RU2315327C1 - Device for analyzing signal spectrum - Google Patents

Abstract

FIELD: measuring technique.

SUBSTANCE: device comprises band amplifier with n outputs connected with n mixers whose outputs are connected with the output of the generator of linearly frequency modulated pulses through the delay line, unit for limiting the amplitude, time comparing circuit, and time-frequency converter. The control input of the generator is connected with the first input of the generator of synchronized pulses. The second output of the generator is connected with the second input of the time-comparing circuit.

EFFECT: enhanced precision.

2 dwg

Description

The invention relates to the field of radio engineering and can be used in analyzers of a spectrum of radio signals with high resolution, for example, for laser Doppler locators with a continuous radiation mode.

Spectrum analyzers are known whose operation is based on one of the following phenomena: interference, refraction in the presence of phase velocity dispersion, resonance. The first two phenomena are used to obtain optical spectra. Spectrum analyzers using resonance are the most versatile. As resonance, oscillatory circuits with lumped parameters or line segments with distributed parameters are used. There are resonant spectrum analyzers of parallel and sequential action. In parallel spectrum analyzers use a set of resonators tuned to different frequencies and simultaneously exposed to the studied oscillations. Serial spectrum analyzers use a single resonator with variable tuning. A parallel spectrum analyzer has an advantage over a serial analyzer in the speed of analysis, but is inferior to it in simplicity. A serial analyzer is suitable for analyzing periodic processes or processes whose nature changes little during the analysis.

Spectrum analyzers allow you to determine the amplitude and frequency of the spectral components that make up the analyzed process. Its most important characteristic is resolution - the smallest interval Δf in frequency between two spectral lines, which are still separated by a spectrum analyzer. The resolution is determined by the bandwidth of the resonator and is associated with the analysis time T by the relation Δf T = const, and the value of this constant depends on the parameters of the resonator. The value of T determines the time of establishment of oscillations in the resonator and is proportional to the selectivity of the latter, that is, its passband.

The selective properties of resonators are described by static resonance characteristics only with a slow tuning of the resonant frequency. In reality, the tuning is carried out at a certain speed, therefore, the concept of dynamic resonance curves is introduced for resonators, and the concept of dynamic resolution is introduced for sequential spectrum analyzers, which depends not only on the properties of the resonators, but also on the analysis time T. Analysis time T = 2F / πμ (Δf) ² , where F is the analysis bandwidth (width of the studied frequency range), μ is the allowable dynamic expansion of the passband.

The spectrum analyzer can give a true spectrum only when the analyzed oscillation x (t) is periodic or exists only within the interval T. When analyzing long-term processes, the spectrum analyzer does not give a true spectrum S (ω) over infinite integration limits, but only its estimate S _T (t ₁ , ω) with integration limits from t ₁ to t ₁ + T, where t ₁ is the moment of switching on, and such an estimate determines the so-called current spectrum of the analyzed oscillation.

Random processes are usually characterized by the energy spectrum G (ω), which determines the distribution on the frequency scale of the rms values of the signal used. Thus, the energy spectrum G (ω) of a stationary random process is related to the current spectrum S _T (ω) by averaging the square of the latter over the set of realizations. If the process is ergodic, then instead of averaging over an ensemble of realizations, one can use averaging over time along one implementation.

In addition to the considered analog spectrum analyzers, there is a wide class of digital analyzers in which instead of continuous implementations x (t), discrete values x (t _k ) = x _{k are used} at discrete time points t _k = k Δt, where k = 0, 1, 2, 3 ..., N-1, Δt = T / N. The samples x _{k are} quantized in magnitude, that is, represented by digital codes, for example binary, with a given number of binary digits. The development of computer technology has contributed to the appearance of spectrum analyzers, the action of which is based on the direct calculation of the expansion coefficients for a certain system of orthogonal, not necessarily harmonic, functions (see, e.g., A. Kharkevich, Spectra and Analysis, 4th ed., M., 1962; Jenkins G., Watte D. Spectral analysis and its applications, transl. From English., M., 1971).

