The present invention relates to fiber optic Fabry-Perot interferometers and more particularly to a method and apparatus for quantitatively measuring the absolute length of a static gap in a Fabry-Perot interferometer. This application claims the benefit of application Ser. Nos. 60/562,492 and 60/562,682, both filed on Apr. 15, 2004.
Fabry-Perot sensors have broad utility for applications where the measurement of the absolute length of an interferometric gap in a Fabry-Perot sensor. These gaps may relate to pressure, temperature, strain or some other physical property of the material which bounds one side of the gap. For example, their simplicity of design allows these sensors to be embedded into large industrial applications including gas turbines, pressure vessels, pipelines, buildings, or other structures, in order to provide information about pressure, temperature, strain, vibration, or acceleration within the structure. Their size, durability and fast response time make these sensors advantageous.
In operation, Fabry-Perot interferometers are capable of spanning a range of gaps to create an interference pattern, regardless of whether via reflected light or transmitted light. Performing an optical cross-correlation of such an interference pattern, by reflecting or transmitting the interference pattern through a second interferometer, produces a distinctive signal that reaches a peak intensity of light when the length of the gap in the optical cross-correlator matches the length of the gap in the Fabry-Perot sensor. This distinctive peak intensity signal forms the basis for measurement of the absolute length of a gap in the Fabry-Perot sensor. Although previous systems known to the inventors use optical cross-correlators to make measurements of the length of gaps in Fabry-Perot sensors, the invention described herein is capable of making quantitative, absolute measurements with better sensitivity, greater dynamic range, greater frequency response, and lower cost than previously known systems.
The invention, at its most basic level, consists of one or more light sources, a first Fabry-Perot sensor spanning a gap which varies in response to changes in the environment (pressure, temperature, strain, etc.) and a second sensor having means for optically cross-correlating modulated light that is reflected by or transmitted through the first Fabry-Perot sensor. This second sensor includes means of controllably varying the length of the gap in the second sensor. A correlation burst signal detector is used, and means for verifying the gap distance of the second sensor are required. Lastly, means for comparing correlation burst signals from the first and second sensor in order to determine the absolute distance of the variable gap in the sensor are also included. Additional light sources may be provided, and the means for verifying the gap distance of the second sensor may comprise a set of known, fixed distance sensors which represent upper and lower limits for the sensitivity of the overall system.
The light sources may consist of a broadband light emitting diode (LED), edge light emitting diode (ELED), super luminescent diodes (SLEDs), wideband lasers such as a vertical cavity surface emitting laser (VCSEL), narrow band lasers such as a HeNe, or various tungsten lamps.
The means for optical cross-correlation of the modulated light reflected by or transmitted from the Fabry-Perot interferometric sensor preferably comes in the form of an optical cross-correlator placed in series with the Fabry-Perot sensor. As used throughout, the term optical cross-correlator should be understood to mean a system element having a variable gap where the gap is bounded on either side by partial reflectors. Preferably, the reflectivity of these boundary surfaces is between 20% and 50%. This optical cross-correlator is preferably configured as a Fabry-Perot interferometer. The amplitude or percentage of light reflected from or transmitted through the Fabry-Perot sensor and reflected from or transmitted through the optical cross-correlator is defined by the cross correlation product of the classic interferometric equation for each interferometer. For further discussion of such modulation, including the various equations that may be used to perform the calculations contemplated by this invention, refer to Principles of Optics, Chapter 7, Born and Wolf which is hereby incorporated by reference. This classic interferometric equation defines the intensity of light as a function of both the length of the gap in the interferometer and the spectral distribution of the light that is transmitted from the light source(s).
The length of the gap in the optical cross-correlator may be variable by oscillating or moving one or both of the reflectors in the Fabry-Perot optical cross-correlator via a lead-zirconate-titanate (PZT) or some other linear or rotary actuator. The means for controlling the position of the optical cross-correlators can be accomplished with any linear or rotary positioner such as stepper motors, PZTs, magnetostrictive actuators, lever arms or any combination thereof.
The resultant correlation may be detected by one or more detectors. The detectors may consist of silicon or InGaAs photodiodes. The detectors may view different light sources with different wavelength bands. The detectors convert the light signals into an electronic output, and an electronic processor converts the electronic signals into representative measures of the Fabry-Perot sensor gap which correspond to the pressure, temperature, strain, vibration, or acceleration of interest. The electronic signals from the detectors are also used to control the frequency and amplitude of the oscillations and/or the length of the gap in the optical cross-correlator.
