CROSS REFERENCE TO RELATED APPLICATIONS

This application claims priority to U.S Provisional Application No. 61/739,282 filed on December 19, 2012 and U.S Provisional Application No. 61/782,161 filed on March 14, 2013. The entire teachings of the above applications are incorporated herein by reference.

BACKGROUND

Technical Field

This application relates to chemical analysis and more particularly to systems and methods that use nuclear quadrupole resonance.

Background Information

It is known that an atom with more than one unpaired nuclear particle (protons or neutrons) will have a charge distribution which results in an electric quadrupole moment. Allowed nuclear energy levels are shifted unequally due to the interaction of the nuclear charge with an electric field gradient supplied by the non-uniform distribution electron density (e.g. from bonding electrons) and/or surrounding ions. This so-called Nuclear Quadrupole Resonance (NQR) effect results when transitions are induced between these nuclear levels by an externally applied radio frequency (RF) field. This electromagnetic field thus induces a magnetic resonance, unique to each material, without using a magnet. A typically NQR detection system consists of a radio frequency (RF) power source, an emitter to produce the electromagnetic excitation field, and a detector circuit which monitors for a RF NQR response coming from the object being analyzed. NQR has a number of practical uses, such as the detection of land mines, or of narcotics or explosives concealed in luggage, or remote monitoring of fluid levels such as in oil wells.

It is also known that irradiation of a sample containing spin-1 nuclei (such as sodium nitrite) by two of three characteristic NQR frequencies can result in free induction decay (FID) and echo signals at a third NQR frequency.

Systems that use radio frequency signals to induce the NQR effect to detect explosive materials are known in the art; see our co-pending U.S. Patent Application Serial No. 12/628,824 filed September 27, 2012 as one example, and U.S. Patent Application Serial No. 13/901,765 filed on May 24, 2013 as another example (the entire contents of each of which are hereby incorporated by reference).

SUMMARY

One of the problems with known existing explosives detection systems is the need for a stable reference to effect high dynamic range cancellation. An approach is to use a reference signal that is the system response to an empty cavity portal. The response of the system with the explosive and its container in the cavity is then subtracted from the reference. However, the system response to a container (or a human being) with no explosive is not quite equivalent to that of a completely empty cavity. A more accurate procedure entails using the empty container in the cavity as the reference mode.

However, since the "container" in a fielded system is often a human being, it is almost impossible to use the empty container as a reference. This situation is compounded by the fact that the system response for each human being is different.

In pertinent aspects an NQR detection system according to the teachings herein detects Rabi transitions in a material being analyzed. The methodology employs a transmitted waveform with two power state illuminations, such as a high and low power at one or more frequencies, which are combined to cancel the incident field. The waveform utilized is preferably a continuous linear frequency modulated chirp signal that provides frequency agility, facilitating the use of matched filter detection. In one embodiment, an enclosed chamber or cavity is used as a detection portal. In specific embodiments, a first measurement is taken using a relatively low radio frequency power level. A second measurement is then taken at a relatively high power level. The detection system determines a measurement such as by calculating S21 parameters for each measurement. The detected responses to the high and low power emissions should be identical as long as the transfer function of the cavity is not nonlinear.

More particularly, a correction for non-linearities in the response may take one of several different forms. The techniques described herein are based on observing that the nuclear quadrupole resonances of explosives involve continuous Rabi transitions are nonlinear processes since stimulated emission occurs.

In a first approach, first order corrections to nonlinearities use one or more empty cavity responses. More specifically, the first method relies on S21 measurements being equal for the high and low power measurements. If the transfer function of the system, excluding the NQR resonances, is nonlinear to greater than, say, one part per billion or 90db dynamic range, then a correction of these non-linearities should be implemented.

In this first method, correction is equal may be determined by first detecting an empty cavity response, such as the S21 measurement, for a high power case, minus an S21 measurement for the low power case using an empty cavity. This difference is used as a correction factor. Measurments are then taken with a human occupied cavity, such as S21 for the high power case minus S21 for the low power case. The above correction factor is then applied to the occupied measurement. In a second method, no

measurements from an empty cavity are needed. Instead, this method uses three emitted signals at frequency ranges of interest, such as three contiguous bands. The NQR resonance of interest is known to exist in a first one of the bands while the second and third bands are known to exhibit no NQR signals of interest. Here, the method determines an S21 high minus S21 low difference for the second and third (non-occupied bands). If the correction may be assumed to be a linear function of frequency over the three bands (highly likely because of the small frequency spread) then the change in the difference between bands two and three should be the same as the changes between bands one and two. This enables an estimate for the correction of responses in the first frequency band, where the resonance is located, by subtracting the observed difference between the measurments at the second and third frequencies.

In at third implementation, two NQR resonances may be simultaneously excited by two radio frequency chirps, which in a preferred arrangement, are centered at, or at least include, two NQR frequencies of interest, and which are expected to elicit a response at a third NQR frequency such that the result is an NQR response signal at a third frequency. Because this approach avoids generation of an excitation waveform at the third frequency, no cancellation or other adjustment of the response detected at the third frequency is required.

BRIEF DESCRIPTION OF THE DRAWINGS

The description below refers to the accompanying drawings, of which:

Fig. 1 is is a high-level block diagram of an NQR detection system.

Figs. 2(a), 2(b) and 2(c) are an example cavity portal that may be used with the detection system described herein.

Figs. 2(d) and 2(e) show another arrangement where a conducting half space layer placed on a floor is used as the portal.

Fig. 3 is a more detailed block diagram of components of the received signal processing.

Fig. 4 is a flow diagram of process steps that may be performed by a controller to operate the system in a calibration mode and in an operating mode.

Fig. 5 is a flow diagram of a detection process where low power and high power measured responses are compared.

Fig. 6 is a system architecture using seven amplifiers and seven digital to analog converters (DACs).

Fig. 7 is architecture using a single DAC and four amplifiers.

Fig. 8 is an architecture using four DACs and four amplifiers.

Fig. 9 is a flow diagram of a first empty cavity method for correcting non- linearities. Fig. 10 is a flow diagram of a second analytic method.

Fig. 11 is a flow diagram of a third method using two simultaneous frequency band emissions with detection at a third frequency band.

DETAILED DESCRIPTION OF AN ILLUSTRATIVE

EMBODIMENT

System Overview

Fig. 1 is a high-level diagram of the components of a detection system according to the teachings herein. An initiate and control function 102 may be implemented by a suitable programmable processor. This function controls generation 104 of a suitable transmit waveform which is then produced as an output signal 106. The output signal 106 is amplified 108 and filtered 110. This signal is then sampled 112 prior to it being radiated 114. The radiated signal enters a portal containing the human being of interest. Responses from the portal 114 are then returned to the sampler 112. Both a reference signal 120 and received signal 122 returned from the portal are then fed into signal processing 124. The outputs from signal processing 124 are then interpreted and reported 126.

The detection system of Fig. 1 is typically architected with a combination of digital and RF components. Initialization 102 begins with the transmission of a series of linear chirp waveforms associated with NQR resonance lines of interest. These signals are then amplified 108, filtered 110, and sent to the sensing portal producing a low-power magnetic field. This magnetic field generated in response to the linear chirp is then incident on whatever is contained in the chamber, causing coherent NQR emissions from such contents of the chamber. The response from the chamber contains the transmitted energy, reflected energy, and the NQR signal. To eliminate the transmit signal, samples 112 are taken and used response signal input to a cancellation algorithm. The response signal is processed by means of a cancellation and matched filter algorithm (via signal portal 124) before being reported 126 as either having an explosive or not. The inspiration for creating an NQR based detection system with detection times of less than < 5.0 seconds stems from the application of continuous wave, chirped signal techniques as typically used in radar applications rather than the pulsed technology which has dominated previous efforts to detect the weak NQR signals. It is well known that when a two (2) state atom is illuminated continuously by an electromagnetic field at resonance, the atom oscillates between state 1 and state 2, alternately absorbing energy from the incident field and emitting coherent energy via stimulated emission as a result of the chirp signal. This process is an attractive way to increase the NQR signal to higher levels.

A transmitter may be operated continuously rather than pulsed if the strong transmitted signal can be separated from the weak signal of interest. A combination of cancellation of the transmitted chirp signal at the receiver and use of directional couplers or circulators are sufficient.

To develop enough cancellation to deal with NQR signals at levels of less than - 70 dBm buried in an incident field of 40 dBm, a combination of directional couplers and a two (2) channel base band digital receiver (120, 122) is utilized in this embodiment. The cancellation methodology employs a chirp waveform with alternating two (2) power state illuminations which are combined to cancel the incident field. Since the frequency range of interest covered is 330 KHz to 5 MHz, a stable, wideband Faraday chamber (which we also call a "cavity" or a "portal" herein) to detect the explosives of interest while maintaining a low return loss (>35 dB) over the bandwidth is important for the cancellation methodology to work well.

The waveform utilized is a continuous linear frequency modulated (FM) chirp to provide frequency agility and facilitate the use of a matched filter (124) for the NQR response to the chirp. The transmitter should generate fields that are in the 10 W/m range with low leakage beyond the cavity chamber.

Figs. 2(a), 2(b) and 2(c) are an isometric, front and side view of a typical portal cavity 300 (also referred to as the "chamber" herein). In a practical implementation, one or more conductive surfaces 302 are arranged to define a space that is to be monitored such as for access control. This cavity type portal uses a generally rectangular space 300 defined by four conductive walls 302-1, 302-2, 302-3, 302-4. Two or more wire loops 306-1, 306-2 are disposed within the space, typically adjacent selected ones of the conductive surfaces 302. The wire loops 306 are each individually electrically terminated through a resistance 310 to the respective conductive wall(s) in this arrangement. A coaxial cable connector 308-1, 308-2 provides connection to the radio frequency (RF) transmitter and receiver. The conductive walls 302 define the space within which a uniform electromagnetic field can be maintained by the wire loop radiators while at the same time protecting the space from outside disturbances.

Other arrangements are possible for the wire loops. For example, they can be implemented as a balanced transmission line driving two wire segments through a balun with the two segments having a resistance disposed at their mid-point.

In another arrangement, the space to be monitored is defined as a conductive half- space such as defined by a metal surface embedded in a floor. In this other arrangement shown in Figs. 2(d) and 2(e), the portal space to be monitored is defined as a conductive half-space 410. A system of wire loops 410 provides excitation to such a conductive half space 400, defined by a metal surface 402 embedded in a floor, as shown. The half space 400 can thus be a corridor or large open public area. In the illustration of Figs. 4(d) and 4(e), the loops 410 are individually fed by coax feeds 408, and terminated by resistors 412. The coax feeds 408 may have alternating polarities, as shown. The excitation loop(s) layer and the conducting half space layer can comprise a composite flexible carpet, in one example. Other arrangements for the space to be monitored are possible.

The portal thus serves two functions— it is both the signal transmission device and the signal sensing device. A low-power magnetic field is generated within the portal cavity and the reflection is received. Some key performance parameters of the portal cavity design included the uniformity of the magnetic field while maintaining acceptable field strength within the cavity, minimal magnetic and electric fields external to the cavity and a cavity sized so that a handicap person can pass through without being impeded.

Simulated results of the detection cavity such as in Fig. 2(a)-2(c) assumed 10.0 Watts of power is applied at a frequency of 5.0 MHz. The power level in the cavity is significantly less than that of the OSHA standard for human safety levels, 100.0 W/m2 for 6.0 minutes.

The uniformity of the magnetic field within the cavity should allow for a body and/or material of interest to be uniformly illuminated within the magnetic field. The low-level magnetic field and electric field external to the cavity, ensure external noise effects are at a minimum where the magnetic field external to the cavity diminishes more rapidly than that of the electric field external to the cavity. To accommodate the majority of travelers and handicap individuals, we prefer a cavity with an opening or walkthrough portion dimensioned at 7.0 feet in height by 3.0 feet in width and 4.0 feet in depth.

Fig. 3 contains a more detailed flow diagram of the signal processing 124 components of the system of Fig. 1. These include functions provided by the sample signals function 112, the receive reference signal 120, received signal 122, and the process signal 124. Here a pair of directional couplers 350-R, 350-S feed a

corresponding pair of analog-to-digital converters 360-R, 360-S and digital down converters 370-R, 370-S. The signal chains provide a reference signal output (REF) and received signal (SIG) output. A signal measurement block 380 then compares the signal (SIG) to the reference (REF) and provides an initial output. According to the teachings herein, the system is operated at a high power level, a low power level, with both a sine and cosine of the illuminating chirp signal, with the cavity empty and occupied, and with a number of different frequency chirps (covering f 1, f2, f3, ..., fN). These outputs are then fed to an end point decomposition block 390 and matched filter 395 prior to a detection process 398.

The receive processing is described at a high level in Fig. 4. In general, with an cavity empty (step 421), the sequences of chirp signals are applied (step 422) to determine a calibration. A first chirp is emitted into an empty cavity at a relatively low power level with sine phase to the chirp (EN _S i _n ). Next, a high power sine signal is applied to the empty cavity and the response is measured (EH _sin ). Next, a relatively low power level cosine signal is applied to the empty cavity and the response (EL _cos ) is stored. The same is done for a high power cosine (EH _cos ). These the responses to these chirps are then detected for the frequencies (step 423) of interest and stored (step 424) as empty cavity references. An operating mode is then enabled (step 450). The cavity becomes occupied such as with a human being. A set of measurements is taken at low and high power levels each for both the sine and cosine chirp phases at each frequency of interest (step 452). Next, an end point decomposition process is applied (step 454) to both the sine and cosine responses. A difference is taken between the the occupied and unoccupied responses from both the high power and low power responses for each of the sine and cosine chirps. A non-linearity correction is then made using any of the techniques discussed below. The sine and cosine responses are then applied to a matched filter (step 458). The matched filter contains an ideal expected response for each of the sine and cosine chirps. The results of the matched filter output are then subjected to a magnitude operation such as may be determined by squaring the sine and cosine responses (step 460) and taking this sum. This process is then repeated for the each frequency of interest (step 462).

Fig. 5 shows the process being repeated for each frequency of interest 502.

More generally, one or more measurements are taken at a frequency of interest (per the process of Fig. 4). These measurements may be taken for high and low power, sine and cosine chirps, and both the empty and occupied cavity. The measurements are then corrected in step 504.

From the detected responses, a conclusion is reached at 506 whether or not the explosive material of interest is not present. This can be the case, in one embodiment, if the high and low power level responses, as corrected per one of the methods described in more detail below, are approximately equal. However if the corrected responses are not equal at step 510 such when the high power response contains more energy than the low power response, it is possible that a resonance of interest may be present at that frequency. At this point in step 510 it may be concluded that there is a positive result. Alternatively, a still higher power may be presented at step 512 to the cavity and subjected to the same sine and cosine chirped end point decomposition and matched filter processing. The magnitude of that still higher power measurement may confirm in step 514 that the resonance of interest is present at the designated frequency. One of the problems noticed with other systems is the measurement to

measurement variability of the results. Analysis has shown that the results depend critically on the frequency of the resonance relative to the start frequency of the chirp. In fact the result depends upon the term cosine (arg), where (arg) is proportional to the frequency differences. By using sine and cosine chirps sequentially and adding the sequential outputs in quadrature, the analytic signal is better captured.

One embodiment of the system herein uses a single-DAC, 4- Amplifier system architecture for generating sixteen (16) nuclear quadrupole resonances, as shown in Fig. 7. Here a sweep consists of four (4) runs: 1) Sine, 2) Cosine, 3) High Power and 4) Low Power. Four (4) sweeps and thus a total of sixteen (16) runs are necessary to cover the sixteen (16) nuclear quadrupole resonant frequencies of interest.

In the Fig. 7 arrangement, a single waveform containing four (4) chirps of varying frequencies is output through the transmit port of the transceiver 702 and split via a 4- way multiplexer or splitter 704. The four (4) signals are then simultaneously sent through four (4) 4-switch amplifier/filter bank blocks 706-1, 706-2, 706-3, 706-4 in parallel so that each of the paths handles one (1) of the four (4) chirps of varying frequency. The four (4) filtered and amplified signals are then 4-way combined or multiplexed 708 and sent through the remaining RF chain, where the two (2) final inputs to the transceiver are the REF and SIG signals. As explained above, the REF signal is a reference signal sampled from the system in order to account for any anomalies the system may incur per run. The SIG signal is the signal sampled from the shielded portal.

An alternative to the 1-DAC, 4- Amplifier system architecture is the 4-DAC, 4- Amplifier system architecture shown in Fig. 8. In this embodiment, a single transceiver 802 with four (4) parallel transmit ports each handle a single waveform containing only one (1) chirp. This transceiver architecture eliminates the need for a 4-way multiplexer or splitter at the input of the four (4) amplifier/filter bank blocks and simplifies the waveform output from the transmit port. The remainder of the system architecture from the amplifier/filter bank blocks 806-1, 806-2, 806-3, 806-4 on is identical to that of the 1- DAC, 4- Amplifier system architecture of Fig. 7. The architectural implementation for the case where five (5) materials are of interest and only one (1) resonance per material can be sampled at a time is the 7-DAC, 7-Amplifier System Architecture shown in Fig. 6. In this embodiment, a transceiver 602 with seven (7) parallel transmit ports and corresponding seven (7) amplifiers (606-1, ... , 606-7) are necessary due to frequency band breaks where each transmit port handles a single waveform containing only one (1) chirp. This transceiver architecture gets rid of the need for a splitter at the input of the amplifier and simplifies the waveform output from the transmit port so that only a single signal waveform is needed per port. The need for filter banks at the inputs and outputs of the amplifiers is unnecessary in this architecture since the reactive combiner/multiplexer at the output of the amplifiers serves as a filtering component for each signal in addition to serving as a low-loss combiner. The combined signals at the output of the reactive combiner/multiplexer 607 (which may include a set of duplexers and triplexers) are sent through the remaining RF chain consisting of two (2) couplers 608, 610 and a shielded portal, where eventually the two (2) final inputs to the transceiver are the REF and SIG signals which come from the two (2) coupled ports of the two (2) couplers. The REF signal is a reference signal sampled from the system in order to account for any anomalies the system may incur per run. The SIG signal is the signal sampled from the shielded portal.

While the above processing architectures of Figs. 7 and 8 are two of the possible embodiments, any of the Stimulated Emission Enhanced Quadrupole Resonance (SEE- QR) based hardware and digital signal processing (DSP) architectures described in the referenced patent applications can be used to implement any of the non-linearity methods described below.

Non-Linearity Correction / Elimination Methods Using Empty Cavity (Method 1 )

In this method, measurements in state 503 involve measurement of parameter S21 at a high and low power. The measurements are made, in a preferred embodiment, with an empty cavity using two successive chirp signals (sine and cosine) covering the frequency range(s) of interest. The assumption here is that the nonlinearities are caused by system components outside the cavity 300 to first order. Thus the corrections made for the empty cavity are assumed to be valid when the cavity is not empty i.e. when a human and/or a human containing explosive material is inside the cavity.

It is assumed that periodic calibration of the system using the reference (REF) and signal (SIG) processing with an empty cavity (e.g., steps 421, 422, 423, 424 of Fig. 4) are performed to maintain accuracy.

In short, this first method involves:

The correction in step 504 may be determined by first taking S21 for the high power case minus S21 for the low power case using an empty cavity.

The final correction result is S21 for the high power case minus S21 for the low power case using a human occupied cavity, with the above correction factor applied.

Fig. 9 is a more detailed flow diagram of a first approach to correcting for nonlinearities that makes use of initial measurements with an empty cavity. In this embodiment, at a first state 901 the cavity 300 is empty. In a next step 902 a relatively high power measurement is taken at a first frequency of interest, f 1. As explained previously, this measurement is preferably taken by emitting a chirp signal preferably centered at the frequency of interest, fl, or least including the frequency of interest, fl. Depending on the detection processing employed, there may be a pair of chirps emitted at the designated power level, such as a sine chirp and cosine chirp. In the event of such sine and cosine processing, the resulting two responses are combined (e.g., the square of each is taken and then summed to arrive at the resulting high power response). In step 903 a low power measurement is taken at the same frequency fl; if necessary this is again done with both the sine and cosine phase and then squaring and summing the results.

In step 904 a correction factor is then determined for measurements taken at frequency f 1. The correction factor is given by the resulting empty cavity measurement at the high power minus the resulting empty cavity measurement at the low power.

Later on, the cavity becomes occupied in state 910. In a subsequent step 911, a high power measurement is taken of the occupied cavity with by transmitting a chirp including frequency f 1 (and for both the sine and cosine phase as may be needed). In a next step 912 a low power measurement is taken also with a chirp including the frequency f 1 and the sine and cosine phases as needed.

In step 913, the resulting response is determined as the difference between the high and low power measurement and by further subtracting the correction factor as was determined in step 904.

Analytic Extrapolation (Method 2)

In an example of this method, measurement 503 involves using three chirps covering three frequency ranges of interest, such as three preferably contiguous 40 KHz bands, the three contiguous bands including respective frequencies fl, f2 and f3. The expected NQR resonance is known to exist in the first band (the band that includes f 1) while the second and third bands (those that include frequencies f2 and f3) are known to harbor no NQR signals. The idea behind the correction implemented by step 504 in this implementation is then to use the S21 high minus S21 low differences for the second and third bands. If the correction can be correctly assiumed to be a linear function of frequency over the three bands (highly likely because of the small frequency spread) then the change in the differences between bands two and three should be the same as the changes between bands one and two. This result enables an estimate for the correction in the first band, where the resonance is located; the estimate is made assuming the correction is linear; that is, a subtraction may be sufficient.

Fig. 10 is a more detailed flow diagram of the processing performed in an analytic extrapolation embodiment that does not require initial empty cavity measurements. Here at first state 1010 the cavity 300 is already occupied. In state 1011, a high power measurement is taken at a first frequency fl, such as with a transmitted chirp that includes frequency fl. In state 1012 a low power measurement is taken with a chirp that includes that same frequency fl. As before these measurement may be taken with both sine and cosine versions of a chirp signal that includes the frequency of interest fl. Next, high power measurement 1020 and low power measurement 1021 are taken at a second frequency f2. As before, these may be taken with chirp signals which may include both sine and cosine chirped signals that includes the frequency f2.

In steps 1030 and 1031, high and low power measurements are taken at a third frequency f3, with the chirp signal that includes a third frequency of interest, f3 (and again for both sine and cosine cases if desired).

The correction at frequency fl is then determined in state 1040. In this state 1040, a difference is determined between the high and low power measurements that were taken at frequency f2 (e.g., in steps 1020 and 1021), and a difference is also determined between high and low power frequency measurements that were taken at f3 (e.g., in steps 1030 and 1031). The difference between these two measurements is then used as the correction factor for measurements taken at frequency f 1.

Finally, in state 1041 the response for fl is determined by taking the difference between the high and low power measurements at frequency f 1 and then applying the correction factor as was determined in step 1040.

Emit Two Simultaneous Chirps, Detect a Third (Method 3 )

In this arrangement, step 503 generates two chirps simultaneously to excite two resonances in the material of interest; in turn the material of interest emits a signal at a third frequency. Because there is no incident wave at the third frequency, detection of the response does not require non-linearity cancellation processing at the third frequency.

The third frequency resonance response is expected to have a signal strength of about

50% that of the single resonance.

The measurement step 503 thus uses excitation by a chirp which includes a first frequency f 1 added to the excitation by a chirp which includes a second frequency f2.

This results in the a induction decaying response at a third frequency, f3.

In the arrangement preferred here, two chirps (increasing or decreasing frequency swept signals) are thus used to simultaneously excite two resonances. The methodology is otherwise basically the same as the single resonance chirp case above where the continuous input waveform in step 503 induces Rabi transitions leading to alternating absorption and stimulated emission at the single resonance frequency. In the case of the two excited resonances, the two stimulating chirps cause stimulated emission of the third frequency instead of the free induction decay that would otherwise result.

Fig. 11 shows a detailed set of steps for this embodiment where corrections are again provided without need for empty cavity measurements. This embodiment relies on underlying NQR physics when two signals simultaneously emitted at two different frequencies provoke a response of interest at a third frequency.

In particular with an occupied cavity in state 1110, a state 1111 is entered in which chirp signal(s) that include at least both f 1 and f2 are emitted simultaneously. These chirp signals may be repeated for both the sine and then cosine case as may be desired. In step 1112, the response to this simultaneous two-frequency emission is detected at a third frequency f3.

In state 1113, no further correction is needed in this instance. However if sine and cosine chirps were both emitted, the response will need to be adjusted (such as by squaring and summing the sine and cosine responses). Finally the adjusted response in state 1114 is given as the signal detected at frequency f3 with no further correction needed.

The fact that no cancellation is required to detect the NQR signal means that the 90 odd db of dynamic range used to perform the operation can be used to detect smaller signals at a distance. For example using the radar range equation it can be shown that at a range of 100 feet the NQR signal to noise will be equal to the signal to noise we are observing in the portal if the power amplifier output is 50 watts.

This three frequency method otherwise basically operates in the same was as for the single resonance case described above. However, this embodiment may be simpler than the other methodsdescribed. For example, it may not be necessary to employ "start- stop" processing as used in the single resonance case.

In one example where the three frequency method may be used, Nitrogen 14 is the material of interest. N14 has a spin 1 nucleus with three possible transition frequencies. When N14 is embedded in explosives the frequencies shift; in many explosives many more than three resonances are present. In another example where this method can be used, sodium nitrite has three resonances.

It is now understood that if the response to the two emitted chirp signals that include a first and second frequency is a chirp centered around (or at least includes) a third frequency, coherent signal processing (such as by using a matched filter for the expected chirp at the third frequency) can thus be used to improve receiver signal to noise ratio (SNR).

What is claimed is: