the catholic university of america monostatic all-fiber
TRANSCRIPT
THE CATHOLIC UNIVERSITY OF AMERICA
Monostatic all-fiber LADAR systems
A DISSERTATION
Submitted to the Faculty of the
Department of Electrical Engineering and Computer Science
School of Engineering
Of The Catholic University of America
In Partial Fulfillment of the Requirements
For the Degree
Doctor of Philosophy
By
Jeffrey H. Leach
Washington, D.C.
2016
Monostatic all-fiber LADAR systems
Jeffrey H. Leach, Ph.D.
Director: Scott Mathews, Ph.D.
Many applications require a LADAR system smaller in size and more power efficient than
those using a polygon, galvo, or Risley-prism beam scanner. A fiber-coupled, monostatic
LADAR system which transmits and receives through the same aperture has many advantages.
These advantages include low cost, easy optical alignment, small size, and low weight. Optical
alignment of the system is greatly simplified since there is no parallax between transmitted and
received beam paths. The direction of the received light is inherently aligned with that of the
transmitted beam. Multiple alignment steps of bulk optical components are one of the major
reasons many laser systems are expensive. Using an all-fiber approach the optical alignment is
simplified by automated splicing of the fibers. A 1-D LADAR consisting of a stationary
monostatic rangefinder with real-time pulse processing is first demonstrated. Then a 3-D
scanning LADAR is demonstrated. A vibrating fiber cantilever tip that is used to scan the laser
beam is mounted on a resonant piezo-electric lead zirconium titanate (PZT) stripe actuator that
requires very little power. A position sensing detector (PSD) is needed to sense the fiber
position since the motion of the scanned fiber may not be fully predictable. The PSD operates in
a two-photon absorption mode to accurately measure the fiber tip position for each laser pulse,
with very low insertion loss.
ii
This dissertation by Jeffrey H Leach fulfills the dissertation requirement for the doctoral degree in Electrical Engineering approved by Scott Mathews, Ph.D., as Director, and by Nader Namazi, Ph.D., and Lew Goldberg, Ph.D. as Readers.
__________________________ Scott Mathews, Ph.D., Director
__________________________ Nader Namazi, Ph.D., Reader
__________________________ Lew Goldberg, Ph.D., Reader
iii
Table of Contents
1.0 Introduction .............................................................................................................................. 1
1.1 Related LADAR systems ..................................................................................................... 3
2.0 Fused-fiber coupler multiplexer ............................................................................................... 7
2.1 Fiber MUX modeling ........................................................................................................... 8
2.2 Fiber MUX fabrication ...................................................................................................... 11
3.0 Monostatic laser rangefinder system ..................................................................................... 22
3.1 Block diagram .................................................................................................................... 22
3.2 Signal processing ............................................................................................................... 24
3.3 Optical link parameters ...................................................................................................... 24
3.4 Results ................................................................................................................................ 25
4.0 Scanning LADAR system ...................................................................................................... 29
4.1 Block diagram and system overview ................................................................................. 29
4.2 Vibrating fiber scanner ...................................................................................................... 31
4.3 Position sensing detector.................................................................................................... 40
4.4 Data acquisition ................................................................................................................. 45
4.5 Results ................................................................................................................................ 49
4.5.1 Beam analysis ............................................................................................................. 49
4.5.2 3-D target measurements ............................................................................................ 55
5.0 Detector and optical reflection considerations ....................................................................... 59
iv
5.1 Zero-time-delay (T0) pulse return ...................................................................................... 59
5.2 Detector .............................................................................................................................. 62
6.0 Conclusion ............................................................................................................................. 65
Bibliography: ................................................................................................................................ 68
Table of Figures
Figure 1. A bistatic system versus monostatic system. .................................................................. 2
Figure 2. Block diagram of all-fiber monostatic rangefinder system. ........................................... 2
Figure 3. Winner of the 2005 DARPA Grand Challenge driverless car competion. ..................... 5
Figure 4. Monostatic System Multiplexer and transmitter/receiver. ............................................. 7
Figure 5. Model of the MUX layout and power transfer. .............................................................. 9
Figure 6. Model of the MUX power transfer. .............................................................................. 10
Figure 7. Diagram of the experimental setup used to create the MUX. ...................................... 11
Figure 8. MUX with an off-the-shelf APD and lens. ................................................................... 12
Figure 9. CO2 laser mirror scanner control. ................................................................................. 13
Figure 10. Fused fiber couplers. .................................................................................................. 14
Figure 11. Coupler performance tradeoffs. .................................................................................. 15
Figure 12. End of fiber taper with CO2 system from 125 µm to 24 µm, made in-house. ........... 16
Figure 13. Coupler power transfer measurement. ........................................................................ 18
Figure 14. Block diagram and physical generation of a monostatic launch field that simulates the
return received signal. ................................................................................................................... 19
Figure 15. Far-field characterization of scanned fiber with fixed detector. ................................. 20
v
Figure 16. Far-field output angular distribution of the MM output fiber in the fused MUX
coupler. The dashed line indicates the 5%-of-max boundary. ..................................................... 21
Figure 17. Block diagram of all-fiber monostatic rangefinder system. ....................................... 22
Figure 18. Signal traces showing residual T0 signal and return pulse (red) and baseline with laser
off (blue). ...................................................................................................................................... 26
Figure 19. Multi-pulse data analysis. (A) The enlarged lower vertical axis clearly displays the
plateaus in the data. (B) The corresponding histogram near 50 m for 3.3 µJ is displayed. .......... 28
Figure 20. All-Fiber scanning LADAR design. ........................................................................... 30
Figure 21. Scanning fiber photo micrograph. .............................................................................. 32
Figure 22. Normalized PZT excitation signal. ............................................................................. 33
Figure 23. Fiber scanning assembly and picture of scanning fiber.............................................. 34
Figure 24. PZT stripe characterization. ........................................................................................ 35
Figure 25. Biaxial fiber oscillation. ............................................................................................. 36
Figure 26. Simulated Lissajous scan. ........................................................................................... 37
Figure 27. Lissajous scan pattern taken with a SWIR camera at 0.2 s exposure (calibration patch
in center). ...................................................................................................................................... 37
Figure 28. Model demonstrating strong single resonant frequency excitation at 290 Hz. .......... 38
Figure 29. Maximum scan before clipping of the TX/RX lens occurs. The figure is not to scale.
