JPH08152329A - Signal detection system for optical fiber gyro - Google Patents

Abstract

PURPOSE: To provide a signal detection system for optical fiber gyro in which a reference signal is taken out from the signal of a phase modulator, two waves are discriminated definitely by the reference signal during one period of phase modulation, and the duration can be measured. CONSTITUTION: A phase modulator 7 is provided at one end of a fiber coil 6 and a rectangular wave clock 8 for driving the phase modulator 7 is also provided. Clockwise and counterclockwise lights in the coil 6 are coupled at a coupler 5 and the intensity of interfering light is detected by a light receiving element 10. The intensity thus detected is fed to an amplifier 11 and the DC component is removed through a capacitor 12 before being converted into a rectangular wave through a comparator 13. The rectangular wave thus converted is fed to the CLK input of a DFF 14 and the clock 8 is sustained at the input D. The output Q subjected to 1/2 frequency division has such waveform as the preceding wave F corresponds to 1 and the following wave corresponds to 0. A counting clock 16 is fed to the CLK inputs of counters 17, 18 while the output Q and the inverted output from an inverter 19 are fed to the EnT input in order to measure the length of the preceding and following waves. A subtractor 20 determines the difference of the length thus determining the rotational angular speed of the coil 6.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal detection system for an optical fiber gyro. In particular, the present invention relates to a signal detection method capable of accurately discriminating asymmetrical durations F and R of an AC component of a light receiving element having a repetition frequency approximately twice that of a phase modulation signal in an optical fiber gyro of a phase modulation method.

An optical fiber gyro is a coil in which a single mode fiber is wound many times, and a monochromatic light is rotated clockwise.
The rotational angular velocity is measured by utilizing the fact that a phase difference proportional to the rotational angular velocity of the coil is generated when passing as counterclockwise light. A light emitting element, a fiber coil, a light receiving element, an optical branching / merging element (beam splitter, fiber coupler), etc. are essential. In addition to this, a polarizer and a depolarizer are provided. The light-emitting element generates monochromatic or quasi-monochromatic light and propagates it in the fiber coil as clockwise light or counterclockwise light. The light receiving element detects the interference light. The photocurrent of the light receiving element is processed to obtain the rotational angular velocity.

[0003]

2. Description of the Related Art A phase modulation system is a phase modulator provided near one end of a fiber coil to periodically modulate the phase of light propagating through the fiber coil and perform synchronous detection from a signal of a light receiving element to obtain a fundamental wave. , Harmonics, etc. are obtained to obtain the rotation speed. In the phase modulation type optical fiber gyro, the following signal I appears in the light receiving element due to the interference of the left-handed light and the right-handed light.

[0004]

[Equation 1]

Here, E ₁ and E ₂ are the amplitudes of right-handed light and left-handed light. Δθ is the phase difference between the left-handed light and the right-handed light generated by the rotation of the coil. This is the target of detection. J _n (ξ) is the n-th order Bessel function, and Ω is the phase modulation frequency. FIG. 2 shows some of the Bessel functions. Ξ contained in the Bessel function is the degree of phase modulation. Ξ = 2bsi, where b is the amplitude of phase modulation in the phase modulator, L is the fiber length, n is the refractive index, and c is the speed of light.
given by n (ΩL / nc).

The first term of the interference signal is a DC component.
The term enclosed by cos Δθ in the second term is all of the DC component and the even harmonics. The part enclosed by sin Δθ in the third term is all the odd harmonics. The rotational angular velocity is obtained from the third term sin Δθ. This is the primary,
Includes all odd harmonics such as third, fifth ... In practice, Δθ is often obtained using the first-order (fundamental wave) term. This is obtained by synchronously detecting the output of the light receiving element with a carrier signal equal to the frequency of phase modulation. In this case, ξ is set to 1.8 so that the value of the first-order Bessel function becomes large.

Alternatively, ξ may be fixed to the zero point of the second-order Bessel function (ξ = 5.2) by controlling the second harmonic to be zero. This makes it possible to eliminate fluctuations in the phase modulation amplitude. Furthermore, a device has been proposed in which the fundamental wave is divided by the fourth harmonic to remove the influence of the fluctuation of the light quantity. Various other improvements have been proposed. In all of them, in the phase modulation method, the output is divided into infinite number of partial waves having a Bessel function as a coefficient, and Δθ is obtained by focusing on one or two or at most three of the partial waves. The method is partial. Only partial waves are extracted by synchronous detection. Let's call these methods collectively the partial wave decomposition method.

A method in which a modulated signal is included in an input and given to a sensor system and an output signal is synchronously detected by the frequency of the modulated signal is widely used not only in optical fibers but also in measuring instruments. This is a well-known method. There is a track record. In the partial wave decomposition method, it is necessary to first match the phase of the carrier signal used for synchronous detection. The output is a voltage. The result will be erroneous due to the carrier phase shift.
The deviation from the set value of ξ causes the result to be incorrect. Further, since the voltage is the output, its accuracy is also limited.

The present inventors have devised a comprehensive detection method that is completely different from partial wave decomposition. It is a method based on a comprehensive method. The lengths of two waves of the alternating signal appearing twice in one period of modulation are measured. Letting T be the period of modulation, the AC part has two waves between them. Neither is a regular sine wave. When Δθ = 0, the wave lengths are equal. Δθ
When is not 0, the two waves have different lengths. The length Ta of the front wave F and the length Tb of the rear wave R are measured, and Δθ is known from the difference. That is, Δθ
Is to seek. Let's call it the AC time difference method.

This does not target the partial waves, but covers the entire AC signal. That is, in contrast to the conventional method that calculates Δθ using only n = 1 as a clue among all outputs, the AC time difference method uses all the components of n = 1 to ∞ equally. Comprehensive and easy to analyze. The AC time difference method does not require a synchronous detection circuit. No preconditioning is required to create the carrier signal and adjust the phase of this and the signal. The output is time. Not voltage. Its accuracy is high because it measures time. It is not affected by fluctuations in the light amount of the light emitting element. In this way, there are many advantages in principle. To extract the AC signal, for example, just insert one capacitor. Besides, it is possible to extract only the AC component also by the circuit shown in FIG.
Here, the output 31 of the light receiving element is put into a low-pass filter 32 to obtain a direct current component, which is input to a comparator 33 together with the original signal 31 to obtain the difference between the two. Difference signal 34
Becomes a signal proportional to the AC component contained in the original light receiving element signal. There is another method for obtaining the AC component, but this is not the subject of the present invention and will not be described further.

The principle of the optical fiber gyro of the present invention is based on a completely new technical idea. It may be difficult for those who are used to the partial wave decomposition method to understand. Then, first, the method of AC time difference method and the underlying principle are explained intuitively. In the AC time difference method, the DC component D is removed and only the AC signal (W = ID) is processed before processing. DC component D

D = (E ₁ ² + E ₂ ² ) / 2 + E ₁ E ₂ cos ΔθJ ₀ (ξ) (2) AC component W is

[0013]

(Equation 3)

It can be expressed as This is a perfect expression. However, the AC time difference method does not process by such an expansion formula. Think of the whole as one. The expansion formula I based on the above Bessel function coefficient is a direct current component, si
It is composed of odd-order harmonic components proportional to nΔθ and even-order harmonic components proportional to cosΔθ.

If the coil is stationary. In this case, Δθ = 0, so all odd harmonics are zero. Only DC component and even harmonics. This AC signal W then contains only even harmonics. That is 2
Only 4x, 6x, ... Harmonics are included. These have a common divisor of 2. These repeatedly change the period T of the phase modulation into two, four, and six. W is the sum of these. Therefore, the repetition frequency is, of course, twice the phase modulation frequency Ω. The period is half of the phase modulation (T / 2). That is, one phase modulation cycle includes two waves. For convenience of description, these are tentatively referred to as a front wave F and a rear wave R. The front wave F and the rear wave R are completely equivalent waves and cannot be distinguished in shape. Thus, when Δθ = 0, the AC component W has a repetition frequency of 2Ω.
There is no distinction between the two waves included in one cycle T.

FIG. 3 shows such an AC signal. The waveform at rest is drawn by the solid line. T ₁ , T
The part ₂ is the front wave F. T ₃ and T ₄ are rear waves R.
These lengths are naturally equal. T ₁ + T ₂ = T ₃ + T
₄ . However, this is not a sine wave because it is the sum of all even harmonics. The 2nd harmonic, the 4th harmonic, etc. are all sine waves, but their sum cannot be a sine wave. It is a vertically asymmetric wave.

It is assumed that the coil starts rotating. Δθ
Increases from 0 little by little. Even-order harmonics gradually decrease in proportion to cos Δθ. But this is small.
On the contrary, in proportion to sin Δθ, the odd harmonics increase little by little as the rotation starts. This changes repeatedly 1, 3, 5, ... In one cycle T of the phase modulation. These have no common divisors other than one. Therefore, the odd harmonics give different contributions to the front wave F and the rear wave R when superposed on a signal having said doubled frequency. That is, the front wave F and the rear wave R are slightly different from each other. Since cos Δθ is larger, the property of having two waves is maintained. But,
One becomes longer and the other becomes shorter. Of course, the sum of the lengths of the two waves is constant because it is equal to the period T of phase modulation.

Due to the mixing of odd harmonics, the front wave F and the rear wave R are not equal. It becomes a front-back asymmetric wave. The waveform is indicated by the broken line in FIG. In this example, T ₁ grows,
T ₃ is getting shorter. The AC component W loses the periodicity of 1 / 2Ω due to the addition of the odd harmonics. The repetition frequency is Ω itself. However, the front wave F and the rear wave R can still be distinguished. Time F of front wave F
= (T ₁ + T ₂ ) and the time R of the after wave R = (T ₃ + T ₄ ).
However, since there is a discrepancy in proportion to Δθ, Δθ can be obtained from the difference between these times.

That being said, the four partial times T ₁ , T
₂ , T ₃ and T ₄ do not change in a chaotic manner. Actually, T ₂ and T ₄ are almost constant even if they are rotated. Moreover, T ₂ = T ₄ always holds. This can be proved by considering symmetry. But I won't prove it here. The lower half-wave does not work for the difference calculation. Eventually, it is the upper half-wave T ₁ and T ₃ that change.
In this case, the sum becomes constant, and the difference increases depending on the rotational angular velocity. The direction of rotation is known by the sign of the difference.

Such an AC time difference method is proposed by Japanese Patent Application No. 6-13145 by the present inventor. The sensitivity Se is defined as a value obtained by dividing a time difference (F−R) between the front wave F and the back wave R by the time sum (F + R), and dividing by Δθ.

Se = (F−R) / {Δθ (F + R)} (4) This is Se = (2 / π) cosecC (5) in the first-order approximation. However, C is cos ⁻¹ {ξ ⁻¹ cos ⁻¹ J ₀
It is a value defined as the main value of (ξ)}.

Thus, by the difference (F−R) between the durations of two waves included in one cycle of phase modulation of the AC signal,
You can know Δθ. Where F = T ₁ + T ₂ , R =
It is T ₃ + T ₄ . (F−R) is eventually (T ₁ −T ₃ ), but here (T ₁ + T ₂ ) and (T ₃ + T ₄ ).
And calculate the sum. Assuming that the rotation of the coil is positive when (FR) is positive, the rotation of the coil is reverse when (FR) is negative. The sign of the difference determines the direction of rotation. This has already been explained.

[0023]

As shown in FIG. 3, the front wave F of the apparent two-cycle portion included in one cycle of phase modulation is obtained.
Of the back wave R (T ₃ + T ₄ ) and the length of (T ₁ + T ₂ ).
Must be measured. This is done as shown in FIG.
In FIG. 4, (1) is the original two waveforms. It is converted into a binary signal in which 1 and 0 alternate when crossing the zero point. This is a binarization process. This can easily be done by amplifying and saturating the signal with a high gain amplifier.

That is, binarization can be performed by a comparator having a threshold value of 0. Let the zeros of the first wave be a, b, h, d, h. Lee, Ha, and Ho cross the zero point from negative to positive. Ro and Ni cross the zero from positive to negative. The binarization of this is (2) in FIG. 1 between Iro,
0 between Loha, 1 between Hani and 0 between Niho. What I want to know is the length of time between I and H.

Then, the frequency is divided by 1/2. The frequency division can be performed by a flip-flop, for example. Using a flip-flop that counts the rising edges of pulses, the result is shown in Fig. 4 (3).
The divided signal can be obtained as follows. Then, it takes a value of 1 in I to C. This is the duration of the front wave F (T ₁ +
Corresponds to T ₂ ). It takes a value of 0 in Ha-e. This corresponds to the duration of the trailing wave R (T ₃ + T ₄ ). That is, the length of the front wave F depends on the time of 1 and the rear wave R depends on the time of 0.
You can know the length of.

When the length of either one of the time of 1 and the time of 0 is calculated, the other time can be known. This is because the sum is equal to one period T of phase modulation. Of course, both may be requested independently. However, it is difficult to distinguish the front wave F and the rear wave R. There is no distinction in the signal itself after binarization. T ₁
And T ₃ and T ₂ and T ₄ must be distinguished. That is, the front wave F must correspond to 1 and the rear wave R must correspond to 0. However, the front wave F and the rear wave R can be distinguished only by specifying the timing at which the frequency division is started. This is possible. However, even if it is initially set, it may be inverted due to noise. Then, the front wave F and the rear wave R are recognized oppositely. In this case, the rotation direction cannot be correctly recognized.

Let us explain with reference to FIG. (1) is the original AC signal waveform. (2) is a binarization of this. Further, the frequency is divided, but the timing of starting the frequency division is important. (3) is a correct frequency division. A to C of the front wave F are associated with 1. This is because the division started at G. G is any one time point included in the rear wave R.

However, there is a possibility of frequency division as in (4). This starts the division at the Chi point. The chi point is an arbitrary time of the front wave F. Front wave F and back wave R
Alternates every cycle of phase modulation. The values of "1" and "0" thereafter are completely opposite depending on whether the frequency division start timing is in the front wave F or the rear wave R. If this is reversed, the sign of the time difference will be reversed, and the direction of rotation will be erroneously recognized. Even though it rotates clockwise, the measurement result is that it is counterclockwise. This should not be. The correspondence must be made as in (3).
Dividing as in (4) must be prohibited. That is, the frequency division must always be started in the rear wave R as in (3). Some kind of ingenuity is needed to do so.

There is another problem. Even if the frequency division is started from the rear wave R and the front wave F is associated with 1 and the rear wave R is associated with 0 as in (3), once noise enters, such a state cannot be maintained. . (5) in FIG. 5 is the signal after binarization. Short noise in the rear wave R
Let's say Nu entered. The frequency divider counts the next rising edge. Here, the pulse of the front wave F ends. The front wave F becomes li. After that, it is "0". This represents the back wave R.
At the next rising c, the output of the frequency dividing circuit becomes "1". The time of the back wave R is recognized as re-ha. This is recognized as the front wave F because the next ha-e becomes "1".

However, in practice, the frequency should be divided as shown in FIG. After the noise noise, the recognition of the front wave F and the rear wave R is wrong. In this way, each time the noise enters, the recognition of the front wave F and the rear wave R is reversed, and the rotation direction of the coil is erroneously recognized. As described above, since there is no certainty in associating the front wave F and the rear wave R with the signals after frequency division,
It is an object of the present invention to solve this. That is, it is an object of the present invention to provide an apparatus capable of accurately distinguishing a front wave F and a rear wave R included in one cycle of phase modulation and measuring their durations F and R. Distinguishing the front wave F from the rear wave R means distinguishing T ₁ from T ₃ . In other words, the present invention can be rephrased as having the purpose of distinguishing T ₁ from T ₃ and accurately measuring the lengths of the front wave F and the rear wave R.

[0031]

According to the present invention, a reference signal S is prepared, T ₁ and T ₃ are distinguished from the phase difference between the interference AC signal W and the reference signal S, and frequency division is performed from the correct time. Let's get started. This reference signal gives the timing of frequency division. This may have a fixed relationship with the phase of the phase modulation signal H. Immediately after the sign of the reference signal S is changed, the positive half wave of the AC signal W is changed to T ₁ or T
_You decide to be one of the _three .

For example, the first positive half wave after the sign change of the reference signal S is determined as T ₁ . The reference signal is determined by reflecting the phase of the phase modulator. The wave of the phase modulator and the front wave F and the rear wave R in the AC component W have a constant relation. A positive / negative change of the phase modulator signal should appear in the AC component of the output of the light receiving element after the optical delay time and the electrical delay time have passed, and this is inevitably the front wave F in one cycle. It is the back wave R. Therefore, the phase relationship between the phase modulator signal and the front wave F and the rear wave R should be different by the delay time. Since the delay time is always constant, there should be a one-to-one correspondence between changes in the phase modulator signal and changes in the front wave F and the rear wave R.

Therefore, in the present invention, the reference signal is obtained from the signal of the phase modulator, the instant of change of the reference signal is detected, and T ₁
And T ₃ are distinguished. In this example, after the reference signal changes from negative to positive, the first positive half wave is determined as T ₁ (T ₃
You may decide). Then, the positive half-wave always having a constant delay time with respect to the phase modulator signal can be determined as T _1, and the other half-waves can be determined as T ₃ . Even if there is noise, you can immediately return to the correct response.

[0034]

The two waves of the AC signal component contained in one cycle T of the phase modulator cannot be known as the front wave F or the rear wave R by themselves. This is originally determined in relation to the phase of phase modulation. In the present invention, the reference signal S is generated from the phase modulation signal H, and the positive half wave immediately after the reference signal changes from negative to positive is determined as T ₁ . As a result, it is possible to determine a half wave that is always in a fixed time relationship with the phase modulation signal. The reference signal S has only to have the same frequency as the phase modulation signal H and have a constant phase delay (or advance). The time when the reference signal S changes from positive to negative or from negative to positive can be used as a timing signal for determining the front and rear waves.

Let the phase modulation signal H be sinΩt. The reference signal S can be written as sin (Ωt + φ). The AC signal W (= ID) included in the light receiving element is given by (3), and the origin of this time includes the delay time τ for phase modulation,

[0036]

(Equation 6)

The expansion formula of the Bessel function is related to cos. cos θ is 0 when θ is 3π / 2, and thereafter becomes positive and increases. Therefore, the distinction between the front wave F and the rear wave R in FIGS. 3 and 4 is actually 3π / 2 or −π /
It corresponds to 2. Therefore, the front wave F here is 2Ω when Δθ = 0 in FIGS.
(T + τ) is −π / 2 + 4πm to 3π / 2 + 4πm
(M is an arbitrary integer), and the back wave R is 2Ω (t
+ Τ) is 3π / 2 + 4πm to 7π / 2 + 4πm.
When Δθ ≠ 0, the range gradually changes. However, they are almost in the same range. Therefore, assuming Δθ = 0, the time ranges of the front wave F and the rear wave R are

Front wave F 2πm−π / 4−Ωτ ≦ Ωt ≦ 2πm + 3π / 4−Ωτ (7) Rear wave R 2πm + 3π / 4−Ωτ ≦ Ωt ≦ 2πm + 7π / 4−Ωτ (8) Of course, the origin of time is given by the phase modulator. The signal of the phase modulator H is sinΩt, and the reference signal S from which the time reference is taken is sin (Ωt + φ). The zero point of this is Ωt
= ΠM−φ (M is an integer). There are two zeros in one cycle. However, the zero that changes from negative to positive is Ωt =
Only 2πM-φ. The zero that changes from positive to negative is Ωt
= 2πM + π−φ. Either may be used as a reference.

Here, a rising zero point that changes from negative to positive is adopted. Then, the difference K from the beginning of the front wave F is K = (2πm−π / 4Ωτ) − (2πM−φ) = φ−π / 4−Ωτ + 2π (m−M) (9). The difference L between the start of the rear wave R and the rising zero point is L = (2πm + 3π / 4−Ωτ) − (2πM−φ) = φ + 3π / 4−Ωτ + 2π (m−M) (10)

It becomes The difference between K and L is always π,
Since it is free to increase or decrease 2π depending on how to select m or M, both can be set to 0 to 2π. Those of 0 and 2π of K and L will be referred to as main values of K and L. It depends on the values of the delay phase φ and Ωτ, but since this is constant, if the main value of K is 0 to π, the main value of L is π.
~ 2π.

That is the principle of the present invention. Since φ and Ωτ are invariant, the positive half wave first appearing from the zero point at which the reference signal S = sin (Ωt + φ) made from the phase modulation signal H = sin Ωt changes from negative to positive is the positive half of the front wave F. Either the wave or the half wave of the rear wave R is determined. Here, the determination of the first positive half wave as T ₁ means that 0 <K <π, π <
It is assumed that L <2π. The first half wave T ₃
Can be defined as π <K <2π, 0 <
It is assumed that L <π. In fact, this is all right. The front wave F and the rear wave R are distinguished between positive rotation and negative rotation. Whether the positive half wave immediately after the rising zero point of the reference signal is T ₁ or T ₃ , this means that the actual rotation direction should be examined and associated. It is important that always remains unchanged a T ₁ if T _1.

There is another thing to note. Δθ =
In the case of 0, the boundary between the front wave F and the rear wave R is -π / 2 and 3π /
In 4. However, in the case of Δθ ≠ 0, the boundary between the front wave F and the rear wave R is slightly deviated from -π / 2 or 3π / 4. This is fine, though. There is a time margin because it is not the zero point of the reference signal itself but the positive half-wave that first appears after the rising zero point (from negative to positive). The phase angle has a maximum of 180 degrees. Therefore, according to Δθ ≠ 0,
A slight deviation of the boundaries does not influence the determination of the positive half wave T ₁ of the front wave F.

[0043]

FIG. 1 shows an embodiment of the present invention. The monochromatic and quasi-monochromatic light emitted from the light source 1 is emitted by the single mode fiber 2
Incident on the edge of. This goes through the coupler 3 and the fiber 4
To the coupler 5. Here, the light is split into two. One rotates counterclockwise in the fiber coil 6, and the other rotates clockwise in the same fiber coil 6. The fiber coil is formed by winding a single mode fiber many times. The purpose is to determine the rotational angular velocity of this.

The phase modulator 7 is provided at one end of the fiber coil 6. This is, for example, one in which a fiber is wound around a piezoelectric vibrator. When the piezoelectric vibrator is expanded and contracted by an AC voltage, the phase of light passing therethrough changes periodically. Phase modulation is represented by bsinΩt. b is the phase amplitude, Ω
Is the modulation angular frequency. One period T of modulation is T = 2π / Ω
Is. A clock 8 is provided to drive the phase modulator 7. This produces a rectangular modulated signal of sin (Ωt + φ). This is given between the electrodes of the phase modulator through the low-pass filter 9. Rectangularization sin (Ωt
Since + φ) includes infinite harmonics of Ω, the low-pass filter 9 is provided to extract the lowest sine wave from this. This gives a sine wave sin (Ωt + φ)
Is obtained.

A sine wave is applied to the phase modulator. Due to the slight delay, the phase modulator oscillates sin Ωt. Here, φ is a constant obtained by multiplying the delay time of the signal from the clock to the phase modulator by Ω. The clockwise light and the counterclockwise light are united by the coupler 5. This is a light receiving element (PD)
8 interferes and the intensity of the interference light is detected. The photocurrent is amplified by the amplifier 11. The voltage signal I includes a DC component D and an AC component W. The DC component D is removed by the capacitor 12. AC portion W remains. Since it is an exchange, the time average is 0.

Further, the AC component W is guided to the comparator 13. This compares 0 V with the AC signal W. W
If> 0, it is converted into a rectangular wave corresponding to “1”, and if W <0, it is converted into a rectangular wave corresponding to “0”. (1) to (2) of FIG. 4 and FIG.
The process from (1) to (2). This is D (data
The clock input CLK of the flip-flop 14 is entered. The D flip-flop is for dividing the rectangular wave by 1/2. This corresponds to the signal processing from (2) to (3) in FIG. 4 or from (2) to (3) in FIG. What is divided by 1/2 appears from the output Q. 4 (3) and FIG.
A waveform in which the front wave F corresponds to 1 and the rear wave R corresponds to 0 as in (3) can be obtained. The clear terminal CLR and the preset terminal PR of the D flip-flop 14 are the power supply voltage +5.
It has been raised to V.

What is important here is that the clock 8 for driving the phase modulator 7 is input to the D input of the flip-flop by the connection line 15. The D flip-flop has a function of outputting the value of the D input to Q at the moment when a pulse is input to the clock terminal CLK. C
The binary AC signal W is input from the LK terminal, but when D = 0, the output Q becomes 0. The output Q can be 1 only when D = 1.

The phase modulator clock 8 is sin (Ωt +
φ) generates a rectangular signal. So the D input is
The half cycle K of −φ + 2mπ ≦ Ωt ≦ −φ + π + 2mπ is “
1 ". -Φ + π + 2 mπ≤Ωt≤-φ + 2π + 2
The half cycle L of mπ becomes “0”. The delay times φ / Ω and τ are adjusted so that T ₁ of the front wave F in FIGS. 4 and 5 falls within the half cycle K. T _{3 of the} back wave R enters another half cycle L.
Then, of the positive half-waves T ₁ and T ₃ of the AC signal W, the output Q of the flip-flop is generated only when T ₁ is input to CLK.
Rises to "1".

When T ₃ enters CLK, D = 0, so the output Q remains 0. Therefore, the output Q of the flip-flop is always only when the positive half-wave T ₁ of the front wave F rises.
Stands up. At the next positive half-wave T ₃ , the output Q drops to 0 because D is 0. That is, the output Q rises at the edge of T ₁ and falls at the edge of T ₃ .
Therefore, all the problems described in FIG. 5 are solved. This will be described with reference to FIG. This is a problem of the timing of the start of frequency division.

In FIG. 6, (1) is an AC binary signal. The zero points in one phase of the phase modulation are a, b, c, d, and e. (2) is a reference signal taken from the phase modulator, which is input to the D input of the flip-flop. Make sure that the rising edge comes just before T ₁ . That is, the reference signal rises during T ₄ or T ₃ . Here we are standing up at T ₄ .

At the rising edge of the positive half wave T ₁ of the front wave F, D = 1. However, at the rising of the positive half wave of the rear wave R, D = 0 is always satisfied. The divided output Q is
As shown in (3), Q in Lee point is the edge of T ₁ = 1
Go up to. Then, at point C, which is the edge of T ₃ , Q = 0. (3) starts dividing at T ₄ .
(4) starts the frequency division at T ₁ . Even in this case, Q does not rise because D = 0 at point C.
Q rises at point a in the next cycle. It is possible to perform selection such that Q = 1 for “a” to “ha” and Q = 0 for “a” to “ho”.

Therefore, it is possible to obtain a frequency division output that rises exactly at T ₁ (a) and falls at T ₃ (c) regardless of the start timing of the division. FIG. 7 is a waveform diagram for explaining the effect of the present invention. It has been explained that the front wave F and the rear wave R are not confused with each other even if noise enters in the middle. FIG. 7A shows the waveform of the AC component when there is no noise. (2) is a reference signal, which is a D input.
(3) is the frequency division output when there is no noise. It rises at the edge of T ₁ and rises at the edge of T ₃ .

(4) is a waveform diagram of an AC signal containing noise. There is a short noise, Linu, between B and C. (5) is a waveform diagram showing a result of dividing a signal including noise by the device of the present invention. The output Q that has risen in B falls at a noise point. It has fallen earlier than the point C. There is an error between re-ha.
However, it does not stand up at the point. This is because D = 0. Therefore, the relationship between the front wave F and the rear wave R is not reversed here.
In the next cycle, the output Q rises at point a. After that, it will function normally. Even if there is noise, the effect stops within that cycle.

Returning to the description of FIG. Counting clock 1
Reference numeral 6 is a clock for measuring time. A clock with a frequency much higher than the frequency of phase modulation is used. The counting clock 16 is the CL of the two counters 17 and 18.
Enter K input. As a counter, for example, TC74HC16
1 can be used. These counters are
When the bull terminal EnT is 0, it is not counted. Pulses are counted when EnT is 1. The frequency-divided output Q described above is input to the EnT of one counter 17. This counts pulses when Q is one. That is, the length of the front wave F (T ₁ + T ₂ ) is measured.

The frequency-divided output Q is supplied to the inverter 19 (for example, T
C74HC04) inverts and puts this in the EnT of another counter 18. This counts the pulses when Q = 0. That is, this measures the length of the back wave R (T ₃ + T ₄ ). The subtractor 20 calculates the difference F−R = (T ₁ + T
₂₎ - to calculate the (T ₃ + T _4). From this, the rotational angular velocity of the fiber coil is obtained. The relationship between the time difference ΔT and Δθ has already been described. The relationship between the phase difference Δθ and the angular velocity Ωc is s
Determined by the agnac effect, Δθ = 4πLaΩc / λ
c. L is the fiber length of the fiber coil, a is the radius of the coil, λ is the wavelength of light, and c is the speed of light.

FIG. 8 is a block diagram showing the second embodiment of the present invention. It uses an integrator circuit to measure the length of time. The same applies up to the point where the flip-flop 14 generates a rectangular wave in which the front wave F corresponds to 1 and the rear wave R corresponds to 0. The output Q of the flip-flop is a resistor 28
And leads to the integrating circuit 27 composed of the capacitor 29. The voltage of the capacitor increases with time. This produces a voltage proportional to the length of the front wave F (T ₁ + T ₂ ). The length (T ₁ + T ₂ ) of the front wave F is known from the voltage. (T ₃ + T ₄ )
Is not measured. Since the sum of the lengths of the front wave F and the rear wave R is equal to one period T of the phase modulator (T = F + R), when F is known, R is also known.

FIG. 9 is a schematic block diagram showing the third embodiment. It is the same as the previous example until the flip-flop 14 generates a rectangular wave in which the front wave F corresponds to "1" and the rear wave R corresponds to "0". A sawtooth wave (triangular wave) is used here to measure the length of the rectangular wave. The length of time is obtained by generating a triangular wave in which the voltage rises at a constant rate and measuring the final voltage. Two sawtooth wave generation circuits 21 and 22 are provided. Further, peak-hold circuits 23 and 24 for temporarily holding these final voltages are connected.

The first sawtooth wave generation circuit 21 generates a voltage that continues to rise for the time Q = 1 (T ₁ + T ₂ ). Pee
The hold circuit 23 temporarily holds a voltage proportional to (T ₁ + T ₂ ). The Q signal is inverted (-Q) by the inverter 25 and then applied to the sawtooth wave generation circuit 22.
It continues to rise for (T ₃ + T ₄ ). The peak hold circuit 24 holds the final voltage. This is (T ₃ + T
It is a voltage proportional to ₄ ). By the subtractor 26 (T
₁ + T ₂₎ - to calculate the (T ₃ + T _4).

[0059]

EFFECT OF THE INVENTION From the signal of the light receiving element, the entire AC component is taken out, and the left-handed light and the right-handed light are detected from the difference in the duration from the zero point to the zero point of the two waves included in one period T of phase modulation. In the AC time difference method for obtaining the phase difference Δθ, the present invention takes a reference signal from the phase modulation signal and permits or prohibits the frequency division by this. Since the frequency division start timing is synchronized with the signal of the phase modulator, the front wave F and the rear wave R are not confused. Even if noise enters, the front wave F and the rear wave R can be correctly recognized in the next cycle. The reference signal can be easily obtained from the drive circuit of the phase modulator. It can also be generated by special circuitry, but can simply be taken from the clock of the phase modulator.

There is a common point in the conventional partial wave decomposition method in that the timing signal is taken from the phase modulator.
But actually it's totally different. In the conventional partial wave decomposition method, of the signals of the light receiving element, only the fundamental wave is extracted and synchronously detected. What is obtained by synchronous detection is a voltage signal. Therefore, the carrier signal must be correctly in phase with the fundamental wave. It does not matter if they are almost the same. If the phase angle is different, this difference is taken as Ψ and the result decreases in proportion to cos Ψ. It requires time consuming and subtle adjustments to avoid this.

Although the present invention also takes the timing signal from the phase modulator, it is not necessary that the phases be in perfect alignment, as the reference signal rises (crosses the zero) before the beginning of T _1. Good. The reference signal can therefore be offset by up to 180 degrees. Initial phase adjustment work is unnecessary. Moreover, the measured value is not voltage but time. The accuracy of time measurement is far superior to that of voltage measurement. Therefore, a highly accurate rotation angular velocity measuring device can be provided.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of an optical fiber gyro according to a first embodiment of the present invention.

FIG. 2 is a graph of a Bessel function. The horizontal axis is the independent variable ξ, and the vertical axis is the value of the function Jn (ξ). The parameter is the order of the Bessel function.

FIG. 3 is a time waveform in one cycle of phase modulation of an AC component extracted from an interference signal detected by a light receiving element. The solid line is when the coil is stationary, and the broken line is when the coil is rotating.

FIG. 4 is a waveform chart showing a waveform of an AC signal detected by a light receiving element, which is binarized and further divided by ½. (1) shows an interference signal, (2) shows a signal obtained by binarizing the interference signal, and (3) shows a signal obtained by dividing the binarized signal by 1/2.

FIG. 5: The AC signal detected by the light receiving element is binarized, and 1 /
FIG. 6 is a waveform diagram illustrating that the result of ½ frequency division is inverted if the frequency division start timing is shifted or noise is generated when frequency division is performed. (1) is the light-receiving element output, (2) is the signal obtained by binarizing (1), (3) is the signal obtained by dividing the binarized signal by 1/2, and (4) is the inverted signal of (3). , (5) is the signal (2)
At (6), a signal obtained by dividing the binarized signal (5) having noise by 1/2 is shown.

[FIG. 6] In the present invention, an AC signal is binarized to ½.
FIG. 8 is a waveform diagram illustrating that the frequency division result is not reversed at the time of frequency division start at the time of frequency division. (1) is a signal obtained by binarizing the light receiving element signal, (2) is a signal at the D terminal input of the flip-flop, (3) is a binary signal obtained by binarizing the light receiving element output, 4) shows a 1/2 frequency-divided signal when the timing of starting frequency division is deviated.

FIG. 7: In the present invention, an AC signal is binarized to be 1/2
FIG. 6 is a waveform diagram for explaining that the division result is not reversed even if noise is input when dividing. (1) sets the light receiving element output to 2
Quantized signal, (2) signal at D terminal input of flip-flop, (3) 1/2 divided signal of light receiving element output when there is no noise, (4) shows light receiving element when there is noise The signal is binarized, and (5) shows the light-receiving element signal binarized and divided by ½ when there is noise.

FIG. 8 is a schematic configuration diagram of an optical fiber gyro according to a second embodiment of the present invention.

FIG. 9 is a schematic configuration diagram of an optical fiber gyro according to a third embodiment of the present invention.

FIG. 10 is a diagram showing an example of a circuit for extracting an AC component from the output of a light receiving element.

[Explanation of symbols]

 1 Light Source (Light Emitting Element) 2 Optical Fiber 3 Coupler 4 Fiber Optical Path 5 Coupler 6 Fiber Coil 7 Phase Modulator 8 Clock 9 Low Pass Filter 10 Photoreceptor 11 Amplifier 12 Capacitor 13 Comparator 14 Flip-Flop 15 Connection Line

Claims (1)

[Claims] 1. A light source for generating monochromatic light or quasi-monochromatic light,
It includes a fiber coil formed by winding an optical fiber in a coil shape many times, a light receiving element for detecting the intensity of light propagating through the fiber coil, and a phase modulator provided at one end of the fiber coil. Split, propagate clockwise and counterclockwise in the fiber coil, merge again, and detect the interference light intensity by the photodetector. From the AC component in the photodetector output, two waves included in one cycle of phase modulation. The lengths of the front wave F and the rear wave R are measured, and from the difference between the two wave lengths,
An optical fiber gyro that detects the rotational angular velocity from the phase difference between the clockwise light and counterclockwise light generated when the fiber coil rotates, and extracts the reference signal from the phase modulator signal. A signal detection method for an optical fiber gyro, characterized in that a front-back relationship between two waves included in one cycle is defined.