Finally, the invention contemplates the processing of the electronic signals from a microprocessor where software is used to read the electronic signal, control the position of the optical cross-correlators, and generate an output signal indicative of the length of the gap in the Fabry-Perot sensor.
One embodiment of the present invention relies upon an optical cross-correlator configured as a Fabry-Perot interferometer with a variable length of gap to make absolute measurements of the length of a gap in a Fabry-Perot sensor at relatively high frequency and at with a higher dynamic range than can be accomplished via other means. In this embodiment, the variable gap optical cross-correlator does not oscillate but is moved via a PZT or similar device through a range of gaps until the length of its gap matches that of the Fabry-Perot sensor. Then the system tracks changes in the length of the gap in the Fabry-Perot sensor by dithering, (oscillating through a very small range of motion). By measuring or otherwise knowing the precise length of gap in the optical cross-correlator where the length of the Fabry-Perot gap is identical to the length of the gap in the optical cross-correlator, one also knows the precise length of the gap in the Fabry-Perot sensor.
In an alternate embodiment, the variable gap optical cross-correlator is configured as a Fabry-Perot interferometer using a PZT element that oscillates at a high rate to sweep through a range of gaps at high frequency. Twice in each oscillation or sweep cycle, the length of the gap in the optical cross-correlator precisely matches the length of the gap in the Fabry-Perot sensor and at these moments a peak in the correlation signal is produced. By precisely knowing or mesuring the time of the occurrence of each match and by knowing the amplitude and frequency characteristics of the oscillations of the optical cross-correlators, one also knows the precise length of the gap in the Fabry-Perot sensor.
The amplitude and frequency of the oscillations and the precise length of the gap in the optical cross correlator can be controlled and known by applying a known voltage to a PZT element. Further embodiments contemplate the use of one or two reference sensors spanning fixed, known gaps along with two or more light sources to increase the accuracy of the system.
In operation, the inventive system comprises a light source, a first Fabry-Perot sensor capable of spanning a range of gaps, an optical cross-correlator configured as a second Fabry-Perot interferometer spanning a gap of a known length and capable of changing the length of that gap in a controllable and known manner, detector means to convert the light signals into electronic signals, and the electronic means to control the length of the gap in the optical cross-correlators and to generate an output signal indicative of the parameter to be measured.
a is a schematic of the invention using a transmissive optical cross-correlator.
b shows an alternative embodiment for the optical cross-correlator which uses a reflective optical cross-correlator.
a shows a typical output curve for a fixed sensor gap where the length of the optical cross-correlator gap varies when a narrowband light source such as an ELED is used according to the invention.
b shows a typical output for a fixed sensor gap where the length of the optical cross-correlator gap varies when a wide bandwidth spectral source is used according to the invention.
a shows an alternate embodiment of the invention which includes the use of three separate light sources to determine the precise length of gap in the optical cross-correlator.
b shows transmission versus wavelength for cut-off filter F1.
c shows transmission versus wavelength for cut-on filter F2.
d shows output signals from detectors D1,D2,D3 that illustrate light intensity versus gap for in
e shows a typical output (light intensity versus gap for VCSEL starting from zero gap) for a fixed sensor gap where the length of the optical cross-correlator gap varies when a laser light source is used according to the invention.
f shows another alternate embodiment of the invention.
a-9d depict alternative arrangements for the invention, also including signal calculation information.
A first embodiment of the inventive system 10 is shown in
For a Fabry-Perot sensor with a fixed length of gap, the intensity of the light will vary as a function of the length of the gap in the optical cross-correlator as shown in
A second embodiment of the inventive system 10 is shown in
Notably, the signals generated in
The PZT actuator 14 may be configured as a stack or as a bimorph as illustrated in
Overall system accuracy can be improved through variations to the elements shown in
While each of the items listed in Table 1 can be used in the present invention, it is important to note the limitations of each. The small bandwidth makes the LVDT and the strain gage less attractive than the optical sensor and capacitance sensor. Both the capacitance and the strain gage may have long-term stability problems due to creep of the adhesive needed to bond these sensors to the PZT. Nevertheless, these options, along with others known to those skilled in the art, are available in configuring an enhanced system according to the schematic of
As seen in
A capacitance sensor 50 which can be used as the positional verification device is depicted in
An alternative to the optical reference sensor would be to make a direct measurement of the length of the gap in the PZT interferometer (rather than the relative position of the PZT as in
Such an alternate embodiment is shown in
Light from sources L2, L3 travels through splitter C1 and the interferometer S2, which does not perform a cross-correlation because the light from sources L2, L3 has not been modulated (i.e., it does not come into contact with interferometer S1). However the interferometer S2 modulates the light from sources L2, L3. After splitting at C2, the light from source L2 passes through the cut-off filter F1, through a second interferometer R1 with a known, stable fixed gap and then onto detector D2 (InGaAs). R1 acts as an optical cross-correlator for the ELED light modulated by the PZT interferometer, and detector D2 converts the cross-correlated ELED light into an electrical current. Notably the long wavelength light from the tungsten lamp source L1 is not filtered by F1 and is transmitted to detector D2 along with the ELED light. Since the intensity from the tungsten lamp is very low compared to the ELED, the tungsten light has a negligible effect on the signal at detector D2.
