddy currents in the plate was found.

Fig. 2 .
Fig. 2. Typical filamentary structures used to model a launcher and plate.
lam en^^ segments, provided each is orix, y, or z directions.Currently it can automatically construct flat coils consisting of concentric rectangles or rectangular spirals, or it can simulate recthgular solenoids with parallel rectangular loops, given the required dimensions.It then calculates magnetic field strength and

Fig. 3 .
Fig. 3. Inductance gradient vs. plate position for a launcher based on 5 turn square spirals.The slight overestimation and offset in the model's results are persistent artifacts associated with the plate model used.The agreement can be improved if a slightly smaller plate is assumed.

Fig. 5 .
Fig. 5. Peak magnitude of the magnetic field as a function of radial distance fiom the center of the launcher coil, for all points studied Simulations indicate that much of the structure can be aitributed to the field associated with the power feed cable.

Fig. 6 .
Fig. 6.Components of the electric field as a function oftime, as measured at the three locations indicated.

Fig. 7 .
Fig. 7. Peak magnitude of the launch phase electric field as a hnction of radial distance ftom the center of the plate connected to the HV terminals of the capacitors.The slope of the fitted line is -3, indicating dipole-like behavior.

launcher\ feed cable sensors
were reoriented and the launcher fired again.For each launch, the capacitors were charged to 5.