RU64386U1 - SYSTEM ANALYSIS OF SPECTRA OF NARROWBAND SPACE RADIO EMISSIONS - Google Patents

Abstract

The invention relates to radio astronomy and can be used for receiving, recording and analyzing narrow-band radio signals from space sources, for example, radio emissions in the spectral lines of gas clouds in space. In order to improve performance, a narrow-band cosmic radio spectrum spectral analysis system comprising a series-connected antenna, a directional coupler, an amplifier channel and an analog-to-digital voltage converter, as well as a modulated noise generator connected to a directional coupler, a computer and a reference frequency generator are introduced in series connected main and additional programmable logic integrated circuits and an interface controller connected to said computer set, and a memory card connected via the system main controller with a programmable logic integrated circuit, and this logic circuit being connected to said reference oscillator with a modulating input of the noise generator and the control input of analog-to-digital converter voltage. Improving the system performance is achieved due to the fact that the structure of the main programmable logic integrated circuit provides operations for calculating the fast Fourier transform of parallel channels. At the same time, it is possible to maximize the number of spectrum implementations averaged during the observation time of the radiation source, which provides higher accuracy of measuring the spectrum parameters at a given observation interval (or reducing the observation time at a given level of accuracy).

Description

The invention relates to radio astronomy and can be used for receiving, recording and analyzing narrow-band radio signals from space sources, for example, radio emissions in the spectral lines of gas clouds in space.

To analyze the spectra of narrow-band space radio emissions, radio telescopes are used equipped with highly sensitive receiving devices (usually with cryogenic cooling of input amplifiers) and analyzers of the spectra of received signals of one type or another. Methods and analysis equipment can be found, for example, in the following works: Ryzhkov N.F. "Hardware methods of radio spectroscopy of the interstellar medium." Astrophysical research. Proceedings of the CAO Academy of Sciences of the USSR, v.6., - L., Nauka, 1974; or Zinchenko I.I. "Technique of millimeter and submillimeter astronomy", http://www.astronet.ru/db/msg/1190067/index.html; or Gosachinsky I.V., Zhelenkov S.R. Digital Autocorrelation Spectrum Analyzer. Preprint No. 96. SAO RAS, St. Petersburg. 1993; or Benz A.O., Grigis PC, Hungerbuhler V., Meyer H., Monstein C., Stuber B., Zardet D. “A broadband FFT spectrometer for radio and millimeter astronomy”, http: //www.astro.phys .ethz.ch / papers / meyerh / FFT_spectrometer.pdf.

The received radio signals from space sources are very weak and usually not visible against the background of a stronger intrinsic noise of the radio telescope. Therefore, to isolate signals and measure their spectra, one has to observe sources and accumulate signals for a sufficiently long time (tens of minutes or more). In addition, it is necessary to measure the parameters of the receive-amplifier channel with high accuracy in order to convert the spectrum parameters measured at the channel output to the spectrum of the signal induced by the radiation source in the antenna.

In radio astronomy systems, various spectrum analyzers were used, but correlators in combination with radiometric power measurement devices were most often used to study narrow-band signals. For example, the systems of spectrum analysis presented in the following publications are constructed by this principle: Ryzhkov N.F. "Hardware methods of radio spectroscopy of the interstellar medium." Astrophysical research. Proceedings of the CAO Academy of Sciences of the USSR, v.6., - L., Nauka, 1974; Gosachinsky I.V., Zhelenkov S.R. Digital Autocorrelation Spectrum Analyzer. Preprint No. 96. SAO RAS, St. Petersburg. 1993. The need to use additional radiometric channels for measuring signal powers, loss of signal accumulation time associated with power measurements, and additional loss of sensitivity due to noise clipping

signals fed to the correlator significantly reduce the effectiveness of analyzers of this type.

In recent years, correlation analyzers have been replaced by more advanced digital spectrum analyzers based on the fast Fourier transform (FFT). Such analyzers are widely used in measurement technology (see, for example, the PXI Product guide. Prospectus from National Instruments, 2004). An FFT spectrometer that analyzes signals in the video frequency band F produces a spectrum at discrete frequencies f _i spaced by a predetermined value w (frequency resolution). The number of discrete frequencies in the spectrum is N = F / w.

In radio astronomy, the use of FFT spectrometers is difficult, since it is necessary not only to conduct amplitude calibration of the spectra with high accuracy, but also to extract signals from the stronger intrinsic noise of the radio telescope. At the same time, the solution of these problems using FFT spectrum analyzers makes it possible to obtain a high frequency resolution, flexibly change the initial conditions (tasks) of the analysis, and also to eliminate additional channels for measuring power and the loss of signal accumulation time associated with these measurements. Examples of the use of FFT spectrometers for studying cosmic radio emissions can be found, for example, in the application for invention No. 2006115184/28 (016507) “Method for measuring the energy spectrum of narrow-band cosmic radio emission”, as well as in the article “A broadband FFT spectrometer for radio and millimeter astronomy” , Benz AO, Grigis PC, Hungerbuhler V., Meyer H., Monstein C., Stuber B., Zardet D., http://www.astro.phys.ethz.ch/papers/meyerh/FFT_spectrometer.pdf ..

The closest in purpose and technical essence is the system for the analysis of narrow-band space radio signals described in the application for invention No. 2006115184/28 (016507) “Method for measuring the energy spectrum of narrow-band cosmic radio emission”.

The known system includes an antenna, receiving and amplifying channel of a radio telescope, converting the received radio signal to video frequencies, a modulated noise generator connected to the input of the receiving and amplifying channel through a directional coupler, and a digital FFT spectrometer, which includes an analog-to-digital voltage converter (ADC) located on the PXI-5620 board, as well as a computer and a highly stable reference frequency generator located in the PXI-8186 cassette. An operator interface (monitor, keyboard and mouse) is connected to this cassette.

In the known system, the ADC reads the analog noise signal at the output of the receiving channel and generates a packet of digital samples of the 2N signal necessary for the FFT. This packet is then transferred to a computer, where the spectrum is calculated by the FFT method. In this way, the spectra are calculated alternately at

turned on and off, the noise generator, which is modulated by the meander voltage and is necessary for amplitude calibration of the parameters of the receiving and measuring channel and subsequent conversion of the spectra at the channel output to the signal spectra in the antenna. The signal spectra obtained during the observation of the source are averaged separately for different half-periods of modulation of the noise generator (with the introduction of calibration noise pulses and without them). Using averaged spectra, the computer calculates the spectrum of noise temperatures (or power spectrum) of the signal induced in the antenna.

In the known system, the operations of reading a signal using an ADC, transferring a packet of samples to a computer and calculating the spectrum by the FFT method are performed sequentially, and all computing operations by the computer are also performed sequentially. During the transmission of the packet and the calculation of the spectrum, the ADC is blocked. Therefore, it takes time t = t _sc + t _subt to get each spectrum implementation, where t _sc is the reading time required for FFT calculations of a packet of samples of a signal of size 2N; t _subt is the time taken to transmit the packet of samples to the computer and to the FFT of the spectrum calculation. Since the signal reading frequency f ₁ is set 2 times greater than the maximum signal frequency F at the ADC input, the reading time t _cf = 2N / f ₁ = N / F = 1 / w is determined by the specified frequency resolution interval w. The computation time t _subt is determined by the well-known empirical formula t _subt = h (2N) log ₂ (2N), where h is the coefficient with the dimension of time (see, for example, Blahut RE “Fast algoritms for digital signal processing”. Addison-Wesley Publishing Company, Inc, 1985). For modern FFT spectrometers, for example, NI-5620, h≈10 ^-5 -10 ^-6 sec. When analyzing with a fairly good frequency resolution (N≥100), the time t _subtract increases when the ADC is blocked and the received signal does not accumulate. The total time t of obtaining one spectrum implementation also becomes larger. At the same time, during a given observation time T of the signal, the number m = T / t of averaged spectrum realization will be less. Accordingly, the residual noise at the output of the spectrometer decreases more slowly and the output signal-to-noise ratio increases.

Thus, the known system, although it provides a reduction in the time of observation of the source in comparison with the systems previously used, does not yet reach the limit possibilities in terms of increasing the speed of the system and does not fully realize the potential possibilities of reducing the time of observation T and increasing the sensitivity.

The purpose of the proposed utility model is to increase the speed of the spectrum analysis system.

To achieve this, a system for analyzing the spectra of narrow-band cosmic radio emissions, containing a series-connected antenna, a directional coupler, a receiving-amplifying channel, and an analog-to-digital voltage converter, as well as a modulated generator

noise connected to the directional coupler, a computer and a reference frequency generator, introduced in series are connected the main and additional programmable logic integrated circuits and an interface controller connected to the said computer, as well as a memory card connected through the system controller to the main programmable logic integrated circuit, this the logic circuit is connected to the reference frequency generator, with a modulating input of the noise generator and with a control input of analog-digital voltage converter.

The introduction of two programmable logic integrated circuits (FPGAs), a system controller with a memory card and an interface controller made it possible to combine in time the process of reading the signal voltage at the output of the receiving-amplifier channel and the process of calculating its power spectrum, which is performed by the main FPGA. During the time t _mid , during which the ADC reads the packet of signal samples necessary for the FFT, the main FPGA manages to calculate the spectrum from the previous packet and prepare for processing the packet currently being read. The result is a continuous operation of the ADC, which reads the signal samples. Improving the system performance is achieved due to the fact that the structure of the main FPGA provides FFT computing operations on parallel channels. At the same time, it is possible to maximize the number of spectrum implementations averaged over the observation interval T. This number in this case is m = 1 / t _cf. This ensures a higher accuracy of the measurement of the spectrum parameters at a given observation interval or reduces the observation time at a given level of accuracy.

The system controller and memory card provide the formation of the necessary configuration (architecture) of the main FPGA. The additional FPGA in accordance with the initial analysis task coming from the computer through the interface controller sets the division frequency of the reference frequency and the clock frequency of reading the signal in the main FPGA. The main FPGA that calculates the implementation of the spectrum, sums them up and counts the number of implementations. In addition, it synchronizes the operation of the noise generator, producing a meander whose edges coincide with the moments of the end of the spectrum calculation cycles.

The spectrum summed over a given interval and the number m of spectra included in it are transmitted via an additional FPGA and an interface controller to a computer, which averages the power spectrum (by dividing the summed spectrum by the number m) and then calculates the desired signal spectrum in the antenna. The spectra obtained with the noise generator turned on and off are summed and averaged separately.

The figure shows a block diagram of the proposed utility model, where it is indicated:

1 - antenna;

2 - directional coupler;

3 - receiving-amplifying channel;

4 - modulated noise generator;

5 - analog-to-digital voltage converter (ADC), for example, AD9218 (10 bits) or AD6640 (12 bits);

6 - the main programmable logic integrated circuit (FPGA), for example XC4VLX25;

7 is an additional programmable logic integrated circuit, for example XC3S400;

8 - USB2 interface controller, for example CY7C68013;

9 - computer;

10 - system controller, for example HSSACE;

11 is a system memory card, for example SysACECF;

12 - reference frequency generator.

Antenna 1, directional coupler 2, receive-amplifier channel 3, ADC 5, main FPGA 6, additional FPGA 7, interface controller 8 and computer 9 are connected in series. The control input of the ADC 5 is connected to the main FPGA 6. A modulated noise generator 4 is connected to the second input of the directional coupler 2, and its modulating input is connected to the main FPGA 6. A memory card 11 is connected to the main FPGA 6 through the system controller 10. The reference frequency generator 12 is connected from the main FPGA 6. The clock inputs of the oscillator 12 and the local oscillators of the receive-amplifier channel 3 are connected to an external generator of a highly stable reference frequency f _sync (for example, to a hydrogen frequency standard).

When the power is turned on, the system controller 10 reads codes from the memory card 11, according to which it sets the configuration of the main FPGA 6. From the computer 9, the input data for spectrum analysis is entered into the additional FPGA 7 through the interface controller 8 — the setup code of the video frequency band F, in which the spectra will be measured , and the time of reception and accumulation of the signal T. In the F band, a portion F _n is allocated in which there is obviously no signal under investigation, and a portion F _s where there are or may be spectral components of the signal.

According to the input data, the additional FPGA 7 sets in the main FPGA 6 the division coefficients of the reference frequency f ₀ received from the highly stable generator 12, and determines the clock frequency f _{1 of} reading the samples of the signal, which is controlled by the ADC 5. This frequency is set 2 times higher than the maximum frequency of the analyzed signal, i.e. f ₁ = 2F. When using the AD9218 ADC, for example, the generator 12 can be tuned to a frequency f ₀ = 64 MHz, which will allow analysis of signals in the band up to 32 MHz. The frequency of the generator 12 and the frequency of the local oscillators of the receiving amplifier channel 3

are synchronized by a frequency standard (for example, hydrogen), which minimizes frequency instability (up to 10 ^-15 ) and provides an opportunity to analyze the spectra of microwave signals with high resolution.

From the input data entered into the main FPGA, the discrete frequency spacing w is also determined, which characterizes the frequency resolution and is rigidly connected with a given analysis band: w = F / N, where N is a fixed number of discrete frequencies in the analysis band. The number N is determined by the configuration of the main FPGA and is limited by the amount of memory blocks built into it. For XC4VLX25 FPGAs, for example, you can set N = 2048 or less.

A system prepared for operation in this way calculates the spectra cyclically. From the main FPGA 6, a logic zero voltage (log.0) is applied to the noise generator 4, which turns off the generator 4. At the same time, the ADC 5 is affected by a mixture of the received antenna signal and the noise of the radio telescope with a temperature T _c . ADC 5 with a frequency f ₁ takes and transmits digital signal samples (for AD6640, for example, 12 bits) to the main FPGA 6. Using the packet of signal samples received from ADC 5, the main FPGA 6 using the FFT method calculates the implementation of the power spectrum s ₁ (f) and saves it in its built-in memory block В ₁ . Subsequent implementations of the spectra calculated with the noise generator turned off are also recorded in block B ₁ and summed with the spectra recorded earlier and stored in it. At the same time, the number m of summed spectrum implementations is fixed.

The FPGA configuration provides FFT spectral calculations in parallel with reading the signal without additional time loss (the so-called conveyor method). The calculation time t of one realization of the spectrum is determined only by the read time of 2N samples of the signal and is equal to t _sc = 2N / f ₁ = 1 / w. To store the spectra in a memory block and during addition, the floating-point number format is used.

At the end of the calculation of the implementation of the spectrum for a given modulation half-cycle, the voltage supplied from the main FPGA 6 to the noise generator 4 changes (log 0 changes to log 1). In this case, the generator 12 is turned on and calibration noise with a small previously measured temperature T _{k is} added to the mixture of the received signal and the noise of the radio telescope. Now the main FPGA 6 calculates and sums the spectra s ₂ (f) with the noise generator turned on, storing them in the second built-in memory block В ₂ . The number m of summed spectrum implementations here is the same as when the noise generator is turned off. Upon completion of this modulation half-cycle, the FPGA operation cycle ends.

Next, the cycles of calculation and addition of the implementation of the spectra s ₁ (f) and s ₂ (f) are repeated. When a given accumulation time T is reached, the summed spectra s _1Σ (f), s _2Σ (f) and the number of summed realizations m

through the interface controller 8 is transmitted to the computer 9 using an additional FPGA 7.

The implementations of the spectrum s _1Σ (f) and s _2Σ (f) _summed during a given accumulation time T by computer 9 are divided by the number of spectral implementations m, as a result of which average power spectra are obtained. The computer 9 from the spectra after averaging for the frequency band F _n , where there is no signal, calculates the average noise power in the elementary frequency channel of width w with the GS turned off (P _1ш ) and with the GS turned on (P _2ш ). For frequencies f _i in the band F _s where the signal is present, the average noise powers in the elementary band w are calculated near the frequency f _i when the main switch is off (P _1i ) and when the _main switch is turned on (P _2i ). Using these 4 power values, the gain of the receiving channel, the noise temperature of the radio telescope, and then the components of the spectrum of noise temperatures of the signal under investigation, recalculated to the antenna of the radio telescope, are calculated: T _si = 0.5T _to [(P _2i + P _1i ) - (P _2ш + P _1sh )] / (P _2sh -P _1sh ). Here, all calculations are carried out according to the same formulas that are used in the known prototype system. Along with the spectrum of noise temperatures, the power spectrum of the signal in the antenna or the spectrum of the flow of electromagnetic energy (in yang) can be calculated.

Compared with the known system, the performance is significantly increased (the time to obtain one spectrum implementation is reduced), which makes it possible to obtain and average a larger number of spectrum implementations over a given time of signal reception and accumulation. In the proposed utility model, each implementation of the spectrum is calculated in a time t = t _sc , and in the known prototype device in a time t = t _sc + t _subt , and usually t _subt > t _sc . Thus, the proposed utility model provides a reduction in signal reception time by M = (1 + t _calc / t _sc ) times compared with the known prototype device. The number m of spectrum realization obtained during the reception and analysis of signal T. increases by the same amount of times (or time T additionally decreases for the same m).

For the NI-5620 spectrometer used in the known prototype device, h≈3 · 10 ^-6 seconds, which at N = 2048 gives t _calc = 135 ms. For example, when analyzing a spectrum with a resolution of w = 200 Hz, when t _sc ≈5 ms, the time gain will be M≈28 times. If the signal reception time T is fixed, then the output signal-to-noise ratio, the sensitivity of the system, and the accuracy of measuring the energy parameters of the spectrum increase by a square root of M times. In this example, these parameters are improved 5.3 times.

The proposed utility model implements the maximum possible (potential) speed of the system for analyzing the spectrum of cosmic radiation, since the signal is read and accumulated continuously (the value of t reaches a minimum equal to t _cf ).

According to the proposed scheme, an experimental model of the proposed utility model was manufactured at RELTA CJSC, which was installed on

32-meter radio telescope at the Svetloye Observatory and tested by the Institute of Applied Astronomy RAS. The system was tested both in laboratory conditions and in real observations of space sources of narrow-band radio emission (when observing radio emissions in spectral lines). The proposed model was compared with a previously manufactured prototype device based on the NI-5620 spectrometer.

The efficiency of the proposed spectrum analysis system and the possibility of reducing the source observation time to obtain the same signal to noise ratio and, accordingly, the same accuracy in measuring the amplitudes of the spectral components of the noise temperature of the signal are fully confirmed. The positive effect is especially noticeable when analyzing weak sources, when a long observation time was required (tens of minutes - hours)

Both analysis systems (both the proposed and the prototype) were significantly superior to the previously used correlation type systems in terms of analysis efficiency, speed and ease of use, but the proposed model provided an additional (compared to the prototype) gain in speed and a reduction in observation time of 10-30 times (depending on the specified analysis parameters).

Claims (1)

A system for analyzing the spectra of narrow-band space radio emissions, comprising a series-connected antenna, a directional coupler, a receiving-amplifying channel and an analog-to-digital voltage converter, as well as a modulated noise generator connected to a directional coupler, a computer and a reference frequency generator, characterized in that, for the purpose of increase system performance, introduced in series connected main and additional programmable logic integrated circuits and interface controller, sub connected to the said computer, as well as a memory card connected through the system controller to the main programmable logic integrated circuit, the main programmable logic integrated circuit connected to the said reference frequency generator, with the modulating input of the noise generator and with the control input of the analog-to-digital voltage converter.