....................................................................................................................................................... 39
Figure 30. Scan voltage versus fiber tip displacement. ............................................................... 39
Figure 31. 2-D position sensing silicon detector for 1.5 µm pulses. ........................................... 40
Figure 32. 2-D position sensor signals. (1 µs / div) ..................................................................... 41
vi
Figure 33. Lissajous scan pattern for a (A) 200 ms camera exposure and (B) a normalized
computer display of the PSD signals in a 400 ms interval. .......................................................... 42
Figure 34. Data taken on PSD signal strength by varying the peak power of the laser—a result of
changing the EYFA pump current. ............................................................................................... 43
Figure 35. (A) On-axis total reflectivity measurement of the silicon in the PSD with a 1560 +/-
10 nm source. (B) On-axis reflectivity of each surface of the PSD. The scale ranges from 0% to
2% reflectivity. .............................................................................................................................. 44
Figure 36. Side view of the PSD structure and interface reflectivities of individual surfaces at
1.55 µm. ........................................................................................................................................ 45
Figure 37. Block diagram of system signal paths. Data from optical sensor inputs is in red. .... 46
Figure 38. LADAR acquisition system synchronization. The thicker arrows appear in Figure 37.
....................................................................................................................................................... 47
Figure 39. Sequence acquisition timing diagram. ........................................................................ 48
Figure 40. Horizontal fringes using a shear plate confirm collimation for on-axis. .................... 49
Figure 41. Oscilloscope return pulses (100 mV/vertical division and 2 ns/ horizontal division)
from 7.9 m, separated by 3.6 ns. ................................................................................................... 50
Figure 42. Images (at range of 8 m) and models (at range of 100 m) used in a field curvature
analysis with the laser focused at the center (upper tiles) and the edge of the scan (lower tiles).
Due to experimental limitations, the edge-of-scan image was taken for 1.2 mm offset rather than
1.5 mm offset used in the model. .................................................................................................. 52
Figure 43. Non-sequential Zemax LADAR model incorporating fiber offset, and tilt for system
optimization. Rays change color as they pass into a new medium. ............................................. 53
vii
Figure 44. Sequential Zemax model demonstrating lens field curvature effects by focusing
diffuse light onto a curved plane. .................................................................................................. 54
Figure 45. 3-D Scanning (Monostatic) LADAR point-cloud output at 8 m with gated out
background. ................................................................................................................................... 55
Figure 46. 3-D Scanning LADAR point cloud at 26 m. The left pedestal is the same used in the
previous figure and the right pedestal consists of four square foam slices, ranging from 7.6 cm
(3”) to 30.5 cm (12”) wide, each 1 cm thick. ................................................................................ 56
Figure 47. Lissajous scan pattern taken with a SWIR camera for two different mounted scanning
fibers. ............................................................................................................................................ 57
Figure 48. Lissajous scan pattern for a (A) 400 ms camera exposure and (B) a normalized
computer display of the PSD signals in a 400 ms interval. .......................................................... 58
Figure 49. The normalized effect of cleave angle versus reflected light. .................................... 60
Figure 50. Varying the PSD to scanning fiber distance effects the T0 pulse due to PSD
reflections. ..................................................................................................................................... 62
Figure 51. High NA, high bandwidth, 125 µm back illuminated detector with 95% estimated
fiber-to-APD coupling efficiency. ................................................................................................ 64
Figure 52. Simulations demonstrating the frame time can be halved and angular resolution
maintained by doubling the laser pulse rate and average power. ................................................. 67
1
1.0 Introduction
Typical Laser Detection and Ranging (LADAR) systems are bulky and expensive, with high
input power requirements [1-6]. The goal is to design a small portable LADAR system that can
be used for unmanned air and ground vehicle navigation. The monostatic all-fiber LADAR
system of this dissertation consists of a typical fiber laser, a multimode fiber coupler multiplexer
(MUX) whose core transmits the laser from a pulsed fiber laser and whose cladding receives a
return pulse and couples it into a high bandwidth fiber coupled avalanche photodiode (APD)
detector, a double-resonant cantilever-mounted flattened fiber that scans in a Lissajous pattern, a
silicon 2-D position sensing detector (PSD), a scan lens, and two analog-to-digital (A/D) systems
to analyze the data. The key novel components of the system are the MUX, the lead zirconium
titanate (PZT) fiber scanner, and the silicon PSD.
Using a single common aperture for transmit (TX) and receive (RX) optical paths instead of
two apertures makes for a smaller system. The former type of system is called monostatic, and
the latter bistatic. Previously, in [5], a shared aperture was used for the TX and RX beam paths
so that the device used a single small polygon scanner to build an image. Another system [7]
also used a single aperture to allow scanning with a micro electromechanical system digital
mirror deflector. A commercially available system also uses a shared aperture and scans with a
rotating mirror [8]. Typical LADARs [9] are bi-static, with separate TX and RX optics that
require precise alignment relative to each other.
