Technical Papers

» Low-cost Interferometric TDM technology for dynamic sensing applications

JEFF BUSH and ALLEN CEKORICH
7652 Haskell Ave.
Van Nuys, CA 91406

Abstract

A low-cost design approach for Time Division Multiplexed (TDM) fiber-optic interferometric interrogation of multichannel sensor arrays is presented. This paper describes the evolutionary design process of the subject design. First, the requisite elements of interferometric interrogation are defined for a single channel sensor. The concept is then extended to multi-channel sensor interrogation implementing a TDM multiplex scheme where “traditional” design elements are utilized. The cost of the traditional TDM interrogator is investigated and concluded to be too high for entry into many markets. A new design approach is presented which significantly reduces the cost for TDM interrogation. This new approach, in accordance with the cost objectives, shows promise to bring this technology to within the threshold of commercial acceptance for a wide range of distributed fiber sensing applications.

Keywords: TDM, TDM interferometer, TDM interrogator, TDM demodulator, fiber interferometer, fiber sensor array

1. INTRODUCTION

The technology for fiber optic distributed sensing for obtaining high quality dynamic measurements has been "available" for the past 15 years, but hasn't experienced significant market acceptance due to intrinsic high cost of the requisite design elements. The majority of the work performed in this area has been focused either on military surveillance or remote sensing in severe environments (sub-sea or sub-surface for the oil and gas industry) where cost sensitivity is not necessarily a first priority.

These two markets, although substantial (as evidenced by investments in fiber sensor technology over the last two decades) do not nearly represent the overall potential market for distributed sensing of dynamic events. There are many other markets heretofore under reached by technically viable, commercially price acceptable fiber sensor offerings. These include: Structural Monitoring (bridges, dams, pipelines, structures); Large Area Surveillance (multi-zone systems for secure facilities or complexes); Seismic Monitoring (USGS and related); and Communications Security (data line tampering or integrity). There are numerous indicators that such markets would embrace fiber distributed sensing solutions if they were cost effective.

Since CY 2000, Optiphase, Inc. has focused on TDM designs for distributed fiber sensing technology and has been methodically developing low-cost solutions for the elements which comprise the interrogator for such systems. A number of breakthroughs have been made which effectively enable the desired cost reductions.

We report on these accomplishments, outline the elements of the design, provide design insight, and represent projections as to the measurement performance capabilities of this reduced cost technology. Further, we anticipate initial commercialization of this technology in CY 2005.

2. INTERFEROMETRIC INTERROGATION

The key components of an interferometric interrogator are shown in figure 1 for a single, remote fiber interferometric sensor.

1. Laser: Typically are single longitudinal line devices such as DFB, External Cavity Laser (ECL), diode pumped YAG or fiber-laser.

2. Modulator: This device either modulates the wavelength, frequency or phase of the light in a precise manner corresponding to its input current or voltage drive signal.

3. Receiver: In its simplest form, this will be a single photodiode configured in a trans-impedance amplifier circuit. However, most remote fiber interferometers constructed of single mode fiber are subject to random interferometric visibility fluctuations 1, 2. In these cases, methods cited by Frigo et al 3are required, which necessitate the use of a Polarization Diversity Receiver (PDR) implementing three simultaneous polarized receivers (at 0, 60 and 120 degrees). Also see section 3.5.

4. Demodulator: This element controls the modulation of the interferometer, the signal sampling and the signal processing to extract the phase data (demodulation) and the flow of data. This discussion is limited to open-loop methods, which employ external modulation, as this is the only type which is applicable to multi-channel interrogation. Further it is assumed that the demodulation approach is capable of measuring optical phase continuously from very small values (sub fringe) to very large values (multi-fringe) using digital processing approaches4. Here, effective interrogation is accomplished using the following controls:

• A modulation servo is used to precisely control the level of modulation.
• For modulation signals which are continuous, a modulation servo is implemented to synchronize the digital sampling process.
• When implemented with a Polarization Diversity Receiver, parallel demodulation is implemented such that “receiver swapping” does not cause a discontinuity in the demodulated output.

2.1. TDM Considerations

This work is concerned with multi-channel interrogation in a Time Division Multiplexed (TDM) format. The generic configuration for the TDM interrogator is shown in figure 2. One difference is that the single channel sensor is replaced by an array of sensors, where each sensor reflects or returns a portion of the input light back to the receiver in a “timed” manner subject to its location in the array. The other difference is the placement of a “pulser” after the CW source. The pulser is implemented by devices capable of providing fast optical switching (10’s of ns or less) having a high optical extinction ratio (preferably 60 dB or better).

