A STUDY OF ELECTROMAGNETIC INDUCTION SYSTEMS FOR THE DETECTION OF UNEXPLODED ORDNANCE Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. ___________________________ Neha Jain Certificate of Approval: ______________________________ ______________________________ Stuart Wentworth Lloyd S. Riggs, Chair Associate Professor Professor Electrical and Computer Engineering Electrical and Computer Engineering ______________________________ ______________________________ Ramesh Ramadoss George T. Flowers Assistant Professor Interim Dean Electrical and Computer Engineering Graduate School A STUDY OF ELECTROMAGNETIC INDUCTION SYSTEMS FOR THE DETECTION OF UNEXPLODED ORDNANCE Neha Jain A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn, Alabama May 10, 2008 iii A STUDY OF ELECTROMAGNETIC INDUCTION SYSTEMS FOR THE DETECTION OF UNEXPLODED ORDNANCE Neha Jain Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. ______________________________ Signature of Author ______________________________ Date of Graduation iv THESIS ABSTRACT A STUDY OF ELECTROMAGNETIC INDUCTION SYSTEMS FOR THE DETECTION OF UNEXPLODED ORDNANCE Neha Jain Master of Science, May 10, 2008 (B. E., Chhotu Ram State College of Engineering, MDU University, 2004) 83 Typed Pages Directed by Lloyd S. Riggs This thesis presents a study of a time domain electromagnetic induction (EMI) system used for detection and discrimination of unexploded ordnance. In general, EMI system components include transmitter and receiver coils, corresponding transmitter and receiver coil amplifiers and a data acquisition system. This thesis explains differences in electronic circuitry of EMI sensors that measure the current or voltage response of a target. In the case of a current measurement it is essential to use a current-to-voltage converter as the first stage before any amplification is done so that the lower 3 dB point of the receiver coil is sufficiently low to capture all important target characteristics. Also using a compensation circuit to lower the 3 dB point makes it possible to acquire the undistorted late time response of targets. This capability is essential for discrimination purposes. In voltage measurement methods, however, only a voltage amplifier is used on v the receiver side and there is no need for compensation circuitry as the 3 dB point of the receiver coil is high enough to capture an undistorted version of the target response. Both voltage and current measurement methods were used to record the target response and it was observed that in general, the voltage measurement method is better than the current measurement method. Comparison of the response collected from a commercial EMI sensor and a constructed EMI sensor (voltage method) is made. Observations are made for possible reasons for superior response from the commercial EMI sensor In addition to the above topics, a software tool is developed for evaluating electromagnetic transmitter coil designs and configurations that are used in EMI sensor arrays. This software tool generates static sensitivity maps in any desired plane when transmitter and receiver coil space coordinates are given as inputs. These maps give a good idea of how effective an EMI sensor layout can be in terms of its sensitivity to a target?s presence at different points in space. The advantage of using static sensitivity maps is that they are independent of target characteristics and thus can be used as an effective tool to measure the performance of a proposed coil layout. Maps have been presented for different coil configurations and important observations and conclusions are made. To verify the fidelity of the software tool output field measurements are taken using a commercial sensor and compared to the same from the software model. In general, good qualitative agreement was obtained though more careful measurements should be done in the future. In addition to sensitivity maps, streamline plots can also be generated from a similar software tool with minor changes in software coding. Streamline plots are helpful in visualizing the direction of the magnetic field due to transmitter or receiver coils in any orientation. vi ACKNOWLEDGEMENTS I would like to express my gratitude to Dr. Lloyd S. Riggs, my advisor, for providing technical direction, support and guidance during the course of this research. I would like to thank Dr. Stuart Wentworth and Dr. Ramesh Ramadoss for agreeing to serve on my thesis committee. My research team members Sailaja, Jithendra also deserve my appreciation for their support. I would like to thank my mother, Kamini Jain for her enduring love, and moral support throughout my life. Finally, my husband, Varun Mahajan deserves credit for his constant encouragement, patience, and sacrifice. vii Style manual of journal used: Graduate School: Guide to preparation and submission of theses and dissertations Computer software used: Microsoft Office 2003, Vista viii TABLE OF CONTENTS LIST OF FIGURES ............................................................................................................ x 1. INTRODUCTION .......................................................................................................... 1 2. BASICS OF ELECTROMAGNETIC INDUCTION SENSOR ..................................... 4 2.1 Introduction to EMI sensors ..................................................................................... 4 2.2 Simple Circuit Model of a Pulsed EMI system ........................................................ 5 2.2.1 Object Current .................................................................................................... 7 2.2.2 Open circuited voltage at the receiver.............................................................. 11 2.3 Summary ................................................................................................................. 13 3. HARDWARE IMPLEMENTATION OF EMI SENSOR ............................................ 14 3.1 CW EMI System ..................................................................................................... 14 3.1.1 Receiver current and output voltage expressions in frequency domain ........... 14 3.1.2 Receiver current measurement using current-to-voltage converter and compensation circuit ................................................................................................. 17 3.1.3 Analytical explanation of compensation circuit ............................................... 