2
A block diagram comparing a bistatic system to a monostatic system is shown in Figure 1. A
monostatic system has the advantages of being smaller, lighter, and more compact compared to a
bistatic system. No alignment to correct for parallax is needed in a monostatic system since the
TX and RX optics share a common aperture and field of view (FOV). This leads to simpler
fabrication and therefore to a lower cost system. Only a single lens is needed for TX and RX
functions. In addition, in the monostatic system the receiver FOV overlaps the area illuminated
by the transmit beam at all distances from the system, whereas in a bistatic system the overlap
occurs only at a limited range of distances from the system.
This dissertation first focuses on one of the key elements of an all-fiber monostatic system,
shown in Figure 2; an all-fiber multiplexer that selectively separates the received and transmitted
Figure 1. A bistatic system versus monostatic system.
Laser
Detector
Bistatic System Monostatic System
Laser and Detector
Figure 2. Block diagram of all-fiber monostatic rangefinder system.
AR-coated Perpendicular Cleave
1.5 µm Pulsed Fiber Laser
APD Receiver Fiber MUX
(TX/RX) lens
Signal Processing
Isolator
3
beams. We then describe the use of this multiplexer to demonstrate a static (non-scanning)
rangefinding system, or a 1-D LADAR system, shown in Figure 2. Later in the dissertation, the
fiber MUX and other critical components, such as a compact fiber scanner and semiconductor
position sensor are used to implement an all-fiber 3-D scanning LADAR. The operation of the
3-D LADAR and description of all critical components will be given. Both systems operate at
the eye-safer wavelength of 1550 nm and have a nominal optical hazard distance of 0 m.
1.1 Related LADAR systems
LADAR systems are currently incorporating the advances and techniques of microwave
RADAR systems and achieving higher angular resolution and range precision due to their shorter
wavelengths and pulse durations. LADAR applications include foliage penetration, target
identification, terrain mapping, adaptive cruise control, and unmanned vehicle navigation (air
and ground).
LADAR systems can be coherent or incoherent. An example coherent system has a
frequency swept laser signal that is split into a ranging signal and a reference signal. The
returned ranging signal is mixed with the reference signal and then optically detected. Signal
processing of the fast Fourier transform of the detected signal results in the range to a target. An
incoherent system emits light whose correlation length is too short to allow efficient optical
mixing of received light with a local oscillator. Only intensity detection is feasible. In an
incoherent LADAR system ranges are calculated from the time-of-flight difference between the
4
transmit and return pulse. A majority of the commercial LADARs are incoherent as is the
LADAR system that this dissertation focuses on.
In many systems, the time-of-flight measurement can incorporate first return pulse only,
multiple returns, or last return only. Digitization of the full return wave-form can also allow
further signal processing advantages and aid in additional feature extraction. Detectors for these
systems can use a single detector, a detector array, or a focal plane array of detectors. These
detectors can be based on PIN diodes, linear-mode APDs, Geiger-mode APDs, and
photomultiplier tubes. The LADAR system in this dissertation uses a single linear-mode APD
detector; it has been operated in first-return mode, but can be configured for multiple returns.
A majority of LADAR systems are scanned, either by a moving platform, or a MEMS,
polygon, galvo, or Risley-prism beam scanner. There has been recent focus on scanning
electronically by using spatial light modulators.[10] This LADAR system uses a double-
resonant cantilever-mounted flattened fiber that scans in a Lissajous pattern
3-D flash LADAR systems illuminate a scene with a single laser pulse and then measure the
varying time of flights using an imaging array of detectors. Advanced Scientific Concepts Inc.
and Voxtel Inc. are both flash LADAR camera manufacturers. Military operational LADAR
systems such as the ALIRT and HALOE system have also been configured to operate in this
manner.[2] Some manufacturers such as Spectrolab and Princeton Lightwave, Inc. use Geiger-
mode APD arrays in their cameras.
There are many commercial LADAR manufacturers and integrators (e.g. Hokuyo Automatic
Co, SICK, Velodyne, Bridger Photonics Inc., and RIEGL). Other companies such as Phantom
Intelligence, Ibeo, and TriLumina are trying to produce inexpensive LADARs. Some of these
5
systems also incorporate advanced signal processing by use of synthetic apertures and waveform
coding of the signals.
Many LADAR systems work well for slow unmanned ground vehicles.[11] The Velodyne
LADAR systems were used heavily in the 2007 DARPA Urban Challenge for autonomous
vehicle navigation. Frequently, multiple LADAR systems are used together to provide
autonomous navigation. Multiple SICK LADAR systems were used on the winner of the 2005
DARPA Grand Challenge vehicle as seen in Figure 3.
Most LADAR systems consist of multiple bulk optics that need precision alignment. The
alignment of bulk optics tends to be a major cost for laser based systems. Assembly of an all-
fiber LADAR system is simplified to the use of readily available automated splicing systems.
A majority of the LADAR systems are bistatic instead of monostatic. The bistatic separation
causes parallax. The TX and RX beams must be aligned at the maximum distance of the system.
Figure 3. Winner of the 2005 DARPA Grand Challenge driverless car
competion.
LADAR Systems
6
The receiver FOV is increased until overlapping the TX beam for the minimum distance needed.