Known spectrum analysis methods based on the use of matched filters (see, for example, B. R. Lezin. Theoretical Foundations of Statistical Radio Engineering, M., Sov. Radio, 1974, book 1-3; V. I. Tikhonov. Optimal reception of radio signals, M., Radio and communication, 1983). For optimal coherent reception, the following conditions must be met: the transmitted signals are fully known and can be accurately reproduced in the receiver, the Gaussian communication channel with constant parameters, there are no signal distortions in the communication channel, the spectral density of additive interference is known, the synchronization of the received and reference signals is ideal . As applied to laser Doppler location, these conditions can be considered practically feasible; therefore, the target detection process is reduced to recording a monochromatic signal in the field of view of the photodetector, whose frequency differs from the frequency of the probing continuous radiation by the value of the Doppler shift, which determines the radial velocity of the target relative to the locator. Analyzing the optimal signal processing algorithm, it should be noted that the ideal receiver (Kotelnikov receiver) determines the distance between the received signal and all other signals in the Hilbert space of the signals and decides that it was the signal that was received closest to it. However, this interpretation of the process of optimal signal processing is based on the fact that the entropy of the signal is not equal to zero, that is, the signal is modulated. In laser Doppler location, unmodulated monochromatic radiation of single-frequency lasers with zero entropy of the probing signal can be used, however, the signal reflected from the target contains complete information about the radial velocity of the target, and in this sense, the entropy of such a signal cannot be accepted as zero. In addition, the presence of the reflected signal itself also contains useful information about the fact that the target was detected by the locator.

The various structural diagrams of the optimal receiver are usually based on linear signal processing circuits. Their difference lies in the hardware implementation of linear processing schemes, which include the optimal correlation receiver, the optimal receiver on matched filters, and others.

When the laser Doppler locator is in the scanning mode along the angular coordinates (in the search mode), the received radiation from the target with an unknown Doppler frequency shift is a radio pulse of duration T _imp = γ / Ω, where γ is the instantaneous angle of view of the photodetector in the plane of angular scanning, Ω - angular scanning speed. When using a parallel spectrum spectrum analyzer, its passband ΔF _∑ is selected from the condition ΔF _∑ = 2 ΔV ν / s, where ΔV is the uncertainty band of the radial velocity of the target, ν is the laser frequency, c is the speed of light, and the number of filters M is determined by the passband each of them Δf ₁ = ΔF _∑ / M. So, with the duration of the received pulse T _imp = 0.5 ms and the analysis bandwidth ΔF _∑ = 10 MHz, we obtain M = 5000, and such an analyzer becomes very cumbersome. According to the sequential analysis scheme with a narrow-band filter (for the above example, Δf ₁ = 2 kHz), another complication arises due to the low speed of the locator in the mode of scanning it by angular coordinates, since the rate of change of the analyzer local oscillator frequency is selected from the condition df _get / dt≈Δf ₁ ² , and the search time in frequency in a given angular direction T _p = ΔF _∑ / Δf ₁ ² significantly exceeds the duration of the received pulse T _p >> T _imp (in this example, T _p = 2.5 s), which leads to the need for sharp lower Ia scanning angular velocity Ω in order not to miss the target.

Given these circumstances, the author proposed a spectral analysis method based on the following sequence of operations:

- the input signal is pre-amplified and divided into n parts with an approximately equal signal to noise ratio in them;

- form a linear-frequency-modulated (LFM) heterodyning signal in the form of radio pulses following with the frequency of the triggering clock signals;

- divide the generated chirp pulse into n parts;

- shift in time each of the n parts of the LFM heterodyning pulse by a time no less than an order of magnitude greater than the inverse of the passband in spectroanalysis using dispersion delay lines (DLS);

- convert the n components of the input signal into the corresponding LFM equivalents by mixing the n components of the input signal with n LFM heterodyning pulses mutually shifted in time;

- carry out spectral-temporal compression of the n obtained LFM equivalents of the input signal in the corresponding n matched filters on the DL;

- delay in time the compressed radio pulses until they are temporarily aligned with each other using the n-channel delay line;

- detect the amplitude of n delayed radio pulses;

- add in the analog adder n received by the detection of video pulses;

- subjecting the resulting video pulse to a minimum threshold limit in the comparator at a given value of the threshold level of restriction;

- measure the time interval between the occurrence of the pulse at the output of the comparator with a clock pulse, creating a chirp-pulse heterodyning;

- determine the frequency of the input signal - the frequency of the Doppler shift - the value of the specified measured time interval taking into account the characteristics of the used DLS.