After splitting at C2, the light from the VCSEL source 51 travels through a cut-on filter F2, which passes the long wavelength VCSEL light but blocks the short wavelengths from the tungsten lamp and ELED (see
d and 6e show the signals generated at each detector as the interferometer S2 is moved through a range of motion corresponding to the gap range (
I=1/[1+F sin2 (2πGR/λ)] (1)
The system operates in two modes, i.e. calibration mode and measurement mode. In calibration mode, the PZT 14 and interferometer S2 are scanned through the range of motion 0 to 30 um. The signal from detector D2 reaches a peak when the length of the gap GR in interferometer S2 is equal to the length of the gap at reference interferometer R1, which has a fixed and known gap of 28 um in the example in
In measurement mode, the voltage to the PZT 14 is changed from its value that resulted in a gap of 28 um until the output signal from D1 reaches its peak as shown in
Another alternative to the optical reference sensor relies on making a direct measurement of the length of the gap in the PZT interferometer and thereby again eliminates the problems associated with calibration, long term drift, resolution, repeatability and accuracy of the PZT and the complexities of the embodiment described above. This alternative employs a very stable light source such as a HeNe laser whose wavelength is more stable than other sources. Through this embodiment, the need for system calibration occurs primarily at system startup.
This alternative is described in
Light from the HeNe source travels through splitter C1 and the interferometer S2, which does not perform a cross-correlation because the light from the HeNe has not been modulated. However the interferometer S2 modulates the light from the HeNe. After splitting at C2, the light from the HeNe travels to detector D2 (Si). Notably the long wavelength light from the tungsten lamp source L1 is transmitted to detector D2 along with the HeNe light. Since the intensity from the tungsten lamp is very low compared to the HeNe, the tungsten light has a negligible effect on the signal at detector D2.
d and 6e show the signals generated at each detector as the interferometer S2 is moved through a range of motion corresponding to the gap range (
I=1/[1+F sin2(2πGR/λ)] (1)
As before, the system operates in two modes, i.e. scan mode and measurement mode. In scan mode, the PZT is scanned through the range of motion 0 to 30 um. The signal from detector D1 reaches a peak when the length of the gap at the interferometer S2 is equal to the length of the gap in the Fabry-Perot sensor S1. Detector D2, which measures light intensity from the HeNe is monitored during scan and software keeps track of the fringe count continuously. Since there is negligible uncertainty in the wavelength of the HeNe, there is no need for a wavelength calibration as there is with a VCSEL or other unstable light source. Once the peak intensity in detector D1 is found, the system changes into measurement mode.
In measurement mode, the sensor gap is calculated in terms of the absolute wavelength of the HeNe using Equation 1. Thereafter in measurement mode, the PZT voltage is dithered so that the correlation pattern signal from D1 is tracked by software. Changes in the peak value are tracked by fractional changes in fringe shift at detector D2.
The frequency response is limited by the PZT scan rate and the absolute measurement accuracy is determined by the repeatability of the gap measurement using the reference sensor.
Yet another means for improving the resolution and accuracy involves the use of a time-based calculation on the absolute position of the Fabry-Perot sensor. This embodiment eliminates some of the hysteresis and creep in the lead-zirconate-titanate (PZT) modulator.
The elements of the time-based system are shown in
The cross-correlated signals from interferometers R1, R2, and the Fabry-Perot sensor S1 are monitored continuously by detector D1, D2, and D3 respectively (although the invention can be configured for fewer than three detectors as shown in
As oscillating interferometer S2 travels through its range of motion, the gap GR provides a range of gaps as a function of time. Each of the three detectors (D1, D2, D3) sees a peak in the correlation burst pattern when the length of the gap GR matches the length of the gap in each respective interferometers (R1, R2, S). The peak detector signals from each interferometer S1, R1, R2 are observed as a precise point in time that is a function of the amplitude and frequency of the oscillation gap GR in the oscillating interferometer S2. Refer to
The Fabry-Perot sensor gap GS is calculated based on the known precise gap and time of occurrence of the peak intensity of reference interferometers R1 and R2 and the sinusoidal functional dependence of the oscillating displacement of the range of gaps from GR.
These three signals as described are plotted in
GS=(t1/t2)[GR2−GR1)+GR1
The equation for above assumes a linear change of the scanned gap with time. In fact, the change is sinusoidal and must be modified accordingly to deal with the nonlinearity. Specifically, the scanned gap is driven sinusoidally at frequency co and can be expressed as
G=A+B cox (ωt)
The correlation peaks of interest occur at the times when:
GR1=A+B cos (ωto)
GS=A+B cos (ωt1)
GR2=A+B cos (ωt2)
Using the equations and information above, it becomes possible to calculate the actual value of GS through accurate time measurement, as achieved by the aforementioned microprocessor, as follows:
GS=GR1+(GR2−GR1)[(cos(ωto)−cos(ωto))/(cos(ωt2)−cos(ωto))]
Notably, the equation above can be manipulated and used to achieve accurate time-based measurements according to any of the alternative embodiments described below.
To maximize the signal the lamp should be a quartz-halogen type that allows high filament temperature while maintaining long life. Exemplary filament temperatures in the range of 2700 K can burn for about 10,000 hours, while temperatures exceeding 3100 K drop that life span to around 100 hours. Notably, manufacturers define the lamp properties in terms of color-temperature, which is approximately 90 degrees higher than the actual filament temperature.
Clearly, there is an advantage to using the higher temperature lamp, but the added power comes at a cost of lifetime so there is a trade-off. One possible arrangement would be to use a lower temperature light source, and if there is a signal level problem, the higher temperature lamp can be substituted or integrated into system 10 as an alternative.
Additional consideration should be given to the radiance of the light source, which impacts the power delivered throughout the system. Further discussion of these principles can be found in the Photonics Handbook, the relevant portions of which are hereby incorporated by reference. As recognized by those skilled in the art, the radiance of the lamp filament can be determined with the temperature and the emissivity of the filament material (preferably tungsten). The radiance is also a function of wavelength, while the total integrated irradiance over the spectral range from 550 mn to 1050 nm is the quantity of interest for tungsten. Of course, these spectral limits are somewhat arbitrary, but are based on the basic fact that the detector response curve falls to about ½ its maximum value at these wavelengths.
The fraction of the input power delivered to the detectors is based on the reflectance from the sensors S1, S2, R1, R2. Ideally, this reflectance measured should be approximately 50% of the input light power.
Additional alternative arrangements of the optoelectronic components for system 100 are possible, although all of these are fundamentally rooted in the comparative calculation principle set forth in system 10. For all of the variations discussed below, the previous designations utilized in
The first such alternative arrangement is presented 9a. Light source L is provided to splitter C1. Notably, all of the sensors, as well as scanned gap GR operate in reflection mode. The reflectance from each gap is indicated by ri and beside the detector is indicated the magnitude of the power delivered through the system arrangement, where Io is the input power. Detector D is used to monitor all three sensors S1, R1, R2 through appropriate routing by splitters C2 and C3, and the power delivered to the detector consists of three terms, one from each reference and one from the sensor. In turn, these terms represent the product of two reflectances depending upon the routing of the light (e.g., rRT×rR1) and are not simple products but rather the correlation product that yield individual burst patterns. Adjusting for these variations, further calculations are consistent with the principles described above (also depicted on
b shows another alternate embodiment that employs one 2×4 splitter in place of the three splitters shown in
c shows yet another configuration requiring a single 2×2 splitter and a single detector D. Significantly, the first two terms in the expression for the power to the detector are simply the feed-through of half the power input to the splitter. These terms do not contain any correlation information and simply add noise.
d shows yet another configuration requiring a single 2×2 splitter and a single detector D where the reference interferometers operate in transmission mode rather than reflection.
The signal levels provided for each configuration are summarized in Table 3. In some configurations, the signal level for the reference interferometers is different from that for the sensors. In these cases, Table 3 lists the worst case. It is assumed that the reflectance and transmittance of all sensors is the same, so the subscripts that are used for clarity in the Figures are omitted in Table 3. It is also assumed that there is no excess loss in the splitters.
To make valid comparisons based on the expressions in Table 3 requires quantification of the cross-correlation terms tt, tr, and rr. Based on 30% reflectance for the separated mirrors that define the gap for each sensor, the cross-correlation products are conservatively estimated to be: tt=0.37; tr=0.31; and rr=0.41. Using these values enables evaluation of the products in column 2 of Table 3 to obtain the fraction of power delivered to the detector, which is given in column 3. The preferred configuration of
It is appropriate to perform a signal-to-noise ratio analysis. Consider, for example a sensor signal update rate of 10 kHz. The scanned gap is sinusoidally driven by the oscillator at 5 kHz, providing the desired 10 kHz update rate. Assuming a total scanning range of 20 μm, the scanned gap is expressed as
g(t)=10 μm(sin ωt)+Q (5)
The scanning rate, dg/dt, is a function of time. The maximum scan rate occurs at the time when sin ωt=0. At this point
10 μm(sin ωt)˜10 μm (ωt)
So, dg/dt=10 μm ω=(10)2π(5000)=3.14×105 μm/sec or 314 nm/μsec (6)
A correlation model can be used to provide results in terms of intensity as a function of correlation element gap. The gap is converted to time using the scan rate Equation (6) above, and the correlation pattern can be viewed as a quasi-sinusoid with a frequency of 0.84 MHz. There are no high frequency features of interest. Accordingly, the photodiode amplifier is designed as a band pass amplifier with a range of 100 kHz to 1 MHz. This is the frequency response of the photodiode signal and not to be confused with the time resolution required to measure the sensor gap with a resolution of 0.1%, (10 nm assuming a full scale range of 10 μm). A 10 nm gap change converted to a time base using Equation (6) reveals a 32 nsec change in time. As a minimum, the time base needs a resolution of 32 nsec.
To quantify the effect of noise, consider 1 MHz sine wave with variable amplitude added to the signal. A 1 MHz noise frequency is considered because higher frequencies are filtered out and lower frequencies do not affect the peak. Noise with frequency content comparable to that of the correlation peak, however, does affect both the amplitude and position of the peak. Ideally, the signal processor should be capable of operating with SNR=50.
The amplifier noise increases with the capacitance of the photodiode. Thus a photodiode is needed with the smallest capacitance possible. A UDT Sensors PIN-020A has an active area with a diameter of 510 μm and a capacitance of 1.0 pf when reverse biased at 10 V.
Other noise sources that need to be considered for systems 10, 11 or 12 include: shot noise due to DC signal current plus dark current; Johnson noise from the feedback resistor; noise due to amplifier input current noise; and noise due to amplifier input voltage noise. Conservative calculations show that the combined noise terms can be estimated and expressed as an RMS value of about iT=3.8×10−10 A. Recall that the estimated light power level delivered to the detector was determined in paragraph 88 to be approximately 5.2×10−8 Watts. The effective detector responsivity is on the order of 0.3 A/W, and the expected signal current is 1.6×10−8 amps. Accordingly, the signal-to-noise ratio is 1.6×10−8/3.8×10−10=42. While this SNR falls slightly below the preferred value of at least 50, it is a worst-case estimate. The SNR can be improved by modifying the arrangement shown in
In the same spirit, the ideal performance ranges for system 10 as shown in
The oscillator for gap GR changes at a set rate, i.e. 1000 Hz and travels through a range of motion of approximately 20 μm. The ultimate range of motion is approximately 5 to 25 μm for the scanned gap, which consists of a moving mirror that maintains parallelism with the reflective end of an optical fiber. The range of motion and the rate of oscillation may be modified for specific applications. The trade off with increased bandwidth is increased noise and reduced dynamic measurement range of the Fabry-Perot sensor gap to be measured.
The detectors D1, D2, D3 may be either silicon or InGaAs photodiodes. For short-range applications, i.e. up to 1000 meters, a quartz-halogen lamp provides the best performance. For long-range applications, i.e. 500 meters to 2500 meters the tungsten lamp provides somewhat better performance.
The inventive systems 10, 11, 12 can be easily multiplexed with several channels of optical data sharing a single oscillator and microprocessor, however, each channel requires its own set of reference interferometers and photodiodes.
Finally, the system electronics are depicted
Number | Date | Country | |
---|---|---|---|
60562492 | Apr 2004 | US | |
60562682 | Apr 2004 | US |