21 3.1.4 Receiver voltage measurement using voltage amplifier .................................. 24 3.2 Receiver current expression in time domain ........................................................... 26 3.3 Summary ................................................................................................................. 31 4. TIME DOMAIN EM63 SENSOR ................................................................................ 32 4.1 General description ................................................................................................. 32 4.2 Major components of the system ............................................................................ 33 4.3 Data Collection ....................................................................................................... 34 4.4 Summary ................................................................................................................. 34 5. TIME DOMAIN EMI SENSOR MEASUREMENTS ................................................. 35 5.1 Measurement Setup ................................................................................................. 35 5.2 Measurements from TD sensors ............................................................................. 38 5.2.1 Response of 4 inch steel sphere ....................................................................... 41 5.2.2 Response of 1 by 4 inch and 2 by 8 inch steel cylinder ................................... 42 5.3 Summary ................................................................................................................. 45 ix 6. SENSITIVITY AND STREAMLINE PLOTS FOR EMI SENSOR ........................... 46 6.1 Introduction ............................................................................................................. 46 6.1.1 Equations for static sensitivity ......................................................................... 46 6.2 Sensitivity and streamline plots for different coil configuration ............................ 48 6.2.1 Simple configuration of two square coils ........................................................ 48 6.2.2 Square transmitting coil and figure 8 receiver coil .......................................... 53 6.2.3 Stacked arrangement of three square coils ....................................................... 58 6.2.4 Two square transmitter and two square receiver coils in same plane (XY plane) ................................................................................................................................... 59 6.3 Comparison of measured to calculated sensitivity plot for the Geonics EM-63 .... 61 6.4 Summary ................................................................................................................. 63 7. CONCLUSION ............................................................................................................. 65 BIBLIOGRAPHY ............................................................................................................. 67 APPENDIX A ................................................................................................................... 69 A.1 Using the software tool ......................................................................................... 69 A.2 Explanation of the Code ........................................................................................ 69 A.3 Limitations of the code .......................................................................................... 70 x LIST OF FIGURES Figure 2.1 Typical electromagnetic induction system consisting of transmitter and receiver coil, shown here in the presence of a buried metallic object. ............................... 5 Figure 2.2 Simple Transmitter Coil Model including parasitic capacitance effects and Rshunt that is used to reduce ringing during turn off of transmitter current. ..................... 6 Figure 2.3 Current Waveforms through the transmitter coil when the closed switch was suddenly opened at 20ms. Red curve is for very high value of Rshunt, green is for low value of Rshunt and blue curve is for very low value of Rshunt. ....................................... 6 Figure 2.4 Magnetic Coupled Circuit representing a TD / CW EMI sensor [4]. ................ 7 Figure 2.5 Normalized transmitter current versus normalized time. .................................. 8 Figure.2.6 Circuit model of object explaining the charge up and decay of object currents. ............................................................................................................................................. 8 Figure 2.7 Plots for (a) Transmitter Current (b) Object Current (c) Object Coupled Receiver Voltage and (d) Direct Coupled Receiver Voltage. Dotted vertical line marks the time t=T. ........................................................................................................................ 9 Figure 3.1 Current-to-voltage converter with gain of 200. ............................................... 18 Figure 3.2 Input current in dB from the receiver coil. ...................................................... 18 Figure 3.3 Output voltage in dB from differential current-to-voltage converter with 3 dB point at 10 Hz. ................................................................................................................... 19 Figure 3.4 Current-to-voltage converter with compensation circuit. ................................ 20 Figure 3.5 Frequency response of output voltage in dB from current-to-voltage converter with compensation circuit. The 3dB point is shifted from 10 Hz to 100 mHz. ................ 20 Figure 3.6 A compensation circuit that compensates for the roll off at 20dB/decade of the low frequency response of the receiver coil. .................................................................... 22 Figure 3.7 Plot of gain in dB versus frequency for the compensation circuit. ................. 22 Figure 3.8 Plot of compensated circuit phase in degrees versus frequency. ..................... 23 Figure 3.9 Voltage amplifier in non inverting configuration connected to receiver coil. The gain of the amplifier is given by ? ? ? ? ? ? ? ? + 3 2 1 R R . ............................................................... 