Using this method of spectral analysis, or rather, measuring the carrier frequency of a received signal containing a Doppler shift, allows you to increase the product of the energy potential of the locator and its speed due to the fact that the number of analysis channels (n << ΔF _∑ / Δf ₁ ) with uncorrelated noise channels is significantly reduced components because the broadband path is used at the input, and the delay time Δτ in the splittable n LFM heterodyning pulses for adjacent channels significantly exceeds the inverse elichinu bandwidth input receiving path Δτ >> 1 / ΔF _Σ, which improves the signal / noise ratio in summable compressed video pulse in comparison with that in each individual radio pulse compressed (in one channel with DLA).

The considered method can be used in the claimed technical solution as the closest analogue (prototype).

The disadvantage of this method is its some cumbersomeness associated with the use of n of the same DLP and an additional n-channel delay line of compressed chirp pulses.

The aim of the invention is to simplify implementing the method of the device.

This goal is achieved in a signal spectrum analyzer containing an input strip amplifier with n outputs connected to n mixers whose heterodyne inputs are connected via an n-tap delay line to the output of a linearly-frequency-modulated pulse generator, the control input of which is connected to the first output of the clock generator , n outputs of mixers, respectively, are connected to n broadband amplifiers, as well as series-connected amplitude limiters to a minimum, a temporary comparison circuit, and a time-frequency converter, the second output of the clock generator being connected to the second input of the time comparison circuit, characterized in that it has an n-channel dispersion delay line connected to n outputs of the broadband amplifiers, having an output electrode common to all its n channels, connected through an output broadband amplifier to an amplitude detector, which is connected by an output to the input of the amplitude limiter to a minimum, and the time delays in the n-channel dispersion line buckle made corresponding delays in n-branch delay lines for the respective channels.

Achieving this goal is explained by a significant reduction in the number of equipment: instead of n separate DLA and the second (as in the prototype) n-channel delay line in the claimed technical solution used integrated DLZ. On its single sound-conducting substrate, n of the same type of DLZ channels are made of n distributed along the length of the input electrodes relative to the output electrode common to all n of its channels for organizing different time delays, consistent with delays in the corresponding channels of the n-tap delay line, in addition, in the claimed the technical solution n times reduced the number of output broadband amplifiers and amplitude detectors. The most complex part of such an integrated DLZ — its output electrode — is common to all n channels, therefore, the complication is associated only with the installation of n input electrodes on the common sound duct and with some extension of the sound duct to disperse these input electrodes along the length. Using lithography methods, the manufacture of such a DLZ does not cause any particular difficulties.

The claimed technical solution is clear from the presented figure 1. Figure 2 shows graphical explanations of the conversion process "frequency-time".

The device contains sequentially installed input strip amplifier with n outputs 1, n mixers 2, 3, ... 4, n wideband amplifiers 5, 6, ... 7, an n-channel dispersion delay line 8, containing n input electrodes 9, 10 , ... 11 and one common output electrode 12, the output broadband amplifier 13, the amplitude detector 14, the amplitude limiter at least 15, the time comparison circuit 16 and the time-frequency converter 17. The device also has a clock generator 18, the first output which is connected to the chirp generator owls 19, and the second with the second input of the temporary comparison circuit 16, and the n-tap delay line 20, the input electrode 21 of which is connected to the output of the chirp pulse generator 19, and n of its output electrodes 22, 23, ... 24 are connected to the heterodyne inputs of n mixers 2, 3, ... 4, respectively.

Consider the action of the inventive signal spectrum analyzer.

The general analysis implementation scheme based on the use of DLZ is reduced to converting the input signal into an LFM equivalent, followed by its compression into a DLM, consistent with the applied LFM heterodyning pulse. After detecting a short radio pulse and its minimum limit with a given limit threshold value, the edges of the video pulse obtained by the indicated processing are temporarily compared with a clock pulse that starts the LFM heterodyning pulse generator, and the input signal frequency is judged by the value of the measured time interval. Thus, the change in the frequency of the input signal is converted into a temporary offset of the pulse compressed in the DLZ [1-6]. It is important to note that the noise component of the broadband path of the LFM equivalent has practically no effect on reducing the signal-to-noise ratio at the output of the decider (i.e., at the output of the amplitude limiter to a minimum), since the amplification of the input signal entering the mixer input is sufficiently selected great.

The main parameters of the DLZ are the duration of its impulse response (delay time) τ _lz and the passband ΔF _lz , the product of which determines the value of the base B = τ _lz ΔF _lz >> 1. The energy of the received signal chirped equivalent determined by the duration of the impulse response τ _LZ, and by compressing the chirp signals in the agreed with him DLA at its output forms a very short radar pulse duration t _imp = 1 / ΔF _LZ, which increases the signal / noise ratio in the (B) ^1/2 times compared with the ratio at the input of the device. The construction of an n-channel analyzer at DLZ allows us to further increase this ratio by an additional (n) ^1/2 time. The latter is due to the uncorrelatedness of the summed noise components at the outputs of n broadband amplifiers 5, 6, ... 7 due to the introduced time delay of the LFM heterodyning pulses for the intervals Δτ >> 1 / ΔF _Σ due to the use of the n-tap delay line 20 in the formation LFM-equivalent input signal in mixers 2, 3, ... 4. The total delay in this line is (n-1) Δτ, which determines the total time delay ΔT _ls in the n-channel dispersion delay line 8 as the sum ΔT _ls ≥ (n-1) Δτ + τ _lz .

On figa shows a sequence of clock pulses that start the operation of the generator of the chirp pulses 19 and determine the beginning of the countdown in the time comparison circuit 16, which are generated in the clock generator 18.

Figure 2b shows the time-frequency diagrams of the process of generating LFM pulses of heterodyning with a linearly increasing frequency f _lchmG (t) in the working area of the analyzer, converting the input signal of frequency f _c = const (t) into signals - LFM equivalents - by mixing its frequency f _c with the frequency f _LChMG (t), as a result of which the LFM equivalents of the frequency f _LCHME (t) are formed. Since the frequency of the input signal is within a certain frequency band ΔF _Σ , with a change in the frequency f _c (with a change in the radial velocity of the target) within this range of ΔF _Σ , the characteristic of the LFM equivalent along the frequency axis will also shift, as shown by dashed lines. The shaded band bounded by the dotted line refers to the spectro-temporal characteristics DLA: its width corresponds to the passband DLA ΔF _LZ, and its length corresponds to the projection inclined line fluctuations chirped equivalents f _lchmE (t) on the x-axis which lies within said band DLA ΔF _LZ, and characterizes the pulse response DLA τ _LZ.

On figv presents the impulse response of the process of "compression" of the chirp-equivalent f f _LFME (t) at the output DLZ - signal u _imp (t) with a duration of t _imp = 1 / ΔF _lz . Its temporary position can vary within the limits indicated by the dotted line when the frequency of the input signal f _c changes in a given frequency range ΔF _Σ .

On fig.2g shows the pulse u _cp (t) of duration τ _back , generated at the output of the temporary comparison circuit 16 and determines the time interval between the moments of action of the fronts of the clock (Fig.2A) and the pulse at the output of the amplitude limiter to a minimum of 15 (Fig.2c) . This pulse can have a different duration τ _ass depending on the value of the frequency of the input signal f _c (shown by a dotted line).

The duration of the generated pulse at the output of the temporary comparison circuit 16 is processed in the converter "time-frequency" 17 by filling it with high-frequency counting pulses, following with a frequency f _SCH . Their number N _sc = τ _ass f _sc uniquely associated with the measured frequency f _c . The period following the counting pulses determines the measurement error of the time interval τ _ass and, therefore, the error in determining the frequency f _c . The value of the frequency of the input signal in a digital code is supplied to the corresponding Doppler locator device, which processes the received information in the interests of the functioning of those means for which the locator is used.