The larger FOV removes most of the parallax error, but can add a significant amount of solar
background to the received signal. Systems that are monostatic typically use a shared aperture or
a beam-splitter. The conventional shared aperture approach has the disadvantage of an occluded
received spot occupied by the transmit laser. The beam-splitter approach typically has a 50%
loss for the transmit beam and for the receive beam, contributing -6 dB to overall system loss.
Savage and H.N. Burns Engineering [5] use a shared aperture for the transmit and receive beam
paths so that a single small polygon scanner can build an image. The NIST Fandango system [6]
also uses a single aperture, scanned with a micro electromechanical system digital mirror
deflector. The Sick LMS 200 also uses a shared aperture and scans with a rotating mirror.[8]
Stann et al. [9] use a bistatic system in their LADAR and therefore have twice as many optics.
All of these systems require alignment of both receive and transmit optics.
This dissertation describes the first demonstration of a monostatic fiber-based LADAR that
has low power consumption with high precision. In addition, the all-fiber LADAR configuration
enables a compact and lightweight system with potentially low cost. In our implementation, the
receiver FOV is inherently co-aligned with the transmitted beam, and all components are fusion
spliced, making the system structurally insensitive to environmental perturbations.
7
2.0 Fused-fiber coupler multiplexer
The MUX, shown in Figure 4, is a key enabling technology for both an all-fiber monostatic
rangefinder and an all-fiber scanning LADAR. The MUX’s development is discussed prior to
detailed descriptions of the all-fiber monostatic LADAR systems. This MUX is an all-fiber
TX/RX structure which reduces cost, size, and weight, and was developed and fabricated at
Night Vision and Electronic Sensors Directorate facilities. The fused fiber MUX structure
consists of a double-clad fiber (DCF) that is transversely fused to a multi-mode (MM) fiber with
a 105 µm diameter and 0.47 numerical aperture (NA). The DCF fiber has a near-single-mode 12
µm diameter, NA = 0.12 core, an NA= 0.47, 90 µm inner cladding, and a low index polymer
Figure 4. Monostatic System Multiplexer and transmitter/receiver.
Multimode fiber-105 µmReceiver
photodiode
Outer coating-250 µm
Innercladding-90 µm
Core-12 µm
TX/RXLens
TX/RXLens
1.5 µm Pulsed Fiber Laser
Single-mode coreCoupling regioncross-section
AR COAT
AR COAT
Fusion splices
8
outer cladding.
The double-clad small-core fiber enables efficient transmission of a low-NA single-mode
output beam in the core and efficient separation of high-NA light received from the target into
the fiber cladding modes. These cladding modes are coupled to a transversely-fused high-NA
multimode fiber that directs the received light to a photodiode. The coupling region of the fused
fiber coupler can be seen circled in Figure 4.
A similar DCF/MM receiver has been previously described [12, 13] and previously
implemented [14] in a fiber scanner/detector for barcode scanning, in which the received light
was de-multiplexed by side-stripping from the cladding of a single-mode fiber (SMF) to a side
detector. Such a system that uses side-stripping tends not to lend itself to a high bandwidth
system. Generally a larger detector, with larger capacitance, would be needed for side stripping
to efficiently collect the light and is not compatible with high-speed pulses. In our MUX, the
received signal light is coupled into the inner cladding of the DCF by the TX/RX lens which has
an NA that significantly exceeds that of the fiber core. Inside the MUX, the received light
transfers into the MM fiber pigtail of the APD.
2.1 Fiber MUX modeling
Finite-difference beam propagation software, RSoft’s BeamPROP, is used to model a fused
fiber coupler in the monostatic all-fiber LADAR system. The fibers are fused over a length of 1
cm, where the circular fiber areas overlap by 5 µm as shown in Figure 5 (A). Power is quickly
transferred from the DCF to the MM fiber as soon as the fibers are contacted as seen in Figure 5
9
(B). A simulated randomized speckle field consistent with an expected target return, focused
with an NA = 0.22 receiver lens onto the core region of the fiber. Since the core’s NA = 0.12 is
substantially smaller than the NA of the incident light, most of the received power is transferred
to the cladding within 500 µm, as can be seen in Figure 6 (A). As shown in Figure 6 (B) and
Figure 5. Model of the MUX layout and power transfer.
105 µm MMFiber
90 µm DCF
MUX cross sectionat end
Amplitude Distribution in Coupler
10
20
30
40
50
Z(m
m)
MUX cross sectionat middle
MUX cross sectionat end
(A) (B)
Air
10
(C), the model predicts that 53% of the power is transferred from the 90 µm inner cladding of the
DCF fiber to the 105 µm MM fiber, 40% of the power remains in the DCF fiber, and 7% scatters
out of the fibers. In the absence of scattering loss, a transfer ratio of 58% can be expected, based
on the geometric ratio of the 105 µm fiber area to the total area of the two fibers (Equation 1).
An overall MUX fiber transfer ratio of ~50% was typically measured, in reasonable agreement
with the RSoft model. There are other models for fused fiber couplers [15-20], however
Equation 1 is sufficient for the MUX used in this dissertation.
Figure 6. Model of the MUX power transfer.