25 xi Figure 3.10 Plot of Vin versus frequency. Pink curve shows critically damped response. Light green curve is for < C(parasitic) and dark green curve is for t >> C(parasitic). Resonance frequency of the coil is set to 10 KHz. ...................................... 25 shunt R shun R Figure 3.11 Interaction between the receiver and object circuits. .................................... 27 Figure 3.12a Plot of normalized receiver current versus time for current measurement with = 0.1Hz and =1000 Hz. ................................................................................... 30 R f O f Figure 3.12b.Plot of normalized receiver voltage versus time with = 10000 Hz and =1000 Hz. .............................................................................................................. 31 R f O f Figure 4.1 EM63 sensor. ................................................................................................... 33 Figure 5.1 Setup used to excite the transmitter coil. ......................................................... 36 Figure 5.2 Differential voltage amplifier with Vout= 1 2 R R (Vin2-Vin1). .......................... 37 )32( RRI in + .Figure 5.3 Current to voltage converter with Vout= .................................. 38 Figure 5.4 Small (Pink curve), medium (Blue Curve) and large (Green Curve) copper 39 43 ring response with current measurements (Log/Linear). .................................................. Figure 5.5 Small (Pink curve), medium (Blue Curve) and large (Green Curve) copper ring response with voltage measurements using differential configuration (Log/Linear). 39 Figure 5.6 Blue, green and red dotted curves are responses from small, medium and large copper rings respectively using EM63 sensor and the corresponding overlapping black curves are from our sensor at 5 inch height. Dashed blue, green and red curves are for 25 inch height using EM63 sensor (Log/Linear). .................................................................. 40 Figure 5.7a Black curves are responses of 4 inch steel sphere from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method at heights of 3, 6, 9 and 12 inches respectively and red dotted curves are from EM63 at heights of 2.5, 5, 10, 15 and 20 inches respectively (Log/Linear). ........................................................... 41 Figure 5.7b Black curves are responses of 4 inch steel sphere from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method at heights of 3, 6, 9 and 12 inches respectively and red dotted curves are from EM63 at heights of 2.5, 5, 10, 15 and 20 inches respectively (Log/Log). ............................................................... 42 Figure 5.8 Black curves are responses of 1 by 4 inch cylinder in vertical positions at heights of 3,6,9,12 inches from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from the EM63 sensor at heights of 2.5,10 and 12.5 inches respectively (Log/Log). ............................................... Figure 5.9 Black curves are responses of 1 by 4 inch cylinder in horizontal positions at heights of 3,6,9,12 inches from our sensor using the high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from EM63 sensor at heights of approximately 2.5 and 10 inches respectively (Log/Log). ............... 43 xii Figure 5.10 Black curves are responses of 2 by 8 inch steel cylinder in vertical positions at heights of 3,6,9,12 inches from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from EM63 sensor at heights of 10 and 25 inches respectively (Log/Log). ....................................................... 44 Figure 5.11 Black curves are responses of 2 by 8 inch steel cylinder in horizontal positions at heights of 3,6,9,12 inches from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from EM63 sensor at heights of approximately 2.5,15 and 20 inches respectively (Log/Log). .......... 44 Figure 6.1 Conductor is shown as dark heavy line and the magnetic field due to it at P is into the page [18]. ............................................................................................................. 48 Figure 6.2 Square transmitter and receiver coils with dimensions 984 mm and 890 mm respectively. Current flows in counterclockwise direction in both the coils. ................... 49 Figure 6.3a Sensitivity contours for coil configuration of Figure 6.2 computed over the plane x=0 cm, -80cm>1, the receiver voltage drops rapidly from the maximum value to the lower limit at t=T, and then the exponential decay starts, as indicated in Figure 2.7(c). There is also a direct coupled voltage at the receiver coil due to the coupling between the transmitter and receiver. The direct coupled voltage can be given as ? ? ? ? ? ? ? > ?? < =?= Tt Tt T IM t tI dt d MV ITR TTRDIRECT 0 0 00 )( (2.14) As mentioned earlier, using a figure eight coil the direct coupling component can be greatly reduced and if the residual coupling between the transmitter coil and the receiver coil is denoted as rscript r represents residual) then r TR M (where the supe 12 13 dt tdI MtV Tr TR r )( )( ?= ? (2.15) The sign depends upon the winding direction of the halves of the receiver coil with respect to that of the transmitter coil and upon which half has the larger direct coupling. Therefore direct-coupled voltage can add or subtract from the object coupled receiver voltage for times 0?t?T [4-5]. 2.3 Summary In this chapter the basic operating principle of an EMI sensor was introduced and circuit analysis was employed to describe the operational characteristics of a time domain EMI system. Also the importance of clean (without oscillation) turn-off of the transmitter current pulse is stressed which if not obtained can lead to distortion in the object response. In the next chapter a frequency domain analysis of EMI sensors is presented. Also some of the important issues related to methods of measuring the object response will be addressed. Finally specific hardware used to construct a pulsed EMI sensor will be discussed. CHAPTER 3 HARDWARE IMPLEMENTATION OF EMI SENSOR In this chapter frequency domain analysis will be used to derive the transfer function that describes both voltage and current measurements of a nonmagnetic finitely conducting object. 3.1 CW EMI System As mentioned earlier a CW system usually employs a discrete set of sine waves to excite the transmitter coil. In general higher frequencies provide better sensitivity than lower frequencies but lower frequencies penetrate deeper into the metallic objects (due to reduced skin effect) and thus provide more information about the internal structure of an object [8]. 3.1.1 Receiver current and output voltage expressions in frequency domain Referring to Figure 2.4 for CW systems, if the source drives a current I T through the transmitter coil resistance and inductance R T and L T , respectively, and the coupling between the transmitter coil and object, and object and receiver coil can be denoted by M TO and M OR , respectively, then the object current can be written as [10] OO TTO O LjR IMj I ? ? + = (3.1) 14 Also the transmitter current can be written as ` TT S T LjR V I ?+ = (3.2) where is the source voltage. Substituting (3.2) in (3.1) yields S V )()( TT S OO TO O LjR V LjR Mj I ?? ? ++ = (3.3) The receiver current is given as the voltage induced in the receiver coil divided by the sum of the receiver coil impedance and load impedance or )( LRR OOR R ZLjR IMj I ++ = ? ? (3.4) Substituting (3.3) in (3.4) we get )()()( TT S OO TO LRR OR R LjR V LjR Mj ZLjR Mj I ?? ? ? ? ++++ = (3.5) Similarly the output voltage across the load is the product of the receiver current and load impedance or )()()( TT S OO TO LRR LOR out LjR V LjR Mj ZLjR ZMj V ?? ? ? ? ++++ = (3.6) In the case of a receiver current measurement (instead of receiver voltage measurement) the load is assumed short circuited and thus can be assumed zero. Using this assumption in (3.5) we get L Z )()()( TT S OO TO RR OR R LjR V LjR Mj LjR Mj I ?? ? ? ? +++ = (3.7) or 15 )/1( / )/1( / )/1( / T TS R R O O RO STOOR R j RV j j j j LL VMM I ???? ?? ?? ?? +++ = (3.8) Where, OOO LR=? , TTT LR=? RRR LR=? are object, transmitter and receiver coil break frequencies, respectively. In a voltage measurement, if is very high (open circuit receiver coil) (3.6) may be written as L Z )()( TT S OO ORTO out LjR Vj LjR MMj V ? ? ? ? ++ = (3.9) or )1()1( O O T T RT STOOR out j j j j LL VMM V ?? ?? ?? ?? ++ = (3.10) Actually, in practice if the load is made too large, current will be forced to flow in the parasitic receiver coil capacitance and undesirable ringing or oscillations will occur. To avoid oscillations the receiver coil is usually terminated in a resistance Rshunt that leads to the following modified form of (3.6) : RshuntR shunt O O T T RT STOOR out LjRR R j j j j LL VMM V ??? ?? ?? ?? ++++ = )(()1()1( (3.11) One can see from (3.8) that in a receiver current measurement the receiver coil behaves as a high pass filter while the transmitter filter characteristics are low pass. Important discrimination information is present in the late time response. Therefore, it is desirable to shift the 3 dB point of the receiver coil to as low a frequency as possible. The break frequency ( ) of the receiver coil is inversely proportional to the number of turns on the receiver coil. Therefore, increasing the number of turns will decrease the break frequency. However, a point is reached where increasing the number of turns does not significantly lower the break frequency (a point of diminishing return is reached). As R f 16 discussed in the next section compensation circuitry may be employed to further reduce the break frequency. In a voltage measurement, however the transmitter coil displays high pass filter characteristics while the receiver displays low pass filter characteristics due to which a compensation circuit on the receiver side is not required. Also, the number of turns for a low pass receiver coil in a voltage measurement is usually much less than that for a high pass receiver coil for a current measurement. 3.1.2 Receiver current measurement using current-to-voltage converter and compensation circuit Figure 3.1 shows the circuit diagram of a current-to-voltage converter that is used for current measurements. The two halves of the receiver coil are wound in opposite directions so that direct coupling between transmitter and receiver coils is minimized (ideally zero). The receiver coil that is used has a break frequency of 10 Hz and as explained above acted as a high pass filter for current measurements. A simple RC filter circuit that has a cut off frequency of 10 Hz may be used to model the receiver coil frequency response. A numerical circuit analysis software called Pspice was used to compute the response. The input current from the coil is assumed to be 10 mA which gives an output voltage of 2V. In accordance with operational amplifier theory the gain of current-to-voltage converter is given as 1.3 )( 23 RR + or 200. The output voltage of the current-to-voltage converter is given by ) 23 R(RIV inout += . Figure 3.2 and 3.3 provide plots of the input current and output voltage as obtained from the Pspice. 1.3 Pspice was developed by MicroSim which was acquired by Orcad and now belongs to Cadence Design Systems. For more information please visit www. Cadence.com 17 18 Figure 3.1 Current-to-voltage converter with gain of 200. from the receiver coil. Figure 3.2 Input current in dB 3 dB point In p u t cu rre nt i n dB 3 dB point Out put v ol t a ge i n dB Figure 3.3 Output voltage in dB from differential current-to-voltage converter with 3 dB point at 10 Hz. A compensation circuit may be used to shift the 3 dB break frequency lower in order to achieve the desired late time response [4], [11]. The final circuit configuration is shown in Figure 3.4. In this circuit the actual break frequency of the coil is given as () CR actualf break 1 2 1 ? = and the desired break frequency is given as CR desiredf break 2 2 1 )( ? = . Thus has to be as large as possible. The gain at higher frequencies is determined by 1 R d is equal to 2 1 R he actual break frequency of the coil is 10 Hz and at arbitrarily high frequencies a gain of 200 is required. Therefore is set to 100 ohms and thus C can be determined using 2 R an T. 1 R CR actualf break 1 2 1 )( ? = . Similarly the desired break frequency is chosen as 0.1Hz and therefore can be calculated using 2 R CR desiredf break 2 2 1 )( ? = which gives =10K ohms. Figure 3.5 shows frequency response of the output voltage from the compensated circuit. The input current frequency response is the same as shown in Figure 3.2. 