The claimed technical solution differs from the considered single-channel analyzer using DLZ in that it uses an n-channel DLZ in order to increase the signal-to-noise ratio at the output of the amplitude limiter to the minimum necessary to increase the energy potential of the locator (increase the target detection range). To organize the n-channel processing of the received signal, it is split into n equal parts, and all these parts are mutually coherent, and the oscillations themselves are considered monochromatic within the analysis procedure (with frequency f _c ). Each of these signals is converted in the mixers 2, 3, ... 4 (Fig. 1) to the corresponding LFM equivalents f _LFME (t) according to the rule f _LFME (t) = f _LFMH (t) -f _with conversion with the frequency of LFM of heterodyning _pulses f _lhmG (t), which are also split into n parts, but time-shifted for adjacent mixers 2, 3, ... 4 by the time interval Δτ using the n-tap delay line 20. It is very important to note that subject to the condition Δτ >> 1 / ΔF _{∑, the} noise components of the Gaussian noise in the signal components of the received signal are mutually uncorrelated we, since it is known that the correlation interval is inversely proportional to the channel band, i.e., equal to Δt _corr = 1 / ΔF _при , which, upon their subsequent addition at the output of DLZ 8, determines the growth of noise components in proportion to the square root of the number of processing channels n, while the accumulation signal components is proportional to the number n, which implies that the signal-to-noise ratio additionally increases in proportion to (n) ^1/2 . Since the LFM pulses of heterodyning in their n parts are also mutually coherent, since they are formed from a single source — the generator of LFM pulses 19, the signal components, when converted to LFM equivalents and subsequent processing in the n-channel DLZ 8, are added coherently to output last. Moreover, the difference in the duration of the LFM pulses of heterodyning is compensated by the same difference in the delays of the corresponding LFM equivalents in the n-channel DLZ 8, as can be seen from Fig. 1, and all n compressed radio pulses occur simultaneously at the output of the DLZ.

где Ф^-1 - обратный оператор от интеграла вероятностей вида:The compressed impulse response after amplitude detection is subjected to a threshold limitation, the limiting level of U _then should be chosen from the given value of the probability of false alarms G _lt , the probability of correct detection of S _obn and the spectral noise density at the output of the DLW σ _w . Relative Threshold Limit

where Ф ^-1 is the inverse operator of the probability integral of the form:

где μ_вых - значение отношения сигнал/шум на выходе ДЛЗ 8, величина которого по сравнению со значением μ₀ на входе приемника растет пропорционально (n В)^1/2.The probability of correct signal detection

where μ _o is the signal-to-noise ratio at the output of the DLZ 8, the value of which increases in proportion to the value of μ ₀ at the input of the receiver in proportion to (n V) ^1/2 .

Consider one example of the implementation of the inventive analyzer. Let a laser Doppler locator on a single-frequency continuous-wave CO ₂ laser (ν = 2.83 · 10 ¹³ Hz) be aimed at detecting a target and measuring its radial velocity V = 300 m / s when the target moves along a course that makes an angle to the optical axis of the locator within [-π / 6≤θ≤π / 6]. Therefore, the frequency of the Doppler shift f _{c of} laser radiation takes values from the range (5.66-4.90) · 10 ⁷ Hz, that is, the reception frequency band ΔF _∑ = 7.6 MHz. We select with a margin the reception band equal to ΔF _∑ = 10 MHz. Let the angular viewing speed and the instantaneous angle of view of the locator be such that the impulse response from the target does not exceed the value of T _imp = γ / Ω = 0.5 ms. We set the parameters of DLZ 8 - the passband ΔF _lz = 40 MHz and the duration of the impulse response τ _lz = 40 μs, that is, we choose a DLZ with the base B = 1600. Let the number of processing channels be n = 64. Then, for the correlation interval for the noise components Δt _corr = 1 / ΔF _Σ = 0.1 μs, we can choose the channel delay Δτ in the n-tap delay line 20 equal to Δτ = 1 μs, i.e., the total delay in this line will be slightly larger than Δτ n = 64 μs, which is feasible. The total delay in DLZ 8 with 64-input electrodes 9, 10, ... 11 will correspond to ΔT _lz ≥ (n-1) Δτ + τ _lz (the sign “more” indicates the presence of a technological gap between the last input electrode 11 DLZ 8 and its output electrode 12). In this case, the delay time of the pulse at the output of the DLZ (output of the amplitude limiter at least 15) τ _ass is about 110 μs. Since the frequency tuning range in the chirp generator 19 must be selected from the condition Δf _lchmH ≥ΔF _{Σ +} ΔF _lz = 50 MHz, this means that during contact time with the target T _imp = γ / Ω = 0.5 ms, k≤ T _imp / τ _ass (1 + ΔF _Σ / ΔF _ls ) = 10 analysis cycles, which further contributes to the refinement of the target parameter studied by the locator - its radial velocity (by averaging data in the time-frequency converter 17). The duration of the compressed video pulse at the output of the amplitude limiter to a minimum of 15 is of the order of t _imp = 1 / ΔF _lz = 25 ns, the resolution of the analyzer in frequency Δf _c = 1 / τ _ass = 25 kHz and corresponds to the number of channels of an equivalent parallel analyzer with a reception band ΔF _Σ = 10 MHz, equal to 400. The indicated frequency resolution corresponds to the accuracy of measuring the radial velocity equal to δV = сΔf _c / 2ν = λΔf _c / 2 = 0.132 m / s, which is quite acceptable. In this case, the signal-to-noise ratio at the output of the solver increases (V n) ^1/2 = 320 times compared to conventional analyzers without the use of DLZ and coherent accumulation. In this case, it is necessary to accurately perform the topology of the n-channel DLZ 8 and the n-tap delay line 20. In FIG. 2, the above example corresponds to a target moving a priori at a speed of V ₀ = 300 m / s when its course deviates from the optical axis of the locator by the angle θ = arccos (280.9 / 300) ≈20.6 ⁰ , since f _c = 53 MHz, which corresponds to the radial velocity of the target V = 280.9 m / s.

The specific characteristics of the energy potential of the locator with the analyzer considered are also determined by the power and short-term stability of the laser radiation frequency. The latter is important in connection with the need for coherent reception of radiations reflected from a diffraction-limited target, so that during the delay time of optical radiation to the target and vice versa (for a range of 5 km this delay is 34 μs in air), the mutual coherence of the mixed radiation is maintained. In order to increase the short-term stability of the laser radiation frequency, automatic frequency control (AFC) systems are used, known from the author [9]. So, for a single-frequency CO ₂ laser (λ = 10.6 μm), the characteristic slope is obtained for short-term frequency stability S _st ~ 100 Hz / μs, so that the frequency of the received and heterodyne radiation does not differ more than at 3.4 kHz, which provides a high degree of coherence when photographing optical radiation.

The claimed technical solution is new, reflects the inventive step and is industrially applicable, that is, meets the criterion of patentability.

Literature

1. Tverskoy V.I. Dispersion-time methods for measuring the spectra of radio signals. - M., Sov. Radio, 1974.

2. Filters on surface acoustic waves, ed. G. Matthews. - M., Sov. Radio 1981, 472 pp.

3. Smaller O.F. Matched Filter. Patent No. 2016493, BI No. 13 for 1994.

4. Smaller O.F. The method of analysis of the spectrum of signals, ed. testimonial. USSR No. 470758 for 1972

5. Smaller O.F. Location method. Patent No. 2296350, BI No. 9 for 2007.

6. Smaller O.F. A device for analyzing the spectrum of signals. RF patent No. 2040798.

7. Smaller O.F. Monopulse signal detector. RF patent No. 2046370.

8. Smaller O.F. The ultrasound microscope. Patent No. 2296350, BI No. 6 for 2006.

9. Smaller O.F. Automatic frequency control device, Aut. testimonial. USSR No. 350125 for 1970

Claims (1)

A signal spectrum analyzer containing an input strip amplifier with n outputs connected to n mixers, the heterodyne inputs of which are connected via an n-tap delay line to the output of the linear-frequency-modulated pulse generator, the control input of which is connected to the first output of the clock generator, n mixer outputs respectively, connected to n broadband amplifiers, as well as series-connected amplitude limiter to a minimum, a time comparison circuit and a time-frequency converter moreover, the second output of the clock generator is connected to the second input of the time comparison circuit, characterized in that it has an n-channel dispersion delay line connected to n outputs of the broadband amplifiers, having an output electrode common to all its channels, connected through the output of the broadband amplifier to an amplitude detector, which is connected by an output to the input of the amplitude limiter to a minimum, and the time delays in the n-channel dispersion delay line are made corresponding delays in the n-branch delay lines for the respective channels.

Images