105 µm MMFiber
90 µm DCF
40% 53%
93%
Power quickly leaves the core
Fraction of Launch Light
Amplitude Distribution in Coupler Fiber Power
Fiber Power Legend:
Power in DCF core
Power in MM Fiber
Power in DCF
Total Power in Fibers
mm
10
20
30
40
50
1
mm
Z(m
m)
(A) (B)
(C)
Air
11
1052
1052 + 902= 58%
(1)
2.2 Fiber MUX fabrication
The fabrication of a fused fiber coupler was performed using a CO2 laser, beam scanning
mirrors, and a pair of computer controlled fiber-holding chucks (Figure 7) similar to Dimmick et
al. [21]. There are other fabrication methods which fuse the fibers in different ways and
incorporate different monitoring feedbacks. [15, 16, 18, 19, 22-26] The equipment shown in
Figure 7 was used to fabricate and characterize the fused fiber couplers. The system is very
flexible and can change the amount of applied laser energy and how it is applied in a number of
ways. The scanning mirrors are controlled from a PC using a National Instruments (NI) DAQ-
6115 card driven by Wavemetrics Igor Pro software.
Figure 7. Diagram of the experimental setup used to create the MUX.
CO2 Laser
X/Y GalvoMirror Scanner
Visible Alignment Laser
CO2CylinderLens
Motorized Fiber Holder Chucks
Power Monitoring of Fiber Input Power
Power Monitoring of Fiber Output Power
12
The performance tradeoffs of the MUX were explored by varying the applied CO2 laser
energy and tension to the fibers. The delivered laser energy determines the amount of fusing
because changing this parameter changes the amount of heat delivered to the fiber. The overall
signal power in the fibers was monitored in-situ. Some of the other performance parameters that
were investigated include core and cladding transmission, and the NA of the light coupled into
the multi-mode fiber.
The coupler is composed of a 105 µm diameter fiber fused to a 90 µm fiber. One reason that
this fiber diameter is chosen is due to the availability of an off the shelf Voxtel fiber-coupled
detector that uses a 105 µm, NA = 0.22 fiber. The receiver lens NA can closely match this
detector fiber NA. This can be seen in Figure 8.
The fibers are center-stripped and then placed in the fiber holding chucks of a Vytran LDS
1250 splicing system. Each edge of the stripped region is placed flush with the fiber holding
chuck. A single 360º twist is performed on the two fibers to keep the fusion region touching
during heating. Before fusing the fibers, there is typically less than 10% transfer of cladding
power after completing the twist, due to optical contact between the fibers.
The CO2 laser scanning is then turned on (SYNRAD Firestar V40 laser, model FSV40KFD,
Figure 8. MUX with an off-the-shelf APD and lens.
Receiverphotodiode
Laser Diode and Fiber Amplifier
Tx/Rx Lens
0.22 NA90 um dia
105 um dia
0.22 NA
13
and UC-2000 controller). The laser beam is focused laterally using a cylinder lens to
approximately 1.5 mm x 5 mm, for heating efficiency and to allow for easy alignment in hitting a
combined fiber diameter as large as 250 µm. The co-aligned visible red laser diode source using
a beam splitter also facilitates the optical setup. Figure 9 shows how the voltage applied to the
scanning mirror corresponds to a typical fused fiber scan length after the laser goes through a
lens. It can also be seen in this figure that the laser is steered off the fiber to prevent excess
dwell time at the ends of the sweep. The applied voltage allows the dimensions of the fused
region to be changed. A longer fused region can also be completed by moving both of the
motors in the Vytran system and thus physically moving the fibers through the beam.
The softening point of typical silica optical fibers range from 1200°C-1800°C [20] depending
on dopants and impurities in the fibers. Tajima et al. shows that the transition temperature of
Figure 9. CO2 laser mirror scanner control.
14
GeO2 glasses is typically over 1600°C [27]. A 40 W continuous wave laser is used in my
experiment. There is more than enough power since it is shown in the literature that 25 W is
sufficient [21, 28]. I typically ran at 31 W CO2 power when fusing.
Performance tradeoffs are investigated by varying the applied CO2 laser energy and tension
to the fibers. If there is too little heat and too much tension, the fiber slips in the fiber chucks. If
there is too much heat with too little tension, then the glass melts and drips down the length of
the fiber. If there is too much heat and too much tension then the fiber is tapered. The amount
of delivered laser energy can also change the degree to which the fibers are welded. Figure 10
shows how as laser energy increases, the more the fibers are welded together. The fiber chucks
Figure 10. Fused fiber couplers.
Tack Welded.Medium “Snowman” Welded.
Side view of Medium “Snowman”Welded.
Ellipse Welded.
15
are slowly pulled apart at 0.0025 mm/s to keep the fibers together during the 2 s that the CO2
laser is fusing the fibers.
The fused sections of the couplers range from 0.4 – 1 cm long. Difficulty can arise in long
heating times due to the absorption of the fiber increasing as the temperature of the fiber
increases. [27, 29] Therefore, I preferred strong short bursts of energy from the CO2 laser
compared to longer heating at a lower energy. After fusing the MUX fibers, the DCF core
transmission loss is measured to be less than 2%.
The fraction of DCF cladding power that can be transferred into the MM fiber can be
increased by tapering the DCF fiber. A MUX coupler geometry incorporating such a taper is
shown in Figure 11. An example of a fabricated fiber taper is shown in Figure 12 where an SMF
fiber is tapered down to 24 µm. Tapering requires lower power compared to fusing two fibers. I
typically ran at 26 W CO2 power when tapering. However, this increases the NA of the light in
the DCF fiber cladding. If the NA of the light in this fiber cladding exceeds what the fiber can
support then the light will couple into the air and out of the fiber.[30] This can be seen in Figure
11.
Figure 11. Coupler performance tradeoffs.