The general concept of TDM interrogation is that the optical pulse output to the array is of such a time period (τ) that sensor returns overlap each other to cause interference. This concept is depicted in figure 3 where one input pulse is sent to the array having six sensors showing how return pulses will overlap to form distinct interference patterns. These patterns are displayed as the shaded regions (bottom of figure 3) indicating which pulse returns are overlapping.

Note the initial and terminal returns do not cause interference at their extreme points.

The scope of TDM sensing encompasses two types of sensing arrays, principally Finite Impulse Response (FIR) and Infinite Impulse Response (IIR) type designs. Figure 4 depicts the FIR type designs and figure 5 depicts the IIR type designs.

The Michelson type of architecture shown in figure 4 shows bidirectional travel of the light to and from the sensors. The Mach-Zehnder sensors are forward propagating, and hence will need a separate return fiber. In figure 5, the top diagram indicates use of wavelength independent reflectors and the bottom diagram implements Fiber Bragg Gratings, which are wavelength selective elements. For both configurations, it is assumed that reflections will be low level (perhaps 0.5% to 5%).

3. STANDARD TDM INTERROGATOR CONFIGURATION

The most effective approach for TDM interrogation is that shown in figure 6. It draws on designs originated at the Naval Research Lab and recently outlined in a recent sensor overview publication5. More exotic and esoteric designs exist, but are omitted from this general discussion. The sensor array shown in this diagram is interchangeable with any of the previously described variants. The elements shown are briefly outlined.

3.1. CW Laser

This must be a coherent device, operating with a single longitudinal line. Design choices in order of increasing cost and decreasing phase noise are: 1) DFB; 2) ECL (external cavity laser diode); 3) fiber laser; and 4) diode pumped YAG. Currently, system designs with cost constraints will be limited to DFB or ECL designs.

3.2. Pulser, Integrated Optic Circuit Pulse Generator

This is a semi-custom (LiNbO3) integrated-optic device which has two serial Mach-Zehnder elements, each with modulation electrodes and optical tap circuits. It functions as a switch by driving the modulation electrodes from the “off” state to the “on” state. The two serial MZ devices are required to obtain a useful extinction ratio, which may range between 50 to 70 dB for high performance devices. The most effective designs for these devices with respect to achieving a high extinction ratio involve the Annealed Proton Exchange (APE) technique (for integrated optics) which yields a single polarization waveguide. This requires the CW laser to emit light into a PM fiber. Currently the approximate cost of these devices for high extinction switch operation (50-65 dB) exceeds $10,000 USD.

APE LiNbO3 devices are known to drift, as their optical phase bias is sensitive to both temperature and static electric fields. Active control circuits (optical pickoff based servos for each MZ element) are required to maintain optical extinction ratios greater than 50 dB. Succinctly stated, these are not trivial circuits, as they must be made to operate simultaneous with the switching operation.

Generally, the Dual MZ IOC and pulse control circuit is the most expensive item in the TDM interrogator, and correspondingly is the highest cost driver for TDM interrogators.

3.3. Compensator

This device is sometimes known as the “compensating interferometer”. For the configuration shown in figure 6, it is used to generate a double pulse from a single input pulse. The short path propagates through a high-speed phase modulator (depicted as “IOC MOD”) and the long path through a delay, which is ideally equal to the input pulse width. The main expense of the compensator is the IOC phase modulator. It has remained at the few thousand dollar level for the past decade. It is also an APE design to insure a precise phase / voltage response.

3.4. Er Amp

This is an Erbium Doped Fiber Amplifier (EDFA) commonly used for telecommunications applications. It is operated at a somewhat lower than average pump drive level to ensure the output pulses do not experience significant droop. In years past, these devices sold for up to $20,000 (USD). Today, after becoming a mainstream component in telecommunications systems, their price has dropped to well below $5,000 (USD).

3.5. Polarization Diversity Receiver.

This is a device, which splits the input light to three polarized receivers. These receivers have polarization orientations of 0, 60, and 120 degrees. The PDR design developed for interferometric interrogation is based on a 1X3 or 3X3 fused biconic taper coupler with short leads to polarized photodiodes and is depicted in figure 7.

3.6. “Standard TDM Demodulator

This element is depicted in figure 6 as “TDM DEMOD.” The demodulation process is based on US Patent 5,903,3506 entitled “Demodulator and Method useful for Multiplexed Optical Sensors.” Particularly, this is a 5 step modulation approach with each step separated by π/2 radians (M0 through M4 in figure 8). The generalized modulation scheme is shown in figure 8. Although only four steps are required to obtain quadrature terms, the additional step is implemented to allow for a pickoff signal to be generated which corrects for errors in the modulation level.

As can be seen, in figure 8, the modulation levels are made short so as to fit completely within a single light pulse. Use of Integrated optic phase modulators with rise times on the order of a few ns allows for very short pulses if needed.

Figure 9 shows how the modulation scheme is made for each return interference signal and figure 10 shows the associated sampling scheme which (S0 – S4 for each pulse) is aligned to the modulation scheme. If the samples are labeled S0 to S4, the quadrature demodulation terms are simply defined as follows

 

Similarly, the modulation pickoff terms are defined.

Note that there are two modulation error pickoff terms. SM is used when the term SR has a higher magnitude than CR and the CM term is used when CR has a higher magnitude than SM. This strategy assures that the modulation depth correction servo has a relatively constant loop gain independent of the quiescent phase of the interferometer (or sensor) used for the modulation correction servo.

It is worth mentioning that this TDM demodulation approach accomplishes one demodulation data point per sensor per input pulse. This is a faster (and potentially more accurate) sampling approach than other known TDM techniques which appear to require a number of input pulses to gather enough information to obtain the quadrature signals to obtain one demodulation point.

3.8. Current Design TDM Hardware

The current design for the TDM demodulator implements the following elements:

• COTS DSP BOARD (TMS320C6X based);
• COTS ADC Board (3 Parallel ADCs for measurement of PDR output);
• Custom Board with FPGA and fast DAC for Modulation and Timing Control;
• CPCI Host Computer (loads and receives data over PCI Bus, serves as user interface);
• CPCI Chassis.

Originally this type of design approach was deemed as a low-cost approach to develop an integrated TDM hardware/firmware design. This did enable quick turn-around for design modifications or revisions. However, if one were to manufacture TDM interrogation systems, the recurring costs for this hardware set becomes significant.

4. COST REDUCTION DESIGN

The “standard design” TDM interrogator, upon review, provides a comprehensive solution to the technical challenge of TDM demodulation. There are a number of high cost elements in this design however, which preclude this technology from entry into many of the distributed sensing markets (principally structural monitoring and large area intrusion detection). The major elements which drive the cost are:

1. Integrated optic dual MZ device for optical switching (pulsing);
2. The COTS based demodulation hardware;
3. The Integrated Optic Phase Modulator;
4. PM fiber interfaces and PM coupler.

Starting in CY 2000, Optiphase initiated a program to reduce the cost of the “standard design” TDM interrogator to the point where it would be cost effective for the structural monitoring and intrusion detection markets. The objectives of this plan were set as follows.

Main Objective:

Reduce cost as much as possible without significantly degrading function and performance.

The sub-objectives which followed are:

• Elimination of Integrated Optic Components;
• Elimination of PM Components;
• Change demodulation approach (for compatibility with slower speed modulators);
• Replace the demodulation electronics system with a single circuit board solution.

To date we have completed the first three objectives and are making significant progress on the fourth. Details of the results are presented.

4.1. IOC Pulser Replacement

First Investigation: AOM Modulator

This task originally looked at fiber-in, fiber-out Acousto-Optic Modulators (AOM). A design was developed and evaluated which actually yielded a high performance device. Key characteristics were:

• Operational Wavelength:                                        C –Band
• Input / Output:                                                       Single Mode fiber (SMF-28)
• RF Excitation                                                        200 MHz
• Rise / Fall Time                                                     20 ns
• Extinction Ratio                                                     67 dB
• Loss                                                                     -5.6 dB
• Expected Mfg Cost                                                approx $5K, small qty (USD)

Although we were encouraged by the design’s performance and the attractive feature that this device had (an extinction ratio that was relatively wavelength insensitive) there were a number of features that made this choice non-ideal. These were: 1) The RF switching circuitry was very sensitive to stray impedances and varying loads; and 2) the operational and storage temperature range was limited due to crystal CTE mismatches;

This approach was abandoned.

Second Investigation: SOA device

In the 2001 – 2002 time frame, Semiconductor Optical Amplifier (SOA) devices were becoming widely available and at significantly lower costs than in previous years. These devices are well known for their high speed switching characteristics for various telecommunications applications. Further, these devices had other desirable characteristics, such as:

• Rise time of 1 – 5 ns;
• Output will saturate (good for “flat-top” pulse);
• Device has gain rather than loss (relaxes source power requirements);
• Anticipated drive circuitry would be simpler than AOM drive circuit;
• Unpolarized devices were available;
• Low Cost.

There were a few concerns regarding the implementation of these type devices which related to noise figure (cited in literature and device data sheets as 6 to 9 dB) and chirp. Regarding the noise figure, it was determined that the main contributions come from ASE and beat-noise, and both can be mitigated if needed. The ASE can be eliminated by implementation of optical bandpass filters (commonly available) and the beat noise is only noticeable when there are many wavelengths being amplified (such as DWDM systems). Regarding chirp, some testing (see figure 11) was performed indicating this was a negligible effect. This test involved pulsing a sample SOA (C-Band with DFB) and monitoring the output signal through a path mismatch interferometer. The results indicate no observable chirp. Note: the leading edge peak shown is related to the pulse generator and not chirp.

Some sample SOA devices were then evaluated for other pulse parameters with the following results:

• Rise time faster than our drive electronics (est. 1-2 ns);
• Flat-top pulse in saturation;
• Extinction greater than 70 dB.

Further testing using SOAs in pulse mode in laboratory TDM interrogator arrangements revealed that these devices perform as good or better than the Dual MZ IOC devices.

Conclusion: SOA was selected as the pulser element.

4.2. Compensator Design Change

The compensator design was changed to eliminate PM components and incorporate a piezoelectric fiber stretcher. This design change is reflected in figure 12.

This new design implements an all SMF Michelson arrangement to enable use of a Faraday Reflector Mirror (FRM) for the modulator leg. This insures birefringence modulation caused by the piezoelectric modulation approach is held to an absolute minimum. The delay line is shown with a simple reflector (in accordance with the low cost objectives).

This design change also results in a substantial cost reduction.

4.3. New Demodulation Approach: “Dual Slope

The original modulation / demodulation approach (discrete modulation steps) previously discussed is incompatible with a piezoelectric fiber based modulator as such devices are only capable of working with continuous signals. A new demodulation approach implementing continuous modulation drive signals is now required. This new approach is based on US Patent 6,778,7207 entitled “Dual Slope Fiber Optic Array Interrogator. It implements a sine wave drive where the modulation slope at the zero crossings (linear portion of the sinusoidal drive signal) is 3π/τ radians per second. The sampling scheme is close to identical as the original demodulation approach. Here 5 samples of the phase signals are made during the pulse width τ (for each channel in the array). A simple depiction is provided in figure 13.

The sampling approach implements the same quadrature and modulation error signal processing as described for the five step modulation / demodulation approach. The difference in this case is that the sampling occurs during the constant transition of the modulation phase. Sampling is performed on both the up and down slopes of the modulation signal. This “dual-slope” sampling approach enables modulation timing corrections to be made “on the fly” to ensure correct alignment of the pulse generation with respect to the linear portions of the modulation drive.

This approach has been demonstrated in the lab by making appropriate firmware modifications to the standard (5 step ) demodulation hardware. Performance measurements (distortion and noise) evaluated show no changes when configured with a 10 element IIR array (lab simulator with 100 m length sensors) with DFB and ECL sources. Average phase noise measurements observed are as follows.

• External Cavity Laser (C-band)                         50 urad/rt-Hz
• DFB Laser (C-band)                                        300 urad/rt-Hz

These noise levels are the same as experienced with the dual MZ-IOC and original compensator and are wholly due to the coherent Rayleigh backscatter from the IIR array topology implemented. Here, narrow line width lasers exhibit lower noise.

4.4. Dual Slope Interrogator Block Diagram

The new low cost interrogator design is shown in figure 14.

4.5. Considerations for Large Channel Count (IIR Arrays)

The concept of a low cost interrogator to meet the needs of price sensitive markets (intrusion detection, and structural monitoring) is not valid unless the sensor arrays are also low cost. This likely eliminates FIR arrays (fig 4) leaving only IIR array types (fig 5) as candidates. A recent unpublished study conducted by Optiphase assessed the likely maximum channel count per array assuming one array is formed by multiple sensors on a single fiber, all operating at the same wavelength. The conclusion of this study indicated lower than expected channel counts as indicated below

• High Performance Sensing                                            6 – 8 channels per λ or per fiber
  • Low Noise, ~ 25 urad/rt-Hz average noise
  • Low Distortion, THD < 0.2%
  • Low Crosstalk
• Moderate-Performance Sensing                                     16 channels per λ or per fiber
  • Average noise ~ 300 urad/rt-Hz
  • Moderate Distortion, THD < 1%
  • Moderate Crosstalk

Many of the sensing applications considered would require substantially more channels and the Interrogator and array topology would have to change to accommodate the higher channel counts. Two approaches may be implemented as shown in figure 15. The first involves multiple fiber arrays; the second involves multiple wavelengths.

Both of the configurations shown in figure 15 despite the added elements (receivers, WDMs and multi-wave sources) still present cost effective solutions given the additional channel capability. Further if the wavelengths are relatively close to each other (as in DWDM systems), common singular items in each TDM interrogator will be:

• Pulser (can accept multiple wavelengths);
• Modulator / Compensator
  • Active control for modulation depth for one wavelength
  • Passive control for modulation depth for other wavelengths;
• Demodulator (needs additional digitizers to some degree, but can process high channel counts).

This indicates that economies can still be attained at higher channel counts.

4.6. Eliminate Reflectors in the Array Entirely

Despite the economies identified in section 4.5, there are still sensing applications that these approaches cannot address when high channel counts are mandated . These will likely be cost-sensitive intrusion detection systems and very large structure monitoring systems. These require very low cost per channel but high measurement performance is not required.

For these cases a different sensing technique may be implemented where the fiber’s inherent Rayleigh backscatter properties serve as the “reflectors” for the array 8-10. Optiphase has performed preliminary testing with these approaches and has determined the TDM Interrogation techniques (complete with demodulation) outlined in this paper are compatible with such sensing implementations. The advantages of this coherent Rayleigh sensing approach (if somewhat degraded signal performance is acceptable) are:

• Very long sensor arrays 10+ km);
• High channel count (100’s +);
• High spatial resolution (when short pulses are used);
• Simple sensor designs possible (can be just a long fiber cable)

Presently, Optiphase’s research in this area is limited to the determination of the ability to interrogate (with demodulation) this type of sensor. Initial results show that this is possible.

4.7. New Demodulation Hardware

This is still a work in progress. The anticipated schedule is to have a prototype single board demodulator to be “functional” (without USB 2.0 interface) by December 04 time frame and a first article board demodulator by the second quarter 2005. This design is based on the TMS320C6713 DSP.

5. CONCLUSION

A cost effective TDM interrogation approach for dynamic interferometric sensing has been identified and demonstrated. Design and development activities are now underway related to the commercialization of this technology.

6. REFERENCES

1. S. K. Sheem and T. G. Giallorenzi, "Polarization effects on single-mode optical fiber sensors," Applied Physics Letters 35, 914 (1978).

2. D. W. Stowe, D. R. Moore, and R. G. Priest, "Polarization Fading in Fiber Interferometric Sensors," IEEE Journal of Quantum Electronics QE-18(10), 1644-1647 (1982).

3. N. J. Frigo, A. Dandridge, and A. B. Tveten, "Method for reduction of polarization fading in interferometers," 4,653,915 (April 12, 1985 1987).

4. A. C. Cekorich and J. G. Davis, "Demodulator and Method for Interferometric Outputs of Increased Accuracy," 6,556,509 (Nov. 8, 1999 2003).

5. C. Kirkendall and A. Dandridge, "Overview of high performance fiber-optic sensing," Journal of Physics D: Applied Physics 37, R197-R216 (2004).

6. I. J. Bush and A. C. Cekorich, "Demodulator and Method Useful for Multiplexed Optical Sensors," 5,903,350 (Jun. 18, 1998-1999).

7. A. C. Cekorich and I. J. Bush, "Dual Slope Fiber Optic Array Interrogator," 6,778,720 (Apr. 2, 2002-2004).

8. K. N. Choi, J. C. Juarez, and H. F. Taylor, "Distributed fiber-optic pressure/seismic sensor for low cost monitoring of long perimeters," presented at the Unattended Ground Sensor Technologies and Applications, Orlando, FL, April 2003, 2003.

9. R. J. Posey, G. A. Johnson, and S. T. Vohra, "Strain sensing based on coherent Rayleigh scattering in an Optical Fiber," Electronics Letters 36(20), 1688-1689 (2000).

10. H. F. Taylor and C. E. Lee, "Apparatus and Method for Fiber Optic Intrusion Sensing," 5,194,847 (1993).

Reprint: Proceedings of SPIE, OpticsEast 2004, Fiber Optic Sensor Technology and Applications III, paper 5589-19, Courtesy Optiphase, Inc. www.optiphase.com