JP2009257986A - Spectrum measuring device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a spectrum measuring device for measuring the spectrum of a random signal extremely smaller than the instrument noise, in measuring random phenomena. <P>SOLUTION: This spectrum measuring device comprises first and second detectors 11 and 12 for detecting the waves or material coming out of a measuring object, correlators 13, 14 and 15 for calculating the correlation spectrum of signals output from the first and second detectors, and an integrator 16 for integrating the signals output from the correlators. At this time, random signals are the same in two measurement systems, and the instrument noises are statistically independent (uncorrelated), so that the random signal and the instrument noise are distinguished from each other by statistical processings. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a spectrum measuring apparatus that can remove random noise that is instrument noise from a random signal.

  Among the measurements using waves, optical measurement using laser light is one of the most accurate measurement methods available at present, and is used for atomic force microscopes and gravity wave detection. However, optical measurement has inevitable noise called shot noise, which determines the theoretical limit of sensitivity. Scattering noise is noise caused by random photoemission due to the photoelectric effect, and is always accompanied by detection methods that perform photoelectric conversion using photodiodes or photomultiplier tubes, except in special cases. . Fortunately, the scattered noise is uncorrelated white noise and can be reduced using statistical techniques such as integration.

Therefore, when the signal to be measured is a repetitive signal, integration is an effective statistical process. Noise decreases with respect to the number of integrations n according to 1 / √n, and there is no theoretical measurement limit in a situation where the number of integrations can be increased without limit. (For example, refer to Patent Document 1.)
Japanese Patent Application No. 2007-218941

  On the other hand, when the signal to be measured is a random signal, the situation is different. Random signals are not only important in basic physics such as measurement of thermal motion and quantum effects of atoms and molecules, and measurement of radiation from celestial bodies, but also move randomly as seen in ultrasonic echo methods. It is also important for detecting reflected waves from objects. These random signals are difficult to distinguish because there is no statistical difference from noise (instrument noise) associated with measurement such as scattered noise. For example, in simple integration, not only noise but also random signals are reduced at the same time, so that random signals and noise cannot be distinguished. It is difficult to detect a random signal smaller than instrument noise.

  In view of the above problems, an object of the present invention is to provide a spectrum measuring apparatus capable of measuring a spectrum of a random signal much smaller than instrument noise in measurement targeting a random phenomenon.

  According to a first aspect of the present invention, there is provided a spectrum measuring apparatus including first and second detectors for detecting a wave or a substance arriving from a measured object, and a cross-correlation spectrum of signals output from the first and second detectors. And a multiplier for integrating signals output from the correlator.

  Further, the spectrum measuring apparatus according to claim 2 includes a projecting unit that projects the wave onto the object to be measured, and the first and second detectors reflect a reflected wave or a transmitted wave from the object to be measured. It is a thing which detects this. Thus, the spectrum of the object to be measured that does not emit waves can be measured.

  The spectrum measuring apparatus according to claim 3 is characterized in that the wave is light. Thereby, the optical spectrum and the vibration spectrum of the object to be measured can be measured with high accuracy.

  The spectrum measuring apparatus according to a fourth aspect is characterized in that the integrator integrates signals on a time axis. As a result, integration based on long-time measurement data can be performed, and instrument noise can be greatly reduced in spectrum measurement.

  The spectrum measuring apparatus according to claim 5 is characterized in that the integrator integrates signals on the frequency axis. Thus, in addition to the fourth aspect, it is possible not only to further reduce the instrument noise, but also to reduce the instrument noise in the spectrum measurement even in a short time measurement.

  According to the present invention, it is possible to measure a spectrum of a random signal that is much smaller than instrument noise in measurement targeting a random phenomenon. In particular, in the treatment of the eye, the degree of elasticity of each part including the inside of the eye can be diagnosed noninvasively by measuring the thermal vibration spectrum of a predetermined part of the eye.

  The best mode for carrying out the present invention will be described below in detail with reference to the accompanying drawings.

  The present invention uses two measurement systems and simultaneously measures the same object independently. At this time, since the random signals are the same in the two measurement systems and the instrument noise is statistically independent (non-correlated), the random signal and the instrument noise can be distinguished by statistical processing.

FIG. 1 is a diagram for explaining the principle of the present invention. The spectrum measuring apparatus of this embodiment includes detectors 11 and 12, FFTs 13 and 14, a multiplier 15, and an averager 16. It is assumed that a random signal S (t) is radiated from the sample S that is the object to be measured. This random signal S (t) may be emitted from the sample S spontaneously like thermal radiation, or the sample S is irradiated with waves such as radio waves or light as in the case of radar or laser measurement. It may be a wave after interaction, that is, a reflected wave or a transmitted wave. The random signal S (t) is independently measured by the two detectors 11 and 12. The measured signals D1 (t) and D2 (t) include instrument noise N1 (t) and N2 (t) in addition to the random signal S (t).
D1 (t) = S (t) + N1 (t) (1)
D2 (t) = S (t) + N2 (t) (2)
It can be expressed as. When a spectrum analysis is performed on signals detected in the time domain by FFT (Fast Fourier Transformation) 13 and 14, a signal in the frequency domain is obtained.

^c D1 (f) = ^c S (f) + ^c N1 (f) (3)
^c D2 (f) = ^c S (f) + ^c N2 (f) (4)
Here, f indicates an analysis frequency, and ^c indicates a complex number. When a product of the complex conjugate ( ^* ) of ^c D1 (f) and ^c D2 (f) is calculated by the multiplier 15, and an average value thereof is obtained by the averager 16,
< ^c D1 (f) ^c D2 ^* (f)>
^{= <| C S (f)} | 2> + <c N1 (f) c S * (f)> + <c S (f) c N2 * (f)> + <c N1 (f) c N2 * ( f)>
(5)
It becomes. Here, <> represents an average, and a specific averaging method will be described below. Since the signal ^c S (f) and the instrument noise ^c N1 (f), ^c N2 (f) are statistically independent from each other,
^{<C D1 (f) c D2} * (f)> = <| c S (f) | 2> (6)
Thus, a frequency spectrum is obtained. On the other hand, in the conventional frequency spectrum measurement method using only the detector 11,
<| ^C D1 (f) | ² >
^{= <| C S (f)} | 2> + <c N1 (f) c S * (f)> + <c S (f) c N1 * (f)> + <| c N1 (f) | 2>
(7)
≒ <| ^c S (f) | ² > + <| ^c N1 (f) | ² > (8)
It becomes. ^{Therefore, <| c N1 (f)} | 2> in order to measure accurately the frequency spectrum of the smaller signal than ^{is, <| c N1 (f)} | 2> and accurately measured in a different manner, It is necessary to subtract. Since the instrument noise changes in an environment such as temperature, it is difficult to obtain a reliable measurement result with this method.

Next, the average <> in Equation (5), that is, the averager 16 in FIG. 1 will be described. The analog signals output from the detectors 11 and 12 are converted into discrete digital signals at a sampling interval Δt. A spectrum is obtained by performing FFT processing with J data as a group. It is assumed that such measurement is repeated M times. Accordingly, the data obtained by this measurement can be expressed as ^{c D1m (fj), c D2m} (fj). Here, m = 1, 2,..., M−1, M. fj (j = 0,..., J / 2) is an analysis frequency obtained by FFT processing. From the sampling theorem, f0 = 0, f1 = 1 / (ΔtJ),..., fj = j / ( ΔtJ),..., FJ / 2 = 1 / (2Δt). The total for the repetition of these data is
^{<C D1 (fj) c D2} * (fj)> = (1 / M) Σ M m = 1 c D1m (fj) c D2m * (fj) (9)
It becomes.

On the other hand, when the frequency resolution Δf required for measurement is larger than the frequency resolution 1 / (ΔtJ) determined by the sampling theorem, integration can be performed on the frequency axis. If the ratio of both is Q = ΔfΔtJ, the final integration result is
^{<C D1 (fj) c D2} * (fj)> = {1 / (Q + 1)} Σ Q / 2 q = -Q / 2 <c D1 (fj + q) c D2 * (fj + q)> (10 )
It becomes.

  The results of measuring the thermal motion of the solid (elastic body) surface using this method will be described below. The atoms and molecules that make up the object move randomly due to thermal energy. For this reason, the object surface is also randomly undulating. This phenomenon is conspicuous on the surface of the liquid and is called lipron. The reprons on the surface of the liquid have an amplitude of about 0.5 nm, and a signal intensity sufficiently larger than the scattered noise can be obtained by ordinary laser measurement, so that the measurement can be easily performed. Similar thermal motion should have occurred on the surface of the solid, but the reprons on the surface of the solid had a small amplitude and were difficult to detect by optical measurement.

FIG. 2 is a diagram showing a specific configuration of the measurement system. FIG. 3 is a diagram showing a configuration closer to an actual measurement system. Here, the reprons on the solid surface were measured by measuring the tilt angle of the sample surface using an optical lever. As the light source, semiconductor lasers 21 and 22 having wavelengths of 638 nm and 658 nm were used, and light emitted from both was synthesized by a dichroic mirror 25 via isolators 23 and 24. The synthesized light was condensed by the objective lens 28 via the polarization beam splitter 26 and the quarter wavelength plate 27 and then irradiated to the sample. The reflected light from the sample was collected by the same objective lens 28 and then separated by the dichroic mirror 29 into light having a wavelength of 638 nm and a wavelength of 658 nm. Each light was detected by dual-element detectors 30 and 31, and a signal proportional to the tilt angle of the sample surface was obtained. Outputs from the two-divided photodetectors 30 and 31 were converted into digital signals by the AD converter 32. The sampling rate of the AD converter 32 is 8 Msamples / s (Δt = 1 / (8 × 10 ⁶ ) s), and 4 M words (J = 2 ²² ) are measured at a time and integrated after Fourier transform. Note that the pinhole 41 shown in FIG. 3 is for the purpose of accurately superimposing two laser beams and for correcting distortion of the beam pattern of the semiconductor laser light, and is a video camera that shoots through the dichroic mirror 42. 43 is for monitoring the state in which the sample S is irradiated with the laser.

FIG. 4 shows an example in which the thermal motion of a cantilever manufactured for an atomic force microscope is measured. The cumulative number M is 600 times. The nominal resonance frequency is 26 kHz, and vibration at this frequency is observed. The “conventional” shown in the figure indicates <| ^c D1 (f) | ² >, and the “present invention” indicates < ^c D1 (f) ^c D2 ^* (f)>. In the “conventional”, a signal buried in noise is clearly observed in the “present invention”. It can also be seen that there is another resonance at 143 kHz and 365 kHz. Cantilever vibrations at frequencies other than these resonance frequencies are extremely small. For this reason, in the equation (5), with an average value of M = 600 times,
^{<| C S (f) |} 2>«<c N1 (f) c N2 * (f)> (11)
The information other than the peak due to resonance shown in FIG. 4 is accumulated noise. Further, in the measurement example shown here, the frequency resolution is Δf = f / 100. On the other hand, the frequency resolution of the spectrum obtained by FFT is 2 Hz (= 1 / ΔtJ). Therefore, the number of integrations on the frequency axis is Q = f / (100 × 2 Hz). Since the number of integrations on the frequency axis depends on the observation frequency, the degree of reduction in instrument noise also depends on the observation frequency. As a result, the “conventional” spectrum (mainly instrument noise) in FIG. 4 hardly depends on the frequency, but the residual noise intensity in the “present invention” decreases with the frequency.

FIG. 5 is an example of observing a transparent silicon adhesive surface. FIG. 5A shows “conventional” and FIG. 5B shows “present invention”. The surface of the silicon-based adhesive is initially gel-like, but hardens over time and becomes an elastic solid. Such a change in physical properties also appears in the surface thermal vibration. The spectrum of the “invention” is proportional to f ^−4/3 at t = 0. The t = 4 days, f ^-1, f ^-2/3, is proportional to f ^-1 in 100kHz~2.5MHz in 3kHz~100kHz in 70Hz～3kHz. Comparing FIG. 5 (a) and FIG. 5 (b), it can be seen that the instrument noise is effectively reduced by the present invention.

FIG. 6 is an example of observing the epoxy adhesive surface. 6A shows “conventional” and FIG. 6B shows “present invention”. The “invention” has a spectrum of f ⁻² before curing and a spectrum of f ⁻¹ after curing. This is very different from silicon adhesives.

FIG. 7 is an example of observing the thermal motion of the double-sided pressure-sensitive adhesive tape surface. In the “present invention”, the spectrum is f ⁻² in the range of 100 Hz to 300 kHz.

FIG. 8 is an example of observing the thermal motion of the nail surface. “Invention” has a spectrum of f ^−3/2 .

When the method according to the present invention is used, in principle, the instrument noise can be reduced by increasing the number of integrations, and as a result, any small random signal can be detected. However, the degree of reduction in instrument noise in the manufactured measuring apparatus is limited to 1/100. This is mainly due to the crosstalk characteristics of the AD converter. The AD converter used is a device that can AD convert two signals simultaneously, but there is a coupling of about 10 ⁻⁴ between channels 1 and 2. For this reason, the degree of noise reduction is 1/100. In order to improve the crosstalk characteristics, the crosstalk of the measurement system was measured in advance and numerical correction was performed. This improved somewhat.

  In this measurement, the sample is irradiated with 1.05 mW of laser light condensed by an objective lens (numerical aperture NA = 0.4). For this reason, the sample is heated and the surface temperature appears to be higher than room temperature (23 ° C.). However, the temperature is not high enough to evaporate or carbonize the sample.

  In addition, this invention is not limited to the said Example.

  Waves to be observed are not limited to light, and can be applied to a wide range of measurement methods including sound waves and electromagnetic waves (such as radar). In addition, the same concept can be applied to radiation measurement for measuring the radiation of a substance such as an atom or an electron.

  In the above-described embodiment, the number of measurement systems is two, but a large number such as three is possible. In the case of a reflection type such as a radar and when the phase fluctuates randomly as the object moves, the three beams may be better.

  Particularly in optical measurement, when the intensity noise of the light source is extremely small (in a coherent state), the signals involved are only the movement of the object to be measured and the scattered noise of the measuring instrument. In this case, even if two completely independent measurement systems are not used, most of the configurations are one system, the light is divided into two immediately before the photodetector, and photoelectric conversion is performed by the independent photodetectors. Even if the cross correlation spectra of both signals are integrated, the same noise reduction effect can be obtained.

  In the case of optical measurement, as shown in FIG. 9, a pinhole 51 is provided in front of the two-divided photodetectors 30, 31 and a confocal optical system is adopted to select a reflection position in the optical axis direction. It is also possible to adjust the position of the sample S by the translation table ST and measure the movement of the lens and the retina surface within the eye, for example. In addition to measuring the surface of the cornea, which is the surface of the eye, it is possible to measure the degree of elasticity of each part of the eye by measuring the thermal vibration of each part of the eye, such as the lens inside the eye or the surface of the retina. . The results can be used for diagnosis of various eye diseases such as dry eye.

It is a figure explaining the principle of this invention. It is a figure which shows the specific structure of a measurement system. It is a figure which shows the structure nearer to an actual measurement system. This is an example of measuring the thermal motion of a cantilever manufactured for an atomic force microscope. It is an example of observing a transparent silicon adhesive surface. It is the example which observed the epoxy-type adhesive agent surface. It is the example which observed the thermal motion of the double-sided adhesive tape surface. It is the example which observed the thermal motion of the nail | claw surface. It is a figure which shows the structure which employ | adopted the confocal optical system.

Explanation of symbols

11, 12 Detector 13, 14 FFT
DESCRIPTION OF SYMBOLS 15 Multiplier 16 Averager 21 Semiconductor laser 23, 24 Isolator 25, 29, 42 Dichroic mirror 26 Polarization beam splitter 27 1/4 wavelength plate 28 Objective lens 30 2 split photodetector 32 AD converter 41, 51 Pinhole 43 Video camera S Sample ST Translation table

Claims (5)

First and second detectors for detecting waves or materials coming from the object to be measured;
A correlator for calculating a cross-correlation spectrum of signals output from the first and second detectors;
A spectrum measuring apparatus comprising: an integrator for integrating signals output from the correlator. A projection means for projecting the wave onto the object to be measured;
The spectrum measuring apparatus according to claim 1, wherein the first and second detectors detect a reflected wave or a transmitted wave of the projection wave from the object to be measured.   The spectrum measuring apparatus according to claim 1, wherein the wave is light.   The spectrum measuring apparatus according to any one of claims 1 to 3, wherein the integrator integrates signals on a time axis. The spectrum measuring apparatus according to any one of claims 1 to 4, wherein the integrator integrates signals on a frequency axis.

Images