Cladding-0.47 NA
Received light 0.22 NA
Brightness theorem- For a 2x reduction in fiber diameter, the propagating light becomes 0.44 NA
16
Multiple couplers were made with various fibers as seen in Table 1. For trial 1, the light is
launched into the 90 µm fiber and is fused to the 105 µm fiber. For trials 2-4, light is launched
into the 105 µm fiber and then tapered. The tapered fiber is then fused to a 125 µm fiber. Table
1 shows the predicted coupling closely matches the actual coupling. A 90 and 125 µm coupler
was measured using an integrating sphere to detect light that transferred out of the cladding and
into the air. This scattered power was approximately 10% of the input power, in agreement with
the finite difference beam propagation model.
Uncertainty sources that contributed to the geometric coupling column include measurement
error using the microscope and camera to examine the tapered fiber (+/- 0.003), the initial fiber
Trial #
Predicted geometric coupling (+/- 0.007)
Actual coupling (+/-0.10)
1 90 µm fiber fused to 105 µm fiber 0.58 0.50 2 105 fused to 125 0.59 0.61
3 105 tapered to 80 fused to 125 0.71 0.74 4 105 tapered to 70 fused to 125 0.76 0.75
Table 1. The result of varying fiber diameters and percentage of coupling. The
baseline for the measurements is the power coupled into the fiber.
Figure 12. End of fiber taper with CO2 system from 125 µm to 24 µm, made
in-house.
17
diameter (+/- 0.002), and variations in the speed of the stepper motors used to pull the softened
fiber apart (+/- 0.002). Uncertainty sources that contributed to the actual coupling measurement
include laser power drift between measurements (+/- 0.03), alignment of the fiber to the detector
(+/- 0.02), thermal drift of the detector (+/- 0.02), and alignment of the coupled light into the
fiber (+/- 0.03). The detector measurement accuracy could be improved by using ratioed
photodetectors and fiber splitters to remove laser power drift error.
The last step in coupler fabrication was to package the bare fibers, which are fragile if left
exposed to air, in a protective enclosure. Fiber packaging affects the impact of temperature and
mechanical forces on the performance of the coupler [31]. There are many ways to package the
couplers; the fibers can be inserted into Teflon or metal tubing, heat shrink tubing, or re-coated
with various index polymers. Most of the recoats of various index polymers require a UV source
to cure the polymers. My current packaging system of choice is to epoxy the end sections of
both fibers together with a UV curing index-matching polymer while the fibers are still in the
fiber chucks of the Vytran LDS-1250. Next, the fibers are removed from the chucks, and a semi-
hard thick-walled Teflon tubing is slid over the bare glass stripped section of the coupler,
extending onto jacketed (unstripped) section of the fibers. Both ends of the Teflon tubing are
then sealed with the same polymer to prevent the coupler region from slipping out of its
protective sleeve.
The couplers are characterized by inserting the laboratory-simulated LADAR receive signal
into the fiber as seen in Figure 13. The power is then measured out of both ends of the coupler.
Power transfer in the core of the DCF fiber is also measured by inputting a known power into the
core and applying at least 1 cm of carbon stripping paste on the glass cladding at the output end
18
of the DCF; this strips any light that is present in the inner cladding of the DCF to assure that that
the measured output power is coming only from the fiber core. Alternatively, the power in the
core can also be measured by fusion splicing Corning SMF-24e fiber to the end of the DCF; the
SMF fiber jacket is designed (it has slightly higher refractive index than the glass cladding) to
strip any light propagating in the fiber cladding. With this approach, assuming negligible fusion
splicing loss, the power emerging from the SMF fiber is equal to the power propagating out of
the DCF core.
The MUX coupler is characterized during fabrication by filling the DCF RX fiber with a
similar distributed optical field as would be expected for a return rangefinder signal. As shown
in Figure 14, to duplicate this field, light from a 1550 nm single-mode fiber laser source is
focused onto a diffuse surface to create a received speckle pattern resembling that from a 50 m
distant target. A pair of 27 mm diameter, NA = 0.33 lenses, the same as the one in the laser
range finder (LRF) system, collects the scattered light and couples it into the DCF fiber. A lower
Figure 13. Coupler power transfer measurement.
InputX mW
Output 1X mW
Output 2Y mW
19
effective NA = 0.22 is generated by setting an aperture in front of the coupling lens. The
characteristic expected speckle field is also observed at the DCF fiber using a separate short
wave infrared (SWIR) camera. The correct speckle size that reproduces the expected far-field
return pulse is produced by the spot size on the diffuse target. [32]
Using input light generated by method of Figure 14, I measured a typical 45 - 50% signal
transfer efficiency in the 90 µm/105 µm MUX, defined as the ratio of power emerging from the
output end of the MM fiber divided by the power coupled into the inner cladding of the DCF.
This fraction is somewhat lower than a 58% transfer ratio that is allowed by the ratio of the
cross-section areas of the MM fiber divided by the total cross-section of both fibers in the fused
section of the coupler. In theory, a larger diameter MM fiber, as well as tapering of the 90 µm
DCF fiber in the fused coupling zone, should lead to a larger signal coupling efficiency.
In order to understand how to couple the light into the receiver, it is important to understand
the fiber modal distribution. The output modal distribution depends on the input excitation field,
which is why the previously described speckled source field was used to simulate actual system
properties. The lab setup shown in Figure 15 was used to characterize the light distribution in
the laser sources, the light coupled into the fibers, and the output from the couplers. It consists
Figure 14. Block diagram and physical generation of a monostatic launch
field that simulates the return received signal.
Diffuse Target
Launch Fiber
Receive Fiber
20
of a fiber that is placed in the same spot (+/-2 mm in the z direction) and is rotated in Φ and Θ
using a Newport PR50PP motorized stage. A biased germanium photodetector with a 4.6 mm
pinhole in front of it receives a signal from the rotated fiber. The detector is then connected to a
Stanford SR570 current pre-amplifier and measured with a National Instruments DAQ-6110 with
Wavemetrics Igor Pro software package.
The angular distribution of output light from the MM fiber of the MUX as measured in
Figure 15 is shown in Figure 16, where an input (receiver) lens with an NA = 0.33 was used to
couple light into the 90 µm inner cladding of the DCF fiber in the MUX, and the MM and DCF
fibers were well fused together to achieve strong cross-coupling from one fiber to the other. The
NA of the MM intensity distribution, as defined by light within a 5%-of-maximum boundary, is
0.33, similar to the NA of the receiver lens. When MUX fibers were insufficiently fused
(resulting in small fiber cross-section overlap), resulting in weak coupling between the fibers, the
angular distribution of the MM fiber exhibited a donut shape. This was a consequence of much
Figure 15. Far-field characterization of scanned fiber with fixed detector.
2-dof stage Scanned FiberFixedDetector
θ
Φ
A/D converter
Computer Analysis
21
stronger cross-coupling of the high NA light than the low NA light propagating in the DCF fiber;
this is thought to occur due to the high NA light making more bounces inside the fiber and
having a greater chance of crossing over to the MM detector fiber, while the low NA light has
fewer bounces into the high NA light for a given fused region. This becomes apparent when
either using the previous characterization setup or observing, with an IR viewer, the far-field
distribution of the light from the MM fiber illuminating a screen. The doughnut shaped far-field
output pattern, corresponding to weak fiber fusing, was frequently seen when using the Vytran
LDS-1250 large diameter splicing system which used an iridium filament to fuse the fibers; the
current carbon dioxide fusion setup [33, 34] resulted in more complete fiber fusing and a far-
field pattern exemplified by Figure 16. There are other non-destructive means that can also be
used to characterize a coupler in detail.[35]
Figure 16. Far-field output angular distribution of the MM output fiber in the
fused MUX coupler. The dashed line indicates the 5%-of-max boundary.
22
3.0 Monostatic laser rangefinder system
A laser rangefinder (LRF) system was designed and built in conjunction with a LADAR
system to understand the tradeoffs associated with a monostatic system and to investigate
multiple pulse processing methods. The prototype allowed for a thorough exploration of the
design space without the additional features of a scanning LADAR system. Preliminary results
for this system were presented earlier. [36, 37]
3.1 Block diagram
The monostatic, all-fiber rangefinder system is shown in Figure 17. It consists of a pulsed
Figure 17. Block diagram of all-fiber monostatic rangefinder system.
AR-coated Perpendicular Cleave
1.5 µm Pulsed Fiber Laser
APD Receiver Fiber MUX
(TX/RX) lens
Signal Processing
Isolator
23
1.5 µm fiber laser source, a multimode fiber coupler MUX described above that separates
transmitted light from received light, an APD receiver, and a signal processor, which consists of
a fast comparator and time analyzer. The fiber laser utilizes a 1-stage cladding-pumped erbium-
ytterbium doped fiber amplifier (EYFA) Master Oscillator Power Amplifier (MOPA)
configuration that is similar to those described previously by Mindly et al. [38]. A directly
pulse-modulated single mode distributed-feed-back (DFB) laser diode is used as the seed laser
for the Er/Yb amplifier. The maximum pulse power was 600 W, limited by nonlinear effects that
cause saturation of peak power. These originate mainly in the large length (compared to the 3.3
m length Er/Yb DCF itself) of passive fiber following the amplifier. The 1.5 µm MOPA laser
generates 5.6 ns, 3.3 µJ pulses at a 50 kHz repetition rate. Light generated by the fiber laser is
transmitted by the single mode core of the MUX double cladding fiber and is collimated by the
monostatic TX/RX lens. Using a Gaussian-approximated waist parameter of 6.0 µm from the
fiber core gives a full-width (1/e2) waist of 6.4 mm at the Lightpath GRADIUM lens (27 mm
diameter, 40 mm focal length). This lens was selected for its low spherical aberration. After the
lens, the TX beam half-angle divergence is 150 µrad. At 50 m range, standard Gaussian mode
propagation gives a full-width (1/e2) spot size of 16.3 mm (including the near-field correction).
The same TX/RX lens collects the return light and couples it into the cladding modes at the anti-
reflection (AR) coated face of the output fiber. This fiber serves as the input to the MUX, which
couples the cladding-mode light into the MUX output fiber. The output light is lens-coupled to
the CMC Electronics P264-339769-101 InGaAs APD, which has a 200 µm diameter active area
and 200 MHz bandwidth.
24
3.2 Signal processing
The APD receiver signal is input to a fast comparator (Analog Devices LT1711), whose
pulse output is connected to an ACAM GP22 time-to-digital converter (TDC). The TDC
receives an earlier start trigger synchronous with the transmitted laser pulse; it records the time
of the first edge from the signal comparator (if it occurs within the TDC's 2.5 µs timeout) minus
the time of the trigger edge. The return-delay-times from a 1000-pulse burst are recorded in real-
time (at 50 kHz) with a Teensy 3.1 microcontroller using the TDC's Serial Peripheral Interface
operating at a clock frequency of 20 MHz. The range data are logged via USB to a laptop
computer, and then analyzed.
3.3 Optical link parameters
Table 2 gives the system optical link budget. The dominating excess system loss sources are
lens-to-fiber coupling efficiency, reduced APD response due to the effects of a large zero-time-
delay (T0) pulse return caused by residual reflections in the transmit beam optical path, and the
fiber MUX transfer efficiency. The TX/RX lens does not effectively focus to a 90 µm spot. A
larger clad fiber and a better TX/RX lens could reduce this coupling loss. Methods are under
examination for reducing the T0 signal (covered in more detail in section 5.0). Turbulence,
atmospheric attenuation, and target movement have not been included into the link budget table.
A potential additional source of noise is solar background. The mean solar power at the receiver
has been calculated as -69.8 dBm, with a 50 nm wide spectral filter. This and its associated
25
noise should have negligible impact on system performance.
3.4 Results
From Table 2, the predicted signal to noise ratio (SNR), given by subtracting the measured
receiver noise power from the calculated receiver peak power, is 18.2 dB (66.1 on a linear scale).
A typical signal return is shown in Figure 18, where the “baseline” trace at 850 mV, taken with
the laser turned off, represents receiver baseline noise voltage. In Figure 18 the return pulse can
be seen to sit on top of a 780 mV pedestal, which corresponds to a 70 mV offset from the
receiver baseline of 850 mV. The 780 mV pedestal (which is present in absence of any RX
return signal) is caused by the Amplifier Spontaneous Emission (ASE) power from the amplifier
which builds up between the amplified pulses. Due to residual reflection by the TX fiber end, a
portion of the ASE power is coupled into the APD. The ASE power can be significantly reduced
in the future with improved fiber amplifier designs, such as using a multi-stage configuration
Fiber laser peak power 58.1 dBm Sources of loss: dB
Diffuse target reflectivity -9.6 1/R2 range loss (50 m Range, 27 mm RX diameter) -71.4 Lens to fiber coupling -3.9 Fiber MUX -3.0 Receiver sensitivity drop due to T0 recovery -3.0
Total loss -90.9 Calculated receiver peak power -32.8 dBm Measured receiver noise power (standard deviation) -51 dBm
Table 2. Link Budget.
26
with narrow-band filtering in between the stages. The measured receiver noise power is
calculated by dividing the standard deviation, 0.81 mV, from Figure 18, by the APD receiver
responsivity of 103 kV/W. The measured (linear) SNR of 52 is obtained using individual
rangefinder 3.3 µJ pulses of the 50 kHz laser from an 11% reflective target at approximately 50
m. This SNR is the average peak signal level divided by the standard deviation of noise in the
baseline signal.
Each pulse signal serves as an input to the comparator, which triggers the time-to-digital
ACAM converter chip [39]. This was operated in a mode which saved one threshold-crossing
time (if detected) per laser pulse. If a typical range-finder specification of 0.01 probability of
false alarm (PFA) and 0.99 probability of detection (PD) is used, the calculated maximum range
for single-pulse detection with the above SNR conditions is 115 m. In this case, PFA = 0.01
means that the probability of no false alarms (FA) at maximum range is 0.99.
The overall range performance can be greatly improved by using multiple laser pulses to take
Figure 18. Signal traces showing residual T0 signal and return pulse (red)
and baseline with laser off (blue).
27
advantage of the system’s high repetition rate (50 kHz). This improvement can be accomplished
by allowing an increased false alarm rate and analyzing the ACAM pulse temporal statistics
(Figure 19), or by full waveform averaging. With the first method, the temporal data of the
detected signals are acquired with a low comparator threshold voltage, which increases both the
false alarm and detection probabilities; then a microprocessor analyzes the time statistics of the
ACAM time-to-digital converter chip output. The microprocessor sorts the group of time-of-
flight data by delay time, and searches for multiple pulses near the same range (delay time) to
discriminate the multiple-return target range from random false alarm arrivals. The upper data in
Figure 19 falling outside the narrow target range band are due to systemic noise artifacts in the
threshold circuit at ranges (times) exceeding the target range. They are useful in this proof-of-
concept analysis, even though they do not come from random noise. True random false-alarm
events (that may be approximated by a Poisson arrival process) would have broad range
distributions throughout the acquisition interval. After sorting, a graph similar to Figure 19 (left)
would have a target range plateau with steep rising tails on either side. For the 1.8 µJ data, 109
points are within the 50 m plateau. For the 3.3 µJ data, 943 samples are within the 50 m plateau.
An example of post-processed statistics of the measured data is in the right graph of Figure 19.
For this system, a minimum of 81 detected target signals is chosen as being sufficient for
meaningful timing statistics. The data in Figure 19 were measured with 1000-pulse (20 ms)
bursts of the 50 kHz pulsed laser. An AND gate threshold is set to ignore any T0 artifacts before
26 m. The range standard deviation for the data in the lower plot of Figure 19 is 0.0078 m (7.8
mm).
A discussion of using multiple trigger arrival time accumulations that account for false
28
alarms to lower the necessary comparator threshold voltage and increase the range capability of
the system to 269 m is discussed in more detail elsewhere [37].
An alternate approach to processing return times from multiple laser pulses is to perform fast
A/D sampling of each entire return signal, and average the multiple digital waveforms. For 1000
averaged signals, an improvement in SNR of √1000 = 31.6 is expected. The point of maximum
amplitude of the averaged signal gives the most probable location of a potential detected target.
If the signal at this averaged waveform maximum exceeds an appropriate threshold level, it is
judged a valid return. Using the individual single-pulse SNR (52) at 50 m, the calculated
maximