Fiber Optic Gyro : Theory & Applications

Transcription

1 International Conference and Exhibition MELAHA 2014 RESILIENCE NAVIGATION September 1-3, 2014 Alexandria, Egypt RADISSON BLU HOTEL Organized by Arab Institute of Navigation (AIN) Fiber Optic Gyro : Theory & Applications Yves Paturel ixblue S.A.S 55 avenue Auguste Renoir, MARLY-LE-ROI, France yves.paturel@ixblue.com ABSTRACT ixblue developed its own Fiber-Optic Gyroscopes (FOG) since early nineties. Since then, the level of performance of the FOG has been increased by about 3 orders of magnitude, reaching now a bias stability close to 1.x10-4 degree per hour over several days. These gyros are used in strapdown navigation systems that reach a performance of 1 nautical mile over several days of navigation. This paper describes what the key elements in the FOG are for reaching this performance. KEYWORDS: navigation, gyroscope, fiber, gyrocompass 1 INTRODUCTION ixblue FOG design follows the well known reciprocal configuration (Figure 1). A wide range of performance is performed by simply adjusting of the fiber coil dimensions. This fundamental design is refined so that extremely high performance can be reached. In particular, we use an original erbium doped Amplified Spontaneous Emission (ASE) broadband source, a powerful All-Digital signal processing and a high performance Integrated Optical Circuit (IOC). Broadband Light Source Coupler Multi-function IOC Fiber Coil Detector A / D Digital Logics D / A Rotation signal Figure 1 : FOG block diagram ixblue expertise in these 3 elements and in optical fiber design and manufacturing allowed us to become a world leader in this technology. Moreover, ixblue uses optical fiber and IOC that were designed and are manufactured in its factory. 1

2 2 SAGNAC EFFECT GENERAL PRINCIPLE Sagnac effect [1][3] shows that light travelling along a closed ring path in opposite directions allows one to detect rotation with respect to inertial space. To understand Sagnac effect, we can consider the case of a circular path (see Figure 2). When this path is at rest, light entering the path at M is divided into two counter-propagating waves which return perfectly in phase at M after having traveled along the same path in opposite directions. When the path is rotating at angular rate, an observer at rest in the inertial reference frame sees the light travelling at the same velocity in opposite directions, but during the transit time through the loop, the beamsplitter has moved to M. Therefore the observer sees that the corotating wave has to propagate over more than one complete turn while the counter-rotating wave has to propagate over less than one complete turn. The path difference is 2 l v. This difference of transit time may be measured by interferometric means. Over one turn as in the original experiment in 1913 [3], the effect is extremely weak but it can be increased with recirculation in the resonant cavity of a ring laser or using the numerous loops of a fiber coil. 3 OPTICAL GYROSCOPES: FOG & RLG Figure 2 : Principle of Sagnac effect Both optical gyroscopes, ring-laser gyro (RLG) and fiber-optic gyro (FOG), are based on the same Sagnac effect. The RLG was demonstrated only a few years after the invention of the laser in 1960, and it is based on helium-neon (He-Ne) technology. It became very successful in the 80s and has since overcome classical spinning-wheel mechanical gyroscopes because of its improved life time and reliability. It also provided excellent scale factor, making strap-down navigation systems possible. It was clear progress over mechanical gyroscopes but gas lasers still have several drawbacks such as highvoltage discharge electrodes which tend to wear out over the long term or the need for perfect sealing of the gas enclosure. The advent of low-attenuation optical fiber and efficient semiconductor light source in the 70s opened the way for a fully solid-state device. Then, however, the FOG was seen as an approach limited to medium performance, and unable to compete with the RLG for top-grade applications. As we shall see, this is not the case anymore. 4 KEY PARAMETERS FOR A HIGH PERFORMANCE FOG 4.1 Optical source The use of an erbium doped fiber ASE source is very natural for a high performance FOG, because its properties suit very well this application. Its broadband spectrum, several nm, avoids effects of parasitic interferometers due to back reflections by decreasing their contrast. It is also an efficient way to reduce backscattering noise and, most important, nonlinear Kerr effect [6] that induces non-reciprocal phase shift. Its light is naturally nonpolarized, which is not the case of Super Luminescent Diode (SLD). It reduces phase error due to lack of polarization rejection of the IOC by averaging polarization coupling effects. It is also simpler to implement because it does not require the use of PM couplers. Its wavelength stability in temperature is 2

3 very good because it depends on erbium energy levels that are more stable than the ones of SLD. Last but not least, its components are widely used by the Telecom Industry and they are of very good quality for low price. It is also very useful to work at 1550nm to decrease the cost of all optical parts, such as couplers, fiber, IOC. ixblue s ASE Source design is the following : Figure 3 : ASE source design A laser pumps a highly doped erbium fiber. A Fiber Bragg Grating (FBG) rear mirror with a narrower bandwidth than the erbium s one reflects the backward ASE, which is re-amplified by the erbium fiber. This configuration provides a very efficient optical source, which can deliver more than It is possible to multiplex a single source with three fiber coils of several kilometers. Furthermore this is obtained with less than It is therefore possible to use relatively low power pump These are cheaper lasers and, most important, they can be used without thermal regulation on a very large thermal range. The laser diode quality is also the most important parameter regarding the FOG reliability. 980nm pump lasers are used in submarine telecoms and this technology can provide devices with more than 20 years lifetime in difficult conditions. 4.2 All digital processing Generally, signal processing methods are as important as the physical sensor itself to obtain high performances, and this is particularly true with the FOG. Rate measurement must be as accurate as possible, since it is integrated to obtain a rotation angle, and over a very large dynamic range; a high performance INS like Marins measures rates below /h (for precise initial positioning) and up to 50 /s (while navigating), that is a 80 db dynamic! Furthermore, the interferometer response is sinusoidal, while a gyroscope measurement should be linear. This shows that it is important to get a very linear scale factor for the FOG. It is solved with a closed-loop approach using a phase ramp, which provides a linear and stable scale factor that is independent of optical power and of the optical detection chain gain. This method also enables to count several fringes of the interferometer, expanding the dynamic range beyond the unambiguous measurement range. The most effective way to implement this design is to use a digital phase ramp : a phase ramp is generated by adding phase steps s of a duration of g, the transit time of light through the fiber coil; because of this delay, the induced phase difference is equal to the step value, s. The benefits of this technique are numerous. First the ramp is generated by a Digital/Analog Converter (DAC). A possible 3

4 drawback could be the size of the DAC needed to comply with the dynamic range (over 26 bits!). It can be shown that this is not the case thanks to an averaging process that allows to measure s even if s < LSB, DAC smallest phase step. The condition is that LSB shall be in the linear part of the sinusoidal interferometer response, which is very easily achieved with only 12 bits! This condition also ensures the DAC linearity performance. It is thus noticeable that this digital phase ramp design does not imply the use of very high performance DAC, which is an important advantage in term of cost and implementation. Figure 4: Principle of digital phase ramp feed back with g being the transit time through the coil (5µs/km) 4.3 High performance Integrated Optical Circuit IOC is a 3-in-1 device achieving key functions in the so-called Y coupler configuration. A single integrated circuit acts simultaneously as a beam splitter, a phase modulator and a polarization filter. The symmetry of the Y junction gives naturally a 50% splitting ratio between the 2 Y-branches, and the use of integrated optics ensures both a good stability of the device and low production costs in volume. The device is made of Lithium-Niobate (LiNbO3) because of its very good electro-optic properties. The most efficient phase modulation is obtained for an X-cut Y-propagation with light polarization parallel to the Z-axis (TE mode). The integrated optics design makes very high bandwidth possible. This type of device can be driven up to 40 GHz with a very good linearity of the electro-optic response. The FOG bandwidth is close to 4 KHz, which is almost the only gyro technology capable of such high frequency. Last but not least, the IOC is the polarizer of the Reciprocal Configuration. True single mode (both spatial and polarization) filtering is essential to obtain low bias FOG. If it is easy to obtain good spatial filtering with a single mode fiber (which is the case with the Y junction configuration), it is more difficult to find a polarizer with adequate filtering for high-performance FOG. Proton-exchanged LiNbO3 waveguides fulfill this function with impressive performances. It is thus possible to obtain very good extinction ratio with long enough IOC, above 80dB for ixblue ones. 4.4 Noise reduction ixblue FOG 200, used in the satellite navigation systems and in MARINS has a noise performance of 2x10-4 deg/ h. For very high performance navigation, we want to achieve a goal ten times better. We had therefore to analyze the influence of the different types of noise in FOG technology Shot noise Shot noise is due to the particle nature of light: it is proportional to the square root of the actual power. Figure 5 shows: The signal, a sine function of the modulation depth (or the phase shift), The actual optical power on detector, cosine function of modulation depth (or phase shift), The shot noise, proportional to square root of power, The noise to signal ratio, which is therefore proportional to 1/ P, 4

5 amplitude normalized to 1 amplitude normalized to 1 amplitude normalized to 1 The noise to signal ratio decreases as the modulation depth increases. Tendancy as a function of phase shift (or modulation depth) 1.5 Power 1 Signal shot noise 0.5 Shot Noise / Signal ratio phase shift normalized to p Figure 5 : Shot noise Shot noise is the ultimate theoretical noise reduction limit Relative Intensity Noise (RIN) This noise is due to the light source width which is necessary for «destroying» the parasitic coherent effects, but the source width creates beats between the different pulses within the source spectrum. RIN is proportional to the actual power P. This is why power curve and noise curve are superimposed on Figure 6. The Noise to Signal ratio is therefore independent from power. It decreases deeply when modulation depth increases and reaches zero when modulation depth is p. However, signal is also zero, which makes measurement very complicated! Relative Intensity Noise is, in practice, the noise that limits mostly FOG noise performance. Tendancy as a function of phase shift (or modulation depth) phase shift normalized to p Figure 6: Relative Intensity Noise Power Signal RIN Intensity Noise/Signal ratio Thermal noise This noise is due to the resistance of the transimpedance: the power detector provides current intensity proportional to optical power. In order to be measured, this intensity must be transformed into voltage using a resistor. The thermal agitation of the electrons generates the thermal noise. Thermal noise is independent from the actual power P. The Noise to Signal ratio is therefore proportional to 1/P. It can be seen on Figure 7 that noise to signal ratio is minimal for modulation depth p/2 and that it increases when modulation depth is larger than p/2. Tendancy as a function of phase shift (or modulation depth) phase shift normalized to p Figure 7: Thermal noise Power Signal Thermal Noise Thermal Noise/ Signal ratio Power 5

6 standard deviation (deg/h) Other noise sources The 3 above noise sources are intrinsic to the fiber-optic gyro. They do not come from imperfections of the manufacturing but from physics. There are other sources of noise inside the FOG and they come from optical imperfections or from electronics. These other sources can be reduced by a better design or a better manufacturing process Optimization of global noise We have seen above that noise to signal ratio decreases for shot noise and RIN and increases for thermal noise when modulation depth is over p/2. Therefore it is clear that it is possible to optimize noise level by selecting carefully modulation depth. Hence, we tried to optimize: the light power generated by the light source, the power coming back to the detector, the modulation depth, the reduction of other noise sources in the electronics, the parasitic reflections inside the optical path. With such an optimization, we have reached 0.13 µrd/ Hz with a light source power of 160 µw as the phase noise. This phase noise translates into angular random walk on the FOG according to the FOG sensitivity (depending on length of fiber and mean coil diameter). For a FOG 120 (about 1 km of fiber), this phase noise is equivalent to 3.4x10-4 deg/ hr. For a FOG 200 (several km of fiber), this phase noise is equivalent to 7x10-5 deg/ hr. It can be seen on Figure 8 the noise improvement using the same coil but with optimized parameters. Noise is improved by a threefold factor. Allan variance FOG120 standard design FOG 120 optimized design time (seconds) Figure 8: Noise improvement after design optimization 5 FOG NAVIGATION SYSTEM PERFORMANCE Based on the various models of FOG that were designed by ixblue, different performance navigation systems have been developed. They are known as OCTANS, PHINS, MARINS. They are used in multiple applications including from civilian vessels, navy vessels, submarines, howitzers, mining machines, tunnel drilling machines, deep oil & gas wells, autonomous vehicles (subsea, surface, ground and air) and last but not least satellites. The FOG technology is well suited for space applications since it is extremely reliable and has very long lifetime (>15 years). ixblue FOG have cumulated more than 100 years of use in space without any failure. The navigation systems performance varies from one system to the other depending on the class of FOG used inside. The performance extends from 2 nautical miles per hour to 1 nautical in one day. 6

7 In the following paragraph we concentrate on the very high performance navigation system. A navigation system close to MARINS, standard product has been bought by the US Navy and evaluated under the Foreign Compatibility Testing (FCT) Program. The principal Navy objective of the FCT Program has been to determine the availability of Commercial Off the Shelf (COTS) low cost INS that would fill capability gaps in the performance of existing INS (specifically the existing WSN-7/7A aboard Navy ships). Static and dynamic testing was conducted both at the SPAWAR Atlantic Based Test Site (LBTS) in Little Creek, VA and at the Penn State University/Applied Research Lab Quiet Test laboratory (QTL) in Warminster, PA [11]. The final test that was performed was a long term run in pure inertial mode during more than 13 days. During these 13 days, heading angle was periodically changed so that all directions were faced for at least 24 hours. Figure 9 shows how position error varies versus time. Figure 9 : Radial position error (nm) vs. time The Time RMS position error is 2.6 nautical miles after more than 13 days of navigation with no aiding. This unequalled result with fully strapdown system shows how well FOG performs over long period of time. On longitude, which is the unbounded position drift, it corresponds to a gyroscope bias less than 1.4x10-4 deg/h! 6 CONCLUSION Optical gyroscopes have revolutionized the inertial techniques making possible strapdown navigation systems feasible with strapdown design. Ring Laser Gyro was considered for a long time as the ultimate performance technology. But recent developments in FOG improvements have shown that FOG technology can surpass now RLG technology by more than one order of magnitude. Moreover, FOG technology was usually considered as sensitive to shocks and vibrations. This is no more the case, since products using FOG technology has been qualified on howitzers, 105mm guns and mortars. REFERENCES [1] Post, E.J., Sagnac effect, Review of Modern Physics, Vol. 39, (1967). [2] Arditty H.J. and Lefèvre H.C., Sagnac effect in fiber gyroscopes, Optics Letters, Vol. 6, (1981). 7

8 [3] Sagnac G., Comptes rendus à l Académie des Sciences, Vol. 95, (1981). [4] Fabre C., La limite quantique dans les gyromètres optiques, Revue scientifique et technique de la défense, Vol. 7, (1990). [5] Aronowitz, F., Fundamentals of the Ring Laser Gyro, Optical Gyros and their Application, RTO AGARDograph 339, 3-1 to 3-45 (1999). [6] Lefèvre, H., The Fiber-Optic Gyroscope, Artech House, Boston-London (1993). [7] Ezekiel S., Davis J.L. and Hellwarth R.W., Intensity dependent nonreciprocal phase shift in a fiberoptic gyroscope, Springer Series in Optical Sciences, Vol. 32, (1982). [8] Arditty H.J., Graindorge P., Lefèvre H.C., Martin P., Morisse J. and Simonpiétri P., Fiber-optic gyroscope with all-digital processing, OFS 6, Springer-Verlag Proceedings in Physics, (1989). [9] 30th FOG Anniversary Session, OFS 18 Conference, Cancun (2006). [10] Paturel Y. et al., MARINS, the first FOG navigation system for submarines in Proceedings of Symposium Gyro technology, [11] Carr K. et al., Navy Testing of the ixblue MARINS Fiber Optic Gyroscope (FOG) Inertial Navigation System (INS) in Proceedings of ION PLANS Conference, Monterey,