RU2794928C1 - Noise automatic threshold adjustment method - Google Patents

Abstract

FIELD: optical signals.

SUBSTANCE: invention relates to the extraction of impulse signals from fluctuation noise, in particular to the technique for receiving impulse optical signals, and can be used in location, communications and other fields. The method of noise automatic threshold adjustment (NATA) additionally consists in the fact that the average frequency f₀ of the noise process crossing the zero level is preliminarily determined, and after the transient process is established, the threshold value U and the frequency f=N/T of exceeding the threshold by noise emissions in the last stabilization cycle are determined, after which the noise standard deviation is calculated by the formula

and

, where ξ << 1 is the confidence coefficient, the initial value of the threshold U is set as close to zero as possible, and the coefficient λ is chosen within (1-3)/f₀, where f₀ is the average frequency of the noise crossing the zero threshold.

EFFECT: prompt determination of the root-mean-square value of the fluctuation normal noise of photodetectors during the production and maintenance of miniature equipment (including portable and built-in equipment), with simple metrological support in a wide operational range.

1 cl, 6 tbl, 6 dwg

Description

The present invention relates to the extraction of impulse signals from fluctuation noise, in particular, to the technique for receiving impulse optical signals, and can be used in location, communications and other fields.

A well-known technique is the threshold selection of impulse signals against the background of fluctuation noise [1-3].

Technical solutions are known that provide automatic maintenance of the response threshold at the required level using noise automatic threshold adjustment (SHARP) [4-8].

Closest to the proposed technical solution is the method of noise automatic threshold stabilization [8].

The specified method of noise automatic threshold adjustment (SHARP) consists in comparing the noise process with the threshold level U, generating standard output pulses when the threshold is exceeded by noise emissions, determining the average frequency f=N/T of exceeding the threshold by noise emissions by accumulating the number N of output pulses over time the next k-th cycle T and increasing the threshold by ΔU=U _k +λf _k , where k is the sequence number of the cycle, λ is the transfer coefficient of the SHARP scheme, until the difference Δf=f _k -f _k-1 becomes less than the set value.

This procedure ensures that the frequency of noise exceeding the threshold is constant when the parameters of the noise source change, for example, due to changes in temperature and other external factors. However, this solution does not allow controlling the noise level at the input of the threshold device, which is sometimes necessary in the process of manufacturing systems that include threshold devices, for example, photodetectors for location and communication systems [9]. This input may be structurally unavailable in the operating mode, or connecting an external sensor to it is undesirable due to the distortion it introduces. In addition, standard measuring instruments [14-16] do not always provide the ability to measure low-level noise and require special laboratory conditions for operation. Thus, a millivoltmeter [16] is implemented in a laboratory device with dimensions (without an external probe) of 271x191x123 mm and a mass of 2.57 kg, operating in a narrow temperature range. Such equipment is unsuitable for embedding in miniature photodetectors, especially as part of portable communication systems, locations, etc.

The objective of the invention is to determine the rms value of fluctuation noise at the input of a threshold device as part of the equipment without connecting external means, including in the process of design, production and maintenance, as well as the application of the method for constructing autonomous rms noise meters.

, причем,

, где ξ << 1 - доверительный коэффициент, начальное значение порога U устанавливают как можно ближе к нулю, а коэффициент λ выбирают в пределах (1-3)/f₀, где f₀ - средняя частота пересечения шумом нулевого порога.This problem is solved by the fact that in the known method of noise automatic threshold control (SHARP), which consists in comparing the noise process with a threshold level U, generating standard output pulses when the threshold is exceeded by noise emissions, determining the average frequency f=N/T of exceeding the threshold by noise emissions by accumulation of the number N of output pulses during the next k-th cycle T and increase the threshold by the value ΔU=U _k +λf _k , where k is the serial number of the cycle, λ is the transfer coefficient of the SHARP scheme, until the difference Δf=f _k -f _k-1 will not become less than the specified value, the average frequency f ₀ of the noise process crossing the zero level is preliminarily determined, and after the transient process is established, the threshold value U and the frequency f=N/T of exceeding the threshold by noise emissions in the last stabilization cycle are determined, after which calculate the value of the standard deviation of the noise according to the formula

, moreover,

, where ξ << 1 is the confidence factor, the initial value of the threshold U is set as close to zero as possible, and the coefficient λ is chosen within (1-3)/f ₀ , where f ₀ is the average frequency of the noise crossing the zero threshold.

The operating principle of the device is based on the dependence of the average frequency f of crossing the threshold U by fluctuation noise emissions [7, 8, 12], which obeys a normal distribution.

- средняя частота пересечения шумом нулевого порога;Where

- average frequency of noise crossing the zero threshold;

R(τ) is the noise correlation function determined by the frequency response of the receiving-amplifying path.

The frequency f ₀ can be preliminarily determined at the threshold level U=0 by counting the number N ₀ of zero crossings by noise over time T' and calculating the frequency f ₀ using the formula f ₀ =N ₀ T. This parameter is constant for this device and does not require verification at each measurement σ.

From relation (1) follows the expression for calculating σ used in the present invention.

In FIG. 1 shows a structure that implements the method.

In FIG. 2 - SHARP scheme.

In FIG. 3 - transient process during analog accumulation of frequency f.

In FIG. 4 - equivalent circuit of the noise source (the input circuit of the photodetector with transimpedance preamplifier).

In FIG. Figure 5 shows the dependence of the frequency f on the relative threshold level U/σ.

In FIG. 6 is a graph characterizing the error in estimating a as a function of U/σ.

Stabilization of the threshold level using the method used allows you to accurately set the average frequency f of exceeding the threshold U by noise emissions, which is automatically set by adjusting the threshold using the SHARP scheme [4-8]. With known f and U, using expression (2), we can determine the root mean square value of the noise a.

The threshold device that implements the method (Fig. 1) includes an analog comparator 1 with a standard pulse shaper 2, the output of which is connected through the SHARP circuit 3 to the reference input of the analog comparator. The signal input of the latter receives a signal from the noise source. The analog output of the SHARP circuit is also connected to an analog-to-digital converter 4 associated with the decision device 5, the other input of which receives digital information from the SHARP circuit about the measured frequency of noise emissions f and a signal about the end of the transient process. The SHARP circuit 3 (Fig. 2) contains an accumulator 6 connected to a processing unit 7 connected to a threshold control circuit 8 and to a decision device 5.

At the initial low threshold U, the average frequency of exceeding the threshold by noise emissions f is close to the maximum frequency f ₀ . The output pulses of the shaper 2, in addition to the terminal device 5, are fed to the input of the accumulator 6, in which, during the accumulation time T, a signal is generated that is proportional to the number N of pulses accumulated during this time. This signal in the processing unit 7 is compared with the reference value N _op , and the difference through the threshold control circuit 8 increases the threshold U by a proportional value. As a result, the frequency of noise emissions is reduced, and the described cycle is repeated until the difference N _op - N is less than the set value, and the processing unit stops the stabilization process. A signal about this from the processing unit is fed to the input of the decision device. Simultaneously steady value U through the analog-to-digital Converter 4 also enters the decision device, which stores the values of f ₀ and N _op . Based on these data, the solver calculates the value of o according to formula (2) and outputs the calculation result to output 1. Other information can be output to output 2, for example, about the delay of the input signal.

The stabilization process is illustrated by the graphs of Fig. 3. Graph 9 is the process of generating the output pulses of the shaper 2. Curve 10 is proportional to the decreasing frequency f. Graph 11 shows the course of the threshold U from the initial value of 40 units. to a steady level of 70 sr.u. Due to the random nature of the generation of noise emissions obeying the Poisson law [13], curves 10 and 11 are distorted by these fluctuations. The longer the averaging time T, the smaller the deviation of the threshold U and the average frequency f from the average value and the more accurate the result of determining a.

The characteristics of the studied noise source are considered in the special literature [10-12]. In FIG. Figure 4 shows the equivalent noise circuit of the receiving circuit of a photodetector based on an avalanche photodiode with a transimpedance preamplifier [10]. At the input of the amplifier, along with the signal photocurrent I _c, there are intrinsic current noises of the photodiode I _t and the equivalent noise current of the preamplifier I _y , as well as the noise voltage of the preamplifier E _y . The thermal noise of the load resistance R and feedback R _oc are brought to the source I _t . The level a is affected by the frequency response of the receiving-amplifying path, including the parasitic input capacitance C and the capacitance in the feedback circuit C _os. Due to the large number of listed and not taken into account noise sources, when designing the receiving path, it is important to know their total effect, which makes the task of the present invention all the more relevant since it is not always possible to measure the noise level using universal measuring equipment.

In FIG. 5 shows a graph of dependence (1) for f ₀ =10 ⁷ 1/s. The choice of the reference frequency f _op and, accordingly, the number of threshold crossings N _op =f _op T during the accumulation time T is due to several considerations. On the one hand, the higher the frequency f _op , the faster the stabilization mode is reached (Fig. 3). On the other hand, if the frequency f _op is too high, close to f ₀ , the requirements for the speed of the analog comparator 1 and the standard pulse shaper 2 increase. The existing element base imposes certain restrictions on this choice [17, 18]. The requirements for measurement accuracy a determine both the choice of the reference frequency f _op and the choice of the accumulation period T. FIG. Figure 6 shows a graph of the estimate σ* versus the threshold/noise parameter U/σ, which in turn is related to the frequency f _op (Fig. 5). The dependence of Fig. 6 obtained for the true value of σ=1 and the error in determining the value of f/f ₀ equal to 0.1%. It can be seen that the accuracy of the estimate σ* increases with the growth of the threshold/noise ratio. With the value of this parameter U/σ>1, the error in estimating σ does not exceed 0.1%. In this case, the frequency f _op must meet the condition

The criterion for choosing the accumulation period T is the known ratio for the standard deviation on the accumulated amount N [13]

откудаThe requirement σ _n = where ξ << 1 is the confidence factor, is equivalent to the condition

where

The accumulation period T is determined by these parameters

Example 1

f ₀ =10 ⁷ 1/c; σ=1; ξ=0.003.

It can be seen from expression (2) that the relative error in determining U is equal to the given error in calculating σ. This determines the requirements for the choice of capacity and stability of the analog-to-digital converter 4.

Each k-th accumulation cycle starts from the reference level U _k =U _k-1 +λ⋅10 ^-7 f _k-1 , where U _k-1 is the initial threshold of the previous cycle; f _k-1 =N _k-1 /T - emission frequency measured in the previous cycle; λ is the transfer coefficient of the SHARP scheme.

The time for the system to reach the specified accuracy mode depends on the transfer coefficient of the SHARP 3 scheme and the initial setting of the threshold U. Tables 1-6 show the options for the transient process for different values of the specified parameters with the initial data of example 1.

In the considered options, the SHARP enters the mode of the given average frequency f _op =3⋅10 ⁶ 1/s for 4-5 accumulation cycles. Thus, the time T _meas measurement σ is (4-5)T. In the conditions of example 1, this corresponds to T _meas = (0.4-0.5) s.

The proposed technical solution can be built into the standard structure of the photodetector and allows you to measure the noise intensity at the output of the linear path in the working composition of the device during its operation.

Thus, a minimum error (0.1-1)% of the o meter is ensured, sufficient for the operational assessment of the noise level of receiving-amplifying devices in a wide temperature range, both in stand-alone version and as part of functional complexes.

Analog instrumental procedures do not contain non-linear functional transformations, the implementation of which requires complex devices with reliable reproducibility and stability, as well as complex metrological support.

Devices that implement the method do not require temperature stabilization and have a minimum time to reach the operating mode. With an acceptable error in determining a within (1-2)%, the measurement time is several milliseconds.

Thus, the method provides a solution to the problem - prompt determination of the root-mean-square value of the fluctuation normal noise of photodetectors during the production and maintenance of miniature, including wearable and embedded equipment, with simple metrological support in a wide operational range.

Information sources

1 Reception of impulse signals in the presence of noise Collection of translated articles, ed. A.E. Basharinov and M.S. Alexandrova. - M.: State. energy publishing house, 1960. - 384 p.

2 Ya.D. Shirman, V.N. Golikov. Fundamentals of the theory of detection of radar signals and measurement of their parameters. - Moscow: Sov. radio, 1963. - 278 p.

3 Gower J. Optical communication systems: TRANS. from English. - M: Radio and communication, 1989. - 504 p.

4 US Pat. No. 3516751. Optical radiation pulse control receiver. 1970.

5 A.C. No. 320067 Device for stabilizing the average frequency of noise emissions above the threshold level. 1972.

6 Pat. RF No. 2755601. Method for detecting optical signals. 2021.

7 RF Patent No. 2732004. Pulse threshold device with noise threshold stabilization. 2019.

8 Vilner V.G. Design of threshold devices with noise threshold stabilization. // Optical-mechanical industry. - 1984 - No. 5, - S. 39-41. - prototype.

9 RF Patent No. 2759262. The receiver of the laser rangefinder.

10 Paul Horovinz, Winfield Hill. The art of electronics. third edition. Cambridge University Press, 2015. - p. 539.

11 I.D. Anisimova and others. Ed. IN AND. Stafeeva. Semiconductor photodetectors. Ultraviolet, visible and near infrared ranges of the spectrum. - M: Radio and communication, 1984 - 216 p.

12 Filachev A.M., Taubkin I.I., Trishenkov M.A. Solid state photoelectronics. Physical bases. Moscow, Fizmatgiz. 2007, - p. 345.

13 Tikhonov V.I. Emissions of random processes. Ch. ed. Phys.-Math. lit., 1970, - p. 392.

14 Nasonov B.C. Handbook of radio measuring instruments. - M.: Soviet radio, 1976, vol. 1., 234 p.

15 F.V. Kushnir. Electroradio measurements.: Uch. allowance for universities. - L., Energoatomizdat. Leningrad. department, 1983 - 320 p.

16 D. Molokov. High frequency RMS millivoltmeter. Radio magazine. No. 5, 2019 - S. 14-21 - prototype.

17 High speed comparator TL714C. Texas Instruments. Datasheet 2003.

18 High speed comparator MAX941/MAX942/MAX944. MAXIM. Datasheet 2004.

Claims (1)

The method of noise automatic threshold adjustment (SHARP), which consists in comparing the noise process with the threshold level U, generating standard output pulses when the threshold is exceeded by noise emissions, determining the average frequency f=N/T of exceeding the threshold by noise emissions by accumulating the number N of output pulses during the next k-th cycle T and increase the threshold by ΔU=U _k +λf _k , where k is the sequence number of the cycle, λ is the transfer coefficient of the SHARP scheme, until the difference Δf=f _k -f _k-1 becomes less a given value, characterized in that the average frequency f ₀ of the noise process crossing the zero level is preliminarily determined, and after the transient process is established, the threshold value U and the frequency f=N/T of exceeding the threshold by noise emissions in the last stabilization cycle are determined, after which the value of the root-mean-square deviation is calculated noise by formula

and

, where ξ << 1 is the confidence factor, the initial value of the threshold U is set as close to zero as possible, and the coefficient λ is chosen within (1-3)/f ₀ , where f ₀ is the average frequency of the noise crossing the zero threshold.

Images