Impulse radar device and method

Abstract

A miniaturized impulse radar is provided. The radar uses a slight frequency difference between a local oscillator and an RF signal to generate high quality A- Scans using an inexpensive microcontroller's analog-to-digital converter. A DMA circuit is used to write analog-to-digital converter outputs to a buffer memory. The radar has applications in, for instance, ground penetrating radar for detecting buried utilities, automotive collision avoidance, autonomous robot navigation, and machine vision in general.

Description

IN THE UNITED STATES RECEIVING OFFICE AN INTERNATIONAL PATENT APPLICATION FOR AN IMPULSE RADAR DEVICE AND METHOD

[0001] I. BACKGROUND OF THE INVENTION

[0002] A. FIELD OF INVENTION

[0003] The invention generally relates to the field of impulse radar detection circuits.

[0004] B. DESCRIPTION OF THE RELATED ART

[0005] Since the origins of modem radar technology in the early 20thcentury, radar innovation has been largely driven by military applications. The capability to detect the presence of targets and determine their size, speed, and direction has been viewed as far more important than the cost or size of radar systems. With little downward pressure on costs, many advanced radar systems tend to be thousands to millions of dollars, depending on the application.

[0006] Impulse radar has been known at least since 1960 when John C. Cook published Proposed Monocycle-Pulse Very-High-Frequency Radar for Air-Borne Ice and Snow Measurement, describing an impulse radar for use in measuring the thickness of polar ice caps. This technology was soon adapted to military applications, searching for underground tunnels in Vietnam, and later developed to detect underground geological structures and buried utilities. See, for instance, AM Introduction to Impulse Radar by Merrill I. Skolnick published in NRL Memorandum Report 6755, dated November 20. 1990.

[0007] Modem commercial radars can generally be characterized as one of two broad classes. One class is the low-cost integrated circuit (IC) radar that outputs data through standard bus architectures such as SPI, I2C, and so on. Though low-cost, IC radars operate at high frequencies around 70 to 80 GHz. Consequently this ty pe of radar is operable over very' short ranges in air on the order of a few meters, and no more than a few centimeters in soil.IC radars can be used in connection with directional antennas, and find application in devices such as vehicle collision avoidance.

[0008] Another class of modem commercial radars operate below 10 GHz. They typically operate either in the frequency domain, or in the time domain as impulse radars using asampling technology to achieve high effective bandwidths. Regarding sub- 10 GHz frequency domain radars, some are relatively costly system-on-a-chip (SoC) devices using a direct digital synthesis (DDS) technology to generate precise analog waveforms that can be time- varied without phase locking loops. These SoC radars can be described by a center frequency fcand bandwidth (Eq. 1)where the fractional Bandwidth (FBW) (Eq. 2) is relatively small, permitting theuse of high-Q resonant antennas. These radars suffer from lower range resolution, which is inversely proportional to pulse width; the longer the pulse, the lower the range resolution. Generally, the cost of an SoC radar with an analog-to-digital converter (ADC) operating in the gigasample per second (GSPS) range with built-in DDS that can produce nanosecond pulses is prohibitively expensive for many commercial applications.

[0009] Other sub- 10 GHz frequency domain radars are made less costly because short pulse width packets are not required in a stepped frequency continuous wave (SFCW) radar, which only measures amplitude and phase. Such devices owe their bandwidth and range resolution to measuring the amplitude and phase returned at each frequency as they step through a Fast Fourier Transform (FFT). A typical architecture uses a frequency synthesizer chip, a fractional-N phase locking loop (PLL) with integrated broadband voltage controlled oscillators (VCO) for phase locking. The fractional-N PLL provides fine frequency resolution, and flexibility in generating output frequencies without a costly DDS in a SoC architecture. While fractional-N PPLs are relatively cheap, they require an ADC that can sample an intermediate frequency at about 200 megasample per second (MSPS), which adds significant cost.

[0010] In contrast to frequency domain radars, impulse radars work by sending out a series of pulses and sampling the corresponding reflected signals after a fixed number of picoseconds. The data can be used to reconstruct the reflected signal as one pulse without requiring the more costly equipment used in frequency domain radars. The tradeoff tends to be an extended data acquisition time. A microcontroller's onboard ADC might be used to quantize analog signals, but the ADC is controlled by the microcontroller’s CPU. There are several problems with this architecture that make it untenable for impulse radar applications.

[0011] First, the CPU processing time is significant, which vastly reduces the ADC’s sampling rate. Architectures that use the CPU to control a microcontroller’s ADC and other peripherals can only achieve bandwidths in the 10’s of KHz. Another problem is that sampling becomes unpredictable because the time lag between a pulse’s arrival and the CPU’s response can vaty on the order of milliseconds, while the signal must be captured in less than a microsecond. A third problem is that the CPU’s response time is slow relative to the timescale of the event being measured. The CPU must process an API call, set up the ADC, and coordinate writing the quantized data to memory. The net result of these three shortcomings, in the context of impulse radars, is that sampling is too slow to keep up with pulse arrivals. The solution to this problem has been to employ an off-board ADC, rather than the microcontroller’s on-board ADC, which may also require additional processing such as a field programmable gated array (FPGA) or complex programmable logic device (CPLD).

[0012] Another problem with using a microcontroller’s on-board ADC in an impulse radar is that it is not always on. It must be precisely synchronized with pulse arrival time, which can be a complex problem in an impulse radar where pulse widths can be on the order of nanoseconds or even picoseconds. The high degree of required precision suggests the need for more sophisticated components.

[0013] What is missing is an impulse radar architecture that is sufficiently fast to provide high bandwidth measurements using low bandwidth hardware. Some embodiments of the present invention may provide one or more benefits or advantages over the prior art.

[0014] II. SUMMARY OF THE INVENTION

[0015] Embodiments comprise multipurpose miniaturized impulse radars. Such radars include circuitry for producing an RF pulse and a LO pulse that are substantially the same but differ enough in frequency to permit a period-to-period phase shift. Moreover, the periods of the RF and LO pulse can be arbitrarily large while the pulse width is very' short. The short pulse width provides the desired level of range resolution while the arbitrarily large period permits the use of low-cost ADCs to digitize the RF return. Each period of corresponding RF and LO pulses phase shift by a constant amount, the magnitude of which is selected to ensure that each RF return pulse is sampled at enough points to satisfy Nyquist’s Theorem. The number of pulses per frame, pulse power, and frequency content of the pulses are selected to provide the desired level of range through a given medium. Return RF pulses are thussampled and recorded to a buffer memory through a DMA circuit, and then transferred to an off-board computer for image analysis. Some embodiments are GPS enabled and are capable of geolocating the RF signals.

[0016] Other benefits and advantages will become apparent to those skilled in the art to which it pertains upon reading and understanding of the following detailed specification.

[0017] III. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The invention may take physical form in certain parts and arrangement of parts, embodiments of which will be described in detail in this specification and illustrated in the accompanying drawings which form a part hereof, wherein like reference numerals indicate like structure, and wherein:

[0019] FIG. 1 is a stacked plot of the higher frequency and lower frequency waveforms;

[0020] FIG. 2 is a stacked plot showing full frames of higher frequency and lower frequency pulses;

[0021] FIG. 3A is a plot of a single RF return pulse in isolation;

[0022] FIG. 3B is a plot of a reconstructed RF return pulse

[0023] FIG. 4A is a plot of three successive pulses in a pulse train illustrating the phase shift in each successive period;

[0024] FIG. 4B is a plot showing the effect of phase shift over many pulses;

[0025] FIG. 5 is a pair of plots showing the pulse train before and after amplification;

[0026] FIG. 6 is a plot of a pulse after applying a low pass filter;

[0027] FIG. 7 is a plot of a reconstructed A-Scan;

[0028] FIG. 8 is a circuit diagram of an impulse generator according an embodiment;

[0029] FIG. 9 is a circuit diagram of a sampling mixer according to an embodiment;

[0030] FIG. 10A is a perspective view of an embodiment with the top section of the housing removed exposing the printed circuit board;

[0031] FIG. 10B is a perspective view of an embodiment with the housing assembled;

[0032] FIG. 11 is a drawing of representative A-Scan data showing reflections from a plurality of target at different ranges; and

[0033] FIG. 12 is a plot of representative B-Scan data.

[0034] IV. DETAILED DESCRIPTION OF THE INVENTION

[0035] Definitions

[0036] As used herein the terms “embodiment”, “embodiments”, “some embodiments”, “other embodiments” and so on are not exclusive of one another. Except where there is an explicit statement to the contrary, all descriptions of the features and elements of the various embodiments disclosed herein may be combined in all operable combinations thereof.

[0037] Language used herein to describe process steps may include words such as “then” which suggest an order of operations; however, one skilled in the art will appreciate that the use of such terms is often a matter of convenience and does not necessarily limit the process being described to a particular order of steps.

[0038] Conjunctions and combinations of conjunctions (e.g. “and / or”) are used herein when reciting elements and characteristics of embodiments; however, unless specifically stated to the contrary or required by context, “and”, “or” and “and / or” are interchangeable and do not necessarily require every element of a list or only one element of a list to the exclusion of others.

[0039] Terms of degree, terms of approximation, and / or subjective terms may be used herein to describe certain features or elements of the invention. In each case sufficient disclosure is provided to inform the person having ordinary skill in the art in accordance with the written description requirement and the definiteness requirement of 35 U.S.C. 112.

[0040] Description

[0041] Contrary to the pervasive view in the electrical arts, microcontrollers and their relatively low-performance onboard analog-to-digital converter (ADC) units are capable of functioning in an impulse radar provided that an architecture according to the present invention is used. In part, embodiments accomplish this by limiting the CPU to performing operations that are timing-flexible. For instance, the CPU may be used to setup the impulse radar, but thereafter the CPU does not intervene in the function of the ADC or in writing quantized data to memory. Additionally, embodiments apply a timing technology of theinvention to spread sampling out over wide time intervals. In this arrangement, low- performance ADCs, like those onboard typical microcontrollers, are sufficient. At the low end of sampling rate, the only limitation in ADC selection is whether the use case can tolerate the added time required for data acquisition. At the high end, the limitations are equipment costs and size. Nonetheless, embodiments of the invention can include vety high performance ADCs without departing from the scope of the invention.

[0042] In embodiments of the invention, the CPU instructs a clock circuit to output two periodic square waves having slightly different pulse repetition frequencies. The square waves are each separately communicated to one of two impulse generators having identical architectures, converting the square waves to broadband pulses having a sufficiently narrow pulse width to provide acceptable range resolution as determined by a given application. In some embodiments both the higher and lower frequency pulses may have identical frequency content. However, this is not a requirement of the invention. What is important is that the lower frequency pulse is capable of serving as a local oscillator for triggering a sampling mixer to sample RF returns originating from the higher frequency pulse, and that it is also capable of phase shifting relative to the higher frequency pulse as described herein. Provided these conditions are met. the output of the second impulse generator is substantially the same as the output of the first impulse generator.

[0043] In one embodiment the frequency content of both pulses is between 500 MHz and 5 GHz. The content starts to taper off at 3 GHz and is very low as frequency increases to 5 GHz. According to some embodiments having this frequency content, the most important frequencies are approximately 2 GHz. Suitable frequencies are determined in part by their ability to transmit through a medium. For instance. 2 GHz has a useful penetration of damp soil due its extinction coefficient at that frequency. Therefore, this frequency can be used to detect targets such as buried utilities. Suitable frequencies are also determined in part by signal strength at the given frequency. A frequency with good penetration of a given medium may be unusable if its power is too low. Conversely, even a relatively high power signal may be unusable if the medium’s extinction coefficient at that frequency is too high.

[0044] Embodiments include but are not limited to semiconductor-based impulse generators, such as those using step recovery diodes (SRDs), to generate narrow pulse width impulses from the high and low frequency square waves of the clock generator circuit. Suitable pulse widths provide an acceptable range resolution according to Equation 3, whereAR is range resolution, c is the speed of light, and r is pulse width. Accordingly, a pulse width of 500 picoseconds provides a range resolution of 7.5 cm. c x T (3 x 108m / s)(5 x 10"los)Eq. 3 AR = — — = - — - - = 0.075mM2 2

[0045] FIG. 8 is a diagram of an impulse generator circuit 800 according to one embodiment. A square wave clock signal 802 is shown as input on the left, and a corresponding pair of output impulses 804 is shown above-right. The impulses 804 are relatively narrow in pulse width compared to the clock signal 802. Step recovery diodes (SRD) 806A and 806B are used to narrow' the pulse w idth of the output. Examples of suitable SRDs are MACOM’s MMDB30 and MMDB45 step recovery diodes. According to the manufacturer’s specifications, the MMDB30 SRD has a lifetime betw een 1-4 ns and a transition time between 30-38 ps. The MMDB45 SRD has a lifetime between 3-8 ns and a transition time between 45-58 ps. The output pulse shape of the SRDs can be modified using an appropriate network to adjust the SRD’s input signal. More specifically, where the SRD is the MMDB45, making the resistors (808 and 810) 50 Ohms provides an output pulse width of about 500 ps. The person having ordinary skill will recognize that increasing the resistance of 808 and 810 would increase the pulse width. For example, changing the resistors 808 and 810 from 50 Ohms to 400 Ohms would roughly double the output pulse width. Accordingly, embodiments are not limited to 500 ps pulse widths. Rather, the skilled artisan, having the benefit of this description, will understand how to adjust the circuit to vary pulse width to suit range resolution requirements of a given application.

[0046] With continuing reference to FIG. 8. diodes 806A and 806B are also configured to modify the frequency content of the inputted clock signal 802. More specifically, the diodes function as a frequency doubler and a harmonic generator. Therefore, the clock circuit’s fundamental frequency can be converted to a broadband signal having much richer frequency content, which can be used to improve range resolution. Finally, the impulse generator of FIG. 8 includes a schottky diode 810 that functions to rapidly switch the circuit between charging and discharging states.

[0047] The pulse generated from the higher frequency signal is amplified and communicated to an antenna where it is transmitted at a predetermined pulse repetition frequency (PRF). The individual transmitted pulses are known as radar pulses. These pulses are transmitted in frames of a predetermined number equally spaced in time. The transmittedpulses are reflected from a target, and a portion of the transmitted energy is returned to the antenna.

[0048] The slight PRF difference between the higher and lower frequency pulses causes the waveforms to phase shift at a constant rate so that each successive coinciding period of high and low frequency pulses shift in constant fractions of a period according to Eq. 4 where At is the time step in picoseconds, / ? is the higher frequency, and fi is the lower frequency. This phase shift corresponds to an effective sampling frequency, where the lower frequency pulse triggers a sampling mixer to sample the higher frequency radar return pulse.1 E7q. 4 A A At = - f ~ 5 — fi fi2The frequency difference is within a range limited by Nyquist’s Theorem (Eq. 5), which states that in order to sample a signal without loss of information, the sampling frequency fi must be at least twice the highest frequency fmax contained in the signal.Eq. 5 fi > 2 x fmax Nyquist's TheoremIf the highest frequency content fmaxof a radar pulse is 5GHz then, according to Nyquist, at the minimum sampling frequency fsof 10 GHz a measurement is made every 100 picoseconds. In other words, the difference between the higher and lower frequencies must be small enough to trigger sampling in 100 picosecond steps from period to period. At 3 GHz. the minimum is 6 GHz or 167 picosecond steps, and at 2 GHz the minimum is at least 4 GHz or 250 picosecond steps. Nyquist limits the minimum sampling rate, but the maximum sampling rate is only limited by the frequency resolution of the clock generator circuit. In other words, the minimum frequency difference between two clock signals that a clock circuit is capable of generating defines the upper limit of effective sampling rate. While the exact frequency difference is not critical, what is important is that the difference is set so that the resultant phase shift is small enough to sufficiently sample the returning radar pulses according to Nyquist’s Theorem.

[0049] Embodiments are capable of arbitrarily high effective sampling rates with low-end ADCs capable of no more than 1 or 2 megasamples per second (MSPS) because the individual pulses in a frame can be spaced arbitrarily far apart e.g., one or two microseconds apart. Accordingly, any ADC can be used in an embodiment provided that the ADC’s sampling rate is compatible with the application. Since lower ADC sampling rates correspond to slower A-Scan and B-Scan production, a slower ADC will extend dataacquisition time. Data acquisition times on the order of seconds can be acceptable in applications such as a ground penetrating radar where the target is stationary, but could be unacceptable in applications where the target is in motion at high rate speed e g., a missile traveling at Mach 5. In such cases, a faster ADC would be warranted; however, a sampling technology according to embodiments of the invention may still be used.

[0050] Ultimately, the lower frequency pulse is a local oscillator (LO) used to trigger a sampling mixer to sample the returning higher frequency radar pulse after it is reflected from a target. Therefore, the constant phase shift enables embodiments to sample the reflected wave across the waveform at enough points to satisfy Nyquist’s Theorem, regardless of when the pulse arrives, so long as it arrives during a time window defined by the PRF. The samples are then used to reconstruct a single waveform, representative of the individual RF return pulses.

[0051] FIG. 9 is an illustration of a sampling mixer 900 according to an embodiment of the invention. As shown, the RF return signal is carried by line 904, while the positive and negative components of the local oscillator signal are carried on lines 902A and 902B. When the LO signal arrives at the schottky diodes 906 and 908 it biases the diodes for a period approximating the pulse width of the LO pulse. In some embodiments the pulse width may be about 500 ps. In some periods of an LO pulse train, a reflected RF pulse may coincide in time. In such instances, the energy added by the RF pulse causes the schottky diodes 906 and 908 to switch from the off state to the on state. The resulting signal is sent to an analog to digital converter 910 where it is quantized.

[0052] With continuing reference to FIG. 9, the analog to digital converter (ADC) 910 can be any relatively low performance ADC such as that which is found on a conventional low- cost microcontroller. Such ADCs typically operate around 1 to 2 MSPS. While slow among ADCs in general, this is still too fast for the CPU of such a low-cost microcontroller to handle writing the data to an onboard buffer memory. Therefore, embodiments use DMA circuit 920 to write to a buffer memory 930. From there the data can be transferred by conventional bus architectures such as USB 940 to a personal computer 950 or similar general purpose computing device e.g.. a tablet computer. The PC can then conduct asynchronous image reconstruction processes. For example, the ADC’s output can be used to construct A-Scans as shown in FIG. 11 which may be assembled into B-Scans as shown in FIG. 12.

[0053] The number of pulses transmited in a frame corresponds to the range R of the radar. In general, more pulses in a frame corresponds to greater range. The reason is that the higher and lower frequency pulses phase shift at a constant rate from period to period. FIG. 4B illustrates the progressive phase shift between the 782 kHz and 782.01 kHz impulses, as produced by the impulse generator, and before amplification or transmission of the RF pulse. Plot (a) of FIG. 4B shows a 16.4xl012second phase shift (At) in the time domain between the second period impulses. Each successive period multiplies the phase shift At by an integer corresponding to the period. As show n in plot (b) of FIG. 4B, at period M the total phase shift is M*At, where M is a pulse in the middle of a frame. Similarly, in plot (c) of FIG. 4B at period N the total phase shift Attotai is N*At, where N is the last pulse in the frame, as shown in Eq. 6.

[0054] Total phase shift Attotai corresponds to a time delay between generating the lower frequency pulse and generating the higher frequency pulse, thereby providing time for the RF pulse to travel down-range, reflect from a target, and reach the sampling mixer at about the same time as the lower frequency trigger pulse. If N is 1000, then the time delay Attotai given by Eq. 6 is 1.64xl0'9s, corresponding to a maximum distance to the target i.e., range R, of 2.5 meters according to Eq. 7. In other words, if a frame has 1000 pulses, at period number 1000, the lower frequency pulse is generated 1.64xl0'9seconds after the higher frequency pulse, which provides sufficient time for the RF pulse to travel to a target 2.5 meters away and return to the antenna during the 500 ps window where the lower frequency pulse is forward biasing the schotky diode.Eq. 6 Attotal= N X At= 1000 X 16.4 X 10“12= 16.4 x 10-9sR = 2.5 m

[0055] In this embodiment, the higher frequency is 782.01 kHz, so the period between higher frequency pulses is 1.28xl0'6s. As long as the RF signal can travel to a target and back in less than 1.28x1 O'6seconds, no aliasing will result. This provides a theoreticalmaximum range Rmax and corresponding maximum number of pulses Nmax and a maximum total phase shift Attotai. Given that 1.28x1 O'6seconds corresponds to 384 meters at the speed of light, a target up 192 meters away from the antenna can be detected without aliasing, assuming signal strength is sufficient to return a measurable RF pulse. A range of 192 meters, corresponds to about 780,000 pulses per frame where the pulses shift in increments of 1.64xl012seconds. The person having ordinary’ skill in the art will understand that such a large number of pulses per frame will greatly increase power consumption by nearly three orders of magnitude compared to a frame of 1000 pulses. Therefore, it is generally desirable to use as few pulses per frame as necessary’ to meet range requirements. The person having ordinary skill will understand how to adjust At, signal strength, pulse width, frequency content, and N to produce a desired maximum operable range Rmax and range resolution AR.Eq. 8 Nmax Rmax / c tAmax= 780,000

[0056] With regard to pulse repetition frequency (PRF), the faster the rate of transmission, the shorter the maximum range Rmax, but the faster the rate of reconstructing A-Scans and the faster B-Scans can be generated. However, competing factors place limits on the rate of transmission. For example, as discussed above, faster transmission rates correspond to smaller maximum effective ranges of a radar because a pulse must travel to a target, be reflected, and travel back to the receiver before the next pulse is transmitted. If the rate of transmission is too fast, aliasing will result, producing false signals. Functional transmission rates within the scope of the invention leave sufficient time for a radar pulse’s maximum expected time-of-flight. In the present embodiment, the maximum effective range Rmax is about 192 m. Provided the target is never more than 192 m from the embodiment, a transmission rate of 1.28 MHz is sufficient. Larger maximum ranges would require slower PRFs.

[0057] In certain embodiments, like ground penetrating radar operating in a conductive environment like wet soil, penetration is expected to be relatively’ low, on the order of a few meters at low power levels and broadband frequency content peaking at about 2 GHz. Such a shallow maximum range Rmax requires relatively few pulses per frame, and permits anarbitrarily large PRF. Therefore, the PRF can be set to coincide with the sampling capability of the onboard ADCs of a typical microcontroller e.g., 1 to 2 MSPS.

[0058] Further, if the maximum penetration of a radar is 3 meters, then the maximum time of flight of a radar pulse is 20 nanoseconds. Accordingly, a theoretical maximum transmission rate is 20 GHz, without considering other design parameters such as power consumption, speed of the ADC, and data handling. However, the present invention enables the use of much slow er low-cost ADCs that are generally limited to sampling at rates of a couple MHz e.g., 2 MSPS.

[0059] The rate of transmission also corresponds to the number of frames that can be transmitted per second. The greater the frames per second, the faster that B-Scans can be assembled. Regarding the number of pulses in a frame, as the number increases, the length of time and pow er required to generate a B-Scan may increase to an unacceptable degree. As the number of pulses per frame increases, eventually, this w ould impact an embodiment's capacity to include A-Scan averaging to improve signal-to-noise due to the time required to generate additional pulses. Accordingly, embodiments must balance a number of factors to achieve desired performance results. The process of balancing these factors is within the ordinary skill in the art as a matter of design choice.

[0060] One non-limiting example, comprises a Silicon Labs Mighty7Gecko (EFR32) microcontroller that sets up a Texas Instrument clock generator (CDCM6208). The clock generator creates two frequencies very close to each other, but different enough that the rising edges strobe past each other. In this example the higher frequency pulse is 782.01 kHz and the lower frequency pulse is 782 kHz, a difference 10 Hz. Both pulses are sent to identical impulse generators that convert the clock generator's step function to a 500 picosecond pulse width broadband pulse, approximating a delta function and providing a 7.5 cm range resolution, A / ?.

[0061] The frequency content of both pulses is between 500 MHz and 5 GHz. The content peaks around 2 GHz starts to taper off at 3 GHz and is very low as frequency increases to 5 GHz. According to this embodiment, the higher frequency 782.01 kHz pulse is amplified to about 35 dBm and sent out through a transmit antenna. According to the present embodiment, this level of amplification provides sufficient power to compensate for losses in the return signal. It will be understood by the skilled artisan that some portion of the energy transmitted will interact w ith a target, if present, and be reflected back to the antenna andmeasured. The person having ordinary skill will understand how to adjust amplification to provide sufficient power for making measurements over given ranges in given media.

[0062] The lower frequency 782 kHz pulse is amplified to about 15 dBm. This pulse triggers a sampling mixer. Accordingly, the degree of amplification is selected to sufficiently forward bias the schottky diode of the sampling mixer so that a returning RF pulse will be sampled if it arrives during the roughly 500 ps that the trigger pulse maintains the bias. The sampling mixer captures the value of the higher frequency 782.01 kHz pulse and uses the resulting signal to charge a capacitor. The capacitor’s charge fluctuates until the lower frequency pulse stops forw ard biasing, at which point the value of the capacitor is fixed and can be measured.

[0063] FIG. 1 is a plot 100 showing the higher frequency 104 and lower frequency 102 waveforms. The vertical line 106 show s the point 108 w here the peak of the lower frequency pulse coincides in time with the higher frequency pulse 104. This point is where the sampling mixer is triggered, and therefore where the higher frequency pulse 104 is sampled.

[0064] Pulses may be organized into frames. For instance, in the present example a 782.01 kHz radar pulse may be transmitted in a frame of 512 pulses at a rate of 1 MHz. That is, the identical radar pulse is transmitted 512 times, comprising a frame of 512 pulses, and they are transmitted at a rate of 1 MHz. FIG. 2 is a plot 200 show ing a train of higher 104 and low er 102 frequency pulses spaced apart at approximately Ips intervals, corresponding to the transmission rate of 1 MHz. FIG. 2 shows an entire frame 206 of 1024 pulses spaced at approximately I s intervals. The person having ordinary skill in the art will understand that the number of pulses per frame can vary', as w ell as the rate at w hich they are transmitted.

[0065] Since the higher and low er frequency pulses differ slightly in frequency by 10 Hz, they effectively march along each other in roughly 16.4 picosecond steps according to Eq. 4 above. FIG. 4A is a plot 400 showing the phase shift At of three successive idealized periods of high 104 and low 102 frequency pulses, with the earliest period being at the bottom, and the latest period at the top. Each successive pulse shifts by a constant amount At.782,000.01 - 782,000 Applying Eq. 4 At = - 782 000^ -= 16,4 PS

[0066] Each sample taken at 16.4 ps intervals can then be amplified by an operational amplifier and sent to an ADC to measure its value. The data may be low-pass filtered to remove high frequency noise resulting from capacitor decay. FIG. 5 is a plot 500 of a pulsetrain before 510 and after 520 amplification. The lower plot 510 shows the output of the sampling mixer prior to amplification, while the upper plot shows the pulse train post amplification 520. High frequency noise 522 is still visible in the amplified signal 520, resulting from charge leakage from the capacitors of the sampling mixer. FIG. 6 is a plot 600 of the pulse 610 after applying a low-pass filter. FIG. 7 is a plot 700 of the reconstructed A- Scan pulse 710.

[0067] Data obtained from the individual pulses of a frame is used to reconstruct a single A-Scan waveform, one frame corresponding to one A-Scan. The number of samples across the waveform is given by Equation 9, where PW is pulse width. In embodiments such as this one, having a 500 ps pulse width, the number of samples at unique phase angles (represented in terms of time as At) is a minimum of 30; how ever, the fractional component of N prevents / ; pulses over 30 from triggering the sampling mixer at the same phase angles as the preceding 30. One thousand successive samples will trigger the sampling mixer at one thousand unique phase angles. 30.5

[0068] Since successive pulses are about 1 ps apart, the ADC is required to transmit measurements to memory at a rate exceeding the capacity of the microcontroller’s CPU. This is especially true given the many randomly timed interrupts from other processes. Therefore, a DMA circuit handles writing the ADC’s output to a buffer memory7. From there the data can be transmitted over known bus architectures such as USB 2.0 or USB-C to an off-board computer that assembles the A-Scans (FIG. 11) into B-Scans (FIG. 12) and conducts image analysis.

[0069] Regarding FIG. 12, the vertical axis corresponds to sample number, and the horizontal axis corresponds to A-scan number. A B-Scan is a plot of vertically7arranged A- Scans with the first A-Scan appearing on the left and the last A-Scan appearing on the right. If the antenna moves horizontally at a known constant rate, the horizontal axis can be converted to the distance traveled by the antenna. Generally, traveling down the vertical axis from top to bottom corresponds to increasing distance from the antenna. In some embodiments discussed herein, samples are taken every716.4 picoseconds. Since the radar pulse travels at the speed of light each 16.4 ps sample corresponds 2.5 mm distance from the antenna. Therefore, sample number 2000 corresponds to 5 meters from the antenna. In this particular B-Scan, the zero point is the first moment that the LO pulse triggers the samplingmixer. Of course, no radar can be detected until the first pulse is reflected from a target. So, samples 0 through roughly 550 can be neglected. Embodiments may make a vertical axis correction by subtracting out the samples prior to the first RF pulse being transmitted from the antenna, making zero correspond to the position of the transmitter.

[0070] Continuing with FIG. 12, the gray scale plot indicates signal strength. White and black represent the strongest signals, but have opposite signs. The gray shade midway between black and white is a zero reading. In the B-Scan shown in FIG. 12, the antenna is stationary, resulting in horizontal bands. Additionally, the skilled artisan will understand that such data will contain certain artifacts such as hyperbolic waves 1202. The hyperbolic waves 1202 correspond to a single target. In some embodiments known mathematical methods may be applied to the B-Scan data to reveal the shape of the target; however, in other embodiments a machine learning data model may be applied to recognize targets in the B- Scan data.

[0071] Onboard ADCs according to embodiments of the invention may have an effective number of bits (ENOB) as low as 8 but typically range from 8 to 14. In one non-limiting example, a 12 ENOB ADC provides a SNR, according to Eq. 10, of 74 dB, where N is the effective number of bits. In embodiments where reflected radar signals are expected to be around 13 dB, a 74 dB SNR leaves ample dynamic range.Eq. 10 SNR = 6.02 X N + 1.76

[0072] Since the CPU is not involved in transmitting data to memory, the speed of data transfer is sufficiently fast to complete the task prior to the arrival of the next reflected pulse. Conversely, if the CPU was involved in transferring data from the ADC to memory, the CPU would degrade the ADC’s performance, making it impossible to reliably acquire an A-Scan.

[0073] The position of a target is determined based on the time-of-flight of a transmitted radar pulse (f ), which is determined by the difference in phase angle between the higher and lower frequency waveforms after accounting to phase shift effects due to their slight frequency difference. For example if, as in the present example, the difference between the lower frequency fi pulse and the higher frequency f2 pulse is 10 Hz, then according to Eq. 4 the At from one period to the next is 16.4 ps. If the 2 pulse is detected at 516.4 ps then the extra 500.0 ps is due to time-of-flight. A 500 ps time of flight corresponds to a travel distance dtravei of 1.5 m (Eq. 11) and a target distance dtarget of 0.75 m (Eq. 12).Eq. 11 dtravel= (3 X 108m / s)(500 X 10’12s)travelEQ- 12 dfarget2

[0074] FIGS. 10A and 10B show an embodiment 1000 comprising a small printed circuit board (PCB) 1004 in a two-part anodized aluminum heatsink case 1006 A, 1006B. The circuitry on this board 1004 is described elsewhere herein. As shown here, board 1004 includes connectors 1002A, 1002B to an external power supply, and a USB port 1008. According to this embodiment 1000 the PCB 1004 is a fully self-contained miniaturized impulse radar containing all necessary components except a power source and data analysis capabilities, both of which are off-board. The ruggedized case 1006A, 1006B and small size permits the embodiment 1000 to be used to provide radar capabilities to a wide range of devices including, without limitation, hand-held ground penetrating radar units, warehouse robots, vehicle collision avoidance, and machine vision. The antenna of such a device may be directed in a user-selected direction. For instance, a user may expect a target to potentially be underground and may point the antenna dow nw ard where the target is suspected of being located. Such a direction is a user-selected direction according to the invention.

[0075] Embodiments may also include onboard geolocating components, such as global positioning system (GPS) components, for locating the radar. Accordingly, such embodiments are configured to map the radar data to GPS data so that targets can be associated with a specific geolocation. This may be especially beneficial in ground penetrating radars for locating buried utilities. Once found and recorded, the location of the utility can be stored and recalled as needed.

[0076] It will be apparent to those skilled in the art that the above methods and apparatuses may be changed or modified without departing from the general scope of the invention. The invention is intended to include all such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

[0077] Having thus described the invention, it is now claimed:

Claims

I CLAIM:

1. A miniature impulse radar, comprising: a microcontroller having an onboard analog-to-digital converter and an onboard central processing unit, wherein the central processing unit configures the analog-to-digital converter including setting a sampling rate; a clock generator configured by the CPU of the microcontroller to output a first frequency and a second frequency, wherein the first frequency is different from the second frequency; a first impulse generator configured to receive the first frequency output of the clock generator, and output a pulse; a second impulse generator configured to receive the second frequency output of the clock generator and output a pulse, the output of the second impulse generator being substantially the same output as the output of the first impulse generator; an antenna configured to receive the output of the first pulse generator and transmit the output through space in a selected direction at a predetermined pulse repetition frequency, wherein the antenna is configured to receive reflected pulse and output the received reflected pulse as an analog voltage signal; a sampling mixer configured to receive the analog voltage signal from the antenna at a first terminal, and the output of the second impulse generator at a second terminal, and to produce an output signal when the output of the second impulse generator forward biases a diode of the sampling mixer, wherein the output signal of the diode is received by and charges a capacitor of the sampling mixer; the analog-to-digital converter being in electronic communication with the capacitor of the sampling mixer and configured to receive an output signal therefrom, and convert the signal from the capacitor to a digital signal corresponding to a magnitude of the signal from the capacitor, and to output the converted digital signal; and a direct memory access circuit configured to receive the output of the analog-to-digital converter and write the output of the analog-to-digital converter to a buffer memory'.

2. The miniature impulse radar of claim 1, wherein the sampling rate of the analog-to-digital converter is between 1 MSPS and 2 MSPS.

3. The miniature impulse radar of claim 1. wherein the first frequency and the second frequency differ by an amount corresponding to a time domain shift no smaller than the Nyquist sampling rate.

4. The miniature impulse radar of claim 3. wherein the highest measured frequency is between 1 GHz and 3 GHz.

5. The miniature impulse radar of claim 4, wherein the time domain shift is between 10 picoseconds and 250 picoseconds.

6. The miniature impulse radar of claim 1, wherein the output of the first impulse generator is a broadband pulse having a pulse width betw een 200 and 1000 picoseconds.

7. The miniature impulse radar of claim 1. wherein the output of the second impulse generator is a broadband pulse having a pulse width between 200 and 1000 picoseconds.

8. The miniature impulse radar of claim 1, wherein the pulse repetition frequency of the antenna is between 1 MHz and 2 MHz.

9. The miniature impulse radar of claim 1, wherein the diode of the sampling mixer is a schottky7diode.

Images