2 R 19 R 2 R 1 C R 2 R 1 C Figure 3.4 Current-to-voltage converter with compensation circuit. Out put v ol t a ge i n dB 3 dB point Figure 3.5 Frequency response of output voltage in dB from current-to-voltage converter with compensation circuit. The 3dB point is shifted from 10 Hz to 100 mHz. 20 3.1.3 Analytical explanation of compensation circuit The total impedance Z across the compensation circuit shown in Figure 3.6 is given as 1 21 2 1 1 1 1 Cj RR R Cj R Z ? ? ++ ? ? ? ? ? ? ? ? + = (3.12) After rearranging terms, (3.12) can also be written as () ? ? ? ? ? ? ? ? ? ? ? ? + + + + = 211 11 21 21 1 1 1 1 RRCj RCj RR RR Z ? ? (3.13) Assuming ? , (3.13) reduces to 2 R 1 R ? ? ? ? ? ? ? ? ? ? ? ? + + = 21 11 1 1 1 1 1 RCj RCj RZ ? ? (3.14) or ? ? ? ? ? ? ? ? ? ? ? ? + + = 21 11 1 1 1 RC j RC j RZ ? ? (3.15) Expressing (3.15) in terms of the laplace frequency domain variable s yields ? ? ? ? ? ? ? ? ? ? ? ? + + = 21 11 1 1 1 RC s RC s RZ (3.16) Thus from (3.16) one may observe that the pole of the response is set by and the zero by . Thus 12 CR 11 CR 11 2 1 CR f Z ? = (3.17) and 21 12 2 1 CR f P ? = (3.18) Figure 3.6 A compensation circuit that compensates for the roll off at 20dB/decade of the low frequency response of the receiver coil. Figure 3.7 shows the plot of the gain in dB versus frequency with =0.1Hz and =10 Hz. From the figure we can see that there is a 20 dB/decade roll off between the pole and zero frequencies. p f Z f 10 -4 10 -3 10 -2 10 -1 10 0 10 1 10 2 40 45 50 55 60 65 70 75 80 Frequency pole ga i n ( d b ) gai n ( d B ) Zero Figure 3.7 Plot of gain in dB versus frequency for the compensation circuit. 22 Since the receiver coil current has high a pass filter characteristic with a cut off frequency of 10 Hz below which its response increases by 20 db/decade, the roll off of the compensation circuit will exactly cancel this increase thereby pushing the receiver coil current break frequency to 0.1 Hz. Figure 3.8 shows a plot of the phase of the compensated circuit in degrees versus frequency. From this plot one can observe that at the pole and zero frequencies the phase is 45 degrees. From (3.16) it can also be seen that the low frequency gain of the compensation circuitry depends on and to shift the pole frequency lower by a decade, has to be made 10 times greater. Increasing the gain at lower frequencies limits the bandwidth of the opamp by an amount that is dependent upon the opamp?s gain bandwidth product. 2 R 2 R 10 -3 10 -2 10 -1 10 0 10 1 10 2 10 3 -80 -70 -60 -50 -40 -30 -20 -10 0 Frequency P h as e i n deg r e es Figure 3.8 Plot of compensated circuit phase in degrees versus frequency. 23 3.1.4 Receiver voltage measurement using voltage amplifier According to Figure 3.9, the voltage amplifier is built using a non inverting opamp configuration so that the receiver coil is terminated in a large resistance. Without Rshunt some frequency will be reached where all the coil current will flow in the parasitic capacitance because of the large resistance between the positive terminal of the opamp and ground. From (3.11) it can be seen that the receiver coil frequency response is low pass while the object response is high pass. Therefore, it is desirable to shift the receiver coil break frequency as high as possible so that object response is not attenuated above the 3 dB point of the receiver coil. The break frequency of the receiver coil is modified due to to shunt R R shuntR break L RR f + = . Thus by making very large, a higher break frequency can be achieved but can?t be too large or current will flow in the parasitic capacitance of the coil resulting in oscillations. When is made large in comparison to the impedance of the parasitic coil capacitance a certain frequency will be reached above which currents will start flowing in the parasitic coil capacitance that forms a resonant circuit composed of , and C(parasitic). Under this condition Vin will increase proportionally with frequency at 20 dB/dec up-to the first self-resonance frequency of the receiver coil (typically around 10 kHz). This in-turn distorts the output voltage, which is no longer a faithful representation of the object response. Figure 3.10 shows Vin versus frequency for different values of with respect to C(parasitic). shunt R shunt R shunt R R L R R shunt R 24 Figure 3.9 Voltage amplifier in non inverting configuration connected to receiver coil. The gain of the amplifier is given by ? ? ? ? ? ? ? ? + 3 2 1 R R . Input Voltage in dB Figure 3.10 Plot of Vin versus frequency. Pink curve shows critically damped response. Light green curve is for R < C(parasitic) and dark green curve is for R >> C(parasitic). Resonance frequency of the coil is set to 10 KHz. shunt shunt 25 3.2 Receiver current expression in time domain According to Figure 3.11 the circuit equations for both object and receiver current after the transmitter current turn off can be written as KIMjIRILj ROROOOO =++ ?? (3.19) and 0=++ OORRRRR IMjIRILj ?? (3.20) where K = and is the steady state object current just before transmitter current turn off and and ject and receiver coil currents after transmitter current turn off. Writing (3.19) and (3.20) in matrix form results in ? OO IL O I ? O I R I are ob (3.21) ? ? ? ? ? ? = ? ? ? ? ? ? ? ? ? ? ? ? + + 0)( )( K I I LjRMj MjRLj R O RROR OROO ?? ?? Thus can be written as R I ? ? ? ? ? ? + + ? ? ? ? ? ? + = )( )( 0 )( RROR OROO OR OO R LjRMj MjRLj Mj KRLj I ?? ?? ? ? (3.22) Expanding (3.22) in terms of the laplace transfer variable s yields [] 2 )())(( ORRROO OR R sMsLRRsL KsM I ?++ ? = (3.23) Since is very small therefore neglecting its square in (3.23) leads to OR M []))(( RROO OR R sLRRsL KsM I ++ ? = (3.24) 26 or ? ? ? ? ? ? ? ? ++ ? = ))(( R R O O RO OR R L R s L R sLL KsM I .25) ing in terms of partial fractions leads to (3 Expand )( )( )( )( O ORRO OOR R ORRO ROR fKM R fs ffLL fKM fs ffLL I + ? + + +? = (3.26) where R f = R R L R and O O O L R f = or 27 ? ? ? ? ? ? ? ? + ? ++? = )()()( O O R R ORRO OR R fs f fs f ffLL KM I (3.27) omain results in Converting (3.27) to time d [ ] tf O tf R ORRO OR R KM I = OR efef ffLL ?? ? +? )( (3.28) OR M O I R I Figure 3.11 Interaction between the receiver and object circuits. From (3.28) it can be inferred that the receiver current consist of two exponentials with n the receiver (decay rates that depend o R R L R R =? ) and object time constants ( O O O R L =? ). It was mentioned earlier in connection with the frequency domain analysis that in a 28 eak frequency (specifically from 10 Hz to 0.1 Hz). Thus assuming to be very low as compared to (3.28) can be written as current measurement the receiver coil has high pass filter characteristic and therefore a compensation circuit was used to lower its br R f O f ? ? ? ? ?= O R RO OR R OR ee fLL I (3.29) The information regarding the object is in the exponential that decays with object tim constant 1? o f , however the exponential that decays with receiver time constant is small ?? ?? tftf fKM e in c very slowly due to sma ement the break frequency of the receiver has to be very low to ppress the unwanted receiver time constant exponential so that it does not interfere with the response of the object amplitude and also de ll R f . Thus in a current measurays su during the time of collection of data. From (3.28) the voltage input (Vin) to the voltage amplifier of Figure 3.9 can be written as [ ] tftf breakRO shuntOR f(LL RKM ?? +? Obreak O in break efef )f V ? (3.30) if parasitic capacitance eff O = ects are neglected. In (3.30) R break L f = and shuntR RR + O O L f = . Since in a O R voltage m ent the receiver coil behaves as a low pass filter the 3 dB to igh enough so that high pass object response is not distorted. Assuming >> (3.30) can be written as easurem be hpoint of the coil has break O f f ? ? ? tf O ? ? ? break shunt RO in fLL ? +? ? O tf OR break e f eRV (3.31) = KM 29 ntial with high de then object exponential decays quickly and does not interfere with object response. One way to achieve this is to make sure that is large which is only possible if C (parasitic) is It can be concluded that in general a voltage measurement is much better than a current measurement. However, if the dynamic range of response of an object is large then the more effective method depends upon the break frequency of the receiver coil with respect to object break frequency and also upon the time window over which the data is collected. For illustration purpose, if we consider = 0.1 and =1000 and the time Thus from (3.31) it can be noticed that in a voltage measurement break f has to be as high as possible so that unwanted receiver expone er amplitu shunt R small so that resonance is avoided. C (parasitic) can be reduced by decreasing the number of turns of the receiver coil or by modifying the receiver coil geometry. As in voltage measurement, receiver response decays quickly and does not interfere with the object late time response which is critical for discrimination purposes. R O window for data collection is 20 ms then the plot of normalized R I for a current measurement is shown in Figure 3.12(a). f f 0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 -120 -100 -80 -60 -40 -20 0 Time in sec Rec ei v er C ur r en t i n db Object Receiver d B Figure 3.12a Plot of normalized receiver current versus time for current measurement with = 0.1Hz and =1000 Hz. R f O f For a voltage measurement considering = 10000 and =1000 Hz, the receiver response is as shown in Figure 3.12(b). Clearly form Figure 3.12(a) and (b) it can be inferred that a voltage measurement provides a larger object dynamic range. R f O f 30 31 0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 -200 -180 -160 -140 -120 -100 -80 -60 -40 -20 0 Time in sec R ec ei v er C ur r ent i n db Receiver Object vol t a ge i n dB Figure 3.12b.Plot of normalized receiver voltage versus time with = 10000 Hz and =1000 Hz. R f O f 3.3 Summary In this chapter voltage and current measurement methods are compared using continuous wave EMI system analysis. Also a detailed explanation of the compensation circuit needed in a current measurement is discussed. In either case (voltage or current measurement) a low pass filter can be used to eliminate high frequency noise. In the next chapter the hardware configuration of the EM63 sensor (a commercial sensor) is introduced briefly in order to facilitate comparison of data collected from our sensor and that of the EM 63. 32 CHAPTER 4 TIME DOMAIN EM63 SENSOR 4.1 General description The EM63 is a high power, high sensitivity, and wide bandwidth full time domain UXO detector and consists of a powerful transmitter coil that generates a pulsed primary magnetic field which induces eddy currents in nearby metallic objects. The decay of these eddy currents generates a secondary magnetic field with a specific decay rate that is determined by target size, shape, orientation and composition. This secondary magnetic field decay thus contains important information towards complete characterization of the target. Figure 4.1 shows an assembled EM63 sensor [12]. The EM63 measures the complete transient response over a wide dynamic range. Time measurements are recorded at 26 geometrically spaced gates, covering a time range from 180us to 25ms.The data acquisition is supported by the PRO4000 field computer which is able to simultaneously receive GPS data for position control [13]. Figure 4.1 EM63 sensor. 4.2 Major components of the system The main components of the system are a control console, trailer mounted transmitter coil and sensors, battery pack, preamplifier, acquisition and editing software. The trailer coil assembly consists of a transmitter coil and three receiver coils as shown in Figure 4.1. A bipolar current waveform is used for excitation of the transmitter coil which is 1m?1m in size. The main receiver coil is coincident with the EM63 source and is 0.5m?0.5m in size. The second receiver coil axially mounted with the main coil is used for target depth determination and is 60cm above the main coil and is also 0.5m?0.5m in size. The EM63 control console contains the system computer and associated electronics for drive of the transmitter coil and process signals from the receiver coil preamplifier. 33 34 4.3 Data Collection The gates of EM63 are narrow during the early time of the response when the signal is very strong and become wider during late time because the signal becomes very weak and to get good S/N ratio time gates have to be wider at later times. The data for a target at a particular location is collected in a horizontal row which contains data reading from each of the 26 gates in millivolts. The measuring range of the EM63 extends up to 10,000 mV. 4.4 Summary The configuration of the EM63 sensor has been discussed briefly with the intent to compare and contrast the response of the EM63 sensor with the response of the sensor developed for this thesis. 35 CHAPTER 5 TIME DOMAIN EMI SENSOR MEASUREMENTS In this chapter the configuration of our sensor is described (see Figure 5.1). Measurements of copper rings and spherical targets are acquired with our sensor and the EM63 and the two data sets are compared. 5.1 Measurement Setup A HP 3314A function generator is used to generate a square wave signal (frequency=50 Hz) which is fed to a Powertron Model 250A (250 watts) power amplifier in order to boost the current flowing through the transmitter coil. A resistor is connected in parallel with the transmitter coil in order to reduce ringing. A Tektronix current probe is used to measure the current flowing in the transmitter and a Tektronix TDS3054B oscilloscope is used for display purposes. Figure 5.1 shows the general block diagram of all the instruments and the connections used to excite the transmitter. For voltage measurements two receiver coils were wrapped in the same direction and connected to provide a single coil with a centre tap (three wire option). The ends of the receiver coil provide inputs to the + and ? terminals of a differential amplifier (voltage amplifier) and the center tap was connected to ground as shown in Figure 5.2. The Rshunt must be adjusted to reduce ringing as was described in the previous chapter. Most of the voltage measurements described here were made using a high performance op-amp manufactured by Linear Technology (the LT1028) [14]. For current measurements the receiver coils were wrapped in opposite directions so that the voltage induced from each half cancels each other. The general schematic of the current-to-voltage converter (excluding the compensation circuit) is shown in Figure 5.3. Receiver Coil Function Generator Frequency Amplitude out Power Amplifier(250Amp) Signal i/p Hi o/p Lo o/p Resistor Transmitter Coil Oscilloscope Current Probe Figure 5.1 Setup used to excite the transmitter coil. A MAX430 operational amplifier manufactured by MAXIM was used in the current measurements because it has very low input offset and drift as compared to conventional opamps [15]. The output voltage from the two measurements can be analyzed using an oscilloscope or using a data acquisition card installed in a desktop computer. A 16 bit analog to digital PCI bus card (Compuscope (CS) 1602 from Gage 36 Applied Inc [16]) was installed on a desktop computer and the Gagescope data acquisition software was used for storing and displaying data acquired by CS1602. The software also provides controls for changing the sampling rate, trigger level, data points, etc. Figure 5.2 Differential voltage amplifier with Vout= 1 2 R R (Vin2-Vin1). 37 Figure 5.3 Current to voltage converter with Vout= . )32( RRI in + 5.2 Measurements from TD sensors The first three targets that were used to compare the performance of our sensor to that of the EM63 sensor are three copper rings of approximately the same mean radius but of varying thickness. Their responses are compared with those acquired by the EM63. The small copper ring has the fastest decay rate and therefore has the maximum dynamic range while the large ring has the longest time constant and therefore the least dynamic range. The responses from both current and voltage measurements using our sensor are shown in Figure 5.4 and 5.5. 38 0.0001 0.001 0.01 0.1 1 0 0.005 0.01 0.015 0.02 0.025 Time(seconds) V o l t ag e( v o l t s) Figure 5.4 Small (Pink curve), medium (Blue Curve) and large (Green Curve) copper ring response with current measurements (Log/Linear). 0.0001 0.001 0.01 0.1 1 10 0 0.005 0.01 0.015 0.02 0.025 Time (Seconds) V o l t ag e( v o l t s) Figure 5.5 Small (Pink curve), medium (Blue Curve) and large (Green Curve) copper ring response with voltage measurements using differential configuration (Log/Linear). 39 It can be easily seen that the voltage measurement has significantly more dynamic range than the current measurement. This observation is in agreement with the theory presented in the previous chapters. Furthermore, the voltage measurements were taken out of doors while the current measurements were taken in the laboratory and the latter environment probably has more ambient noise (particularly 60 Hz noise). However our sensor response is still inferior to that of the EM63 possibly due to fact that our transmitter turn off response is exponential and slower than the linear current turn off response of the EM63. Figure 5.6 shows a comparison of the EM63 response for small, medium and large copper rings with respect to the response of our sensor. d B Figure 5.6 Blue, green and red dotted curves are responses from small, medium and large copper rings respectively using EM63 sensor and the corresponding overlapping black curves are from our sensor at 5 inch height. Dashed blue, green and red curves are for 25 inch height using EM63 sensor (Log/Linear). 40 5.2.1 Response of 4 inch steel sphere Figure 5.7a compares the response of a 4 inch steel sphere from each sensor. It can be seen from the figure that the sphere response does not have nearly the dynamic range that the copper rings have and therefore the two measurements are in good agreement. Figure 5.7b presents the same information in log versus log scale. Figure 5.7a Black curves are responses of 4 inch steel sphere from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method at heights of 3, 6, 9 and 12 inches respectively and red dotted curves are from EM63 at heights of 2.5, 5, 10, 15 and 20 inches respectively (Log/Linear). 41 Figure 5.7b Black curves are responses of 4 inch steel sphere from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method at heights of 3, 6, 9 and 12 inches respectively and red dotted curves are from EM63 at heights of 2.5, 5, 10, 15 and 20 inches respectively (Log/Log). It should be noted further that in case of the copper rings the responses are approximately straight lines on a log/linear plot while on a log/linear plot the sphere responses have an exponential shape. Thus the general shape of the response can be used to easily distinguish between ferrous and non ferrous objects. 5.2.2 Response of 1 by 4 inch and 2 by 8 inch steel cylinder Figures 5.8 and 5.9 present data for a 1 by 4 inch steel cylinder when oriented with its axis perpendicular and parallel to the incident magnetic field. Figures 5.10 and 5.11 present similar data for a 2 by 8 inch steel cylinder. 42 Figure 5.8 Black curves are responses of 1 by 4 inch cylinder in vertical positions at heights of 3,6,9,12 inches from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from the EM63 sensor at heights of 2.5,10 and 12.5 inches respectively (Log/Log). 43 Figure 5.9 Black curves are responses of 1 by 4 inch cylinder in horizontal positions at heights of 3,6,9,12 inches from our sensor using the high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from EM63 sensor at heights of approximately 2.5 and 10 inches respectively (Log/Log). Figure 5.10 Black curves are responses of 2 by 8 inch steel cylinder in vertical positions at heights of 3,6,9,12 inches from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from EM63 sensor at heights of 10 and 25 inches respectively (Log/Log). Figure 5.11 Black curves are responses of 2 by 8 inch steel cylinder in horizontal positions at heights of 3,6,9,12 inches from our sensor using high gain INA103KP instrumentation amplifier in voltage measurement method and red curves are from EM63 sensor at heights of approximately 2.5,15 and 20 inches respectively (Log/Log). 44 45 All target responses shown here are with background subtraction. Comparing the responses for different targets it can be observed that using a log versus linear scale yields a straight line response from the nonmagnetic copper rings while the response from ferrous targets like the steel sphere and cylinder are nonlinear on a log versus linear scale. However, when the ferrous targets are presented using a log versus log scale, an approximately linear response results. The performance from our sensor can be improved by using integration techniques to increase the signal to noise ratio especially in the late time when the signal strength is very weak. Also using a higher resolution A/D converter, minimizing noise from different sources including digitization noise, motion induced noise, location uncertainties, etc. can prove to be beneficial in enhancing overall sensor performance. 5.3 Summary In this chapter the components used to construct our EMI time domain sensor were presented. Data collected with our sensor for various targets (ferrous and nonferrous) was compared with the same collected with the commercial sensor (the EM63). In the next chapter sensitivity plots for an arbitrary coil configuration are defined and a software tool is developed as a means of presenting sensitivity data in a graphical form. Sensitivity plots are particularly useful for developing a qualitative understanding of overall EMI system performance. CHAPTER 6 SENSITIVITY AND STREAMLINE PLOTS FOR EMI SENSOR This chapter introduces efforts made to develop and test a software tool for evaluating electromagnetic transmitter coil designs and configurations used in the geophysical detection of unexploded ordnance. The effort is directed towards developing a better understanding of the capabilities and limitations of Wide-Area Transmitter and Multi-Transmitter (WAT-MT) sensor systems. 6.1 Introduction In order to evaluate EMI system performance in terms of sensitivity to different targets, static sensitivity maps can be used. Sensitivity maps provide a large amount of information about how effective a proposed coil layout can be [17]. Static sensitivity maps are a graphical representation that maps sensor response to a standardized infinitesimal object in a static field. These plots take in full account of both the transmitter and receiver coil shapes with no explicit consideration of target characteristics and are compact and are easy to understand. 6.1.1 Equations for static sensitivity The static sensitivity ?()P of a set of coils is defined as the ratio of receiver coil flux linkage change to the flux linkages of the transmitter coil, per unit volume of the target in the limit of vanishing target size and is given as 46 TR TR TT o ii HH L P ? = ? ? 2 3 )( (6.1) where ? (P) is static sensitivity at any space point P, H R is the magnetic field at point P due to the receiver coil currents and H T is the magnetic field at the same point due to the transmitter coil. Note that according to (6.1) the sensitivity is independent of transmitter and receiver currents because H R and H T are proportional to i R and i T , respectively. Therefore, sensitivity plots can be generated for a particular configuration of coils without reference to specific current values. In short, the aim of this graphical representation is to show the relative output change for an infinitesimal target when it is placed at different positions relative to the transmitter and receiver coils so that the uniformity and dynamic range of sensitivity can be assessed without reference to specific targets. The full derivation of (6.1) can be studied in [17]. Also we know that by Bio Savart?s law the field at a point (P) in Figure 6.1 due to a straight current carrying conductor is given as ? ?? ?? a I H )1cos2(cos 4 ?= (6.2) Where, a ? = a l x a ? ,? is perpendicular distance from conductor to point P and I is the current flowing in the conductor [18]. A computer program written using the latest release of Matlab (2006) has been developed using (6.1) and (6.2) for arbitrary configurations of transmitter and receiver coils to generate sensitivity maps. Appendix A provides more details about the software tool. The software allows the user to obtain sensitivity as well as streamline plots. Streamlines are useful when one is interested in visualizing the direction of the magnetic field due to the transmitter or receiver coil [19]. 47 P Z ?2 ?1 a l H (into the page) ? a a ? Figure 6.1 Conductor is shown as dark heavy line and the magnetic field due to it at P is into the page [18]. 6.2 Sensitivity and streamline plots for different coil configuration 6.2.1 Simple configuration of two square coils The first configuration considered is shown in Figure 6.2 and has been examined previously in [17]. It consists of two square symmetrically positioned coils with dimensions 984 mm (transmit coil) and 890 mm (receive coil). Figure 6.3a and 6.3b presents a contour plot of sensitivity on a plane parallel to the z-y plane extending from ?60cm50 cm) it decreases rapidly. Also at the center the sensitivity is less than it is just above the conductors. But in Figure 6.4b at x=60 cm (further away from the plane of the coils) the sensitivity peaks at the center (y=0 cm) and then decreases monotonically on either side. 49 Y Z Contour Plot for X= 0 0. 0 24 909 0 . 0 4 5 2 2 3 0 . 06 5 5 38 0 . 0 8 5 8 5 2 0 . 1 0 6 1 7 0. 20 77 4 0 . 3 0 9 3 1 0 .8 5 7 8 0. 61 40 3 -80 -60 -40 -20 0 20 40 60 80 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 Figure 6.3a Sensitivity contours for coil configuration of Figure 6.2 computed over the plane x=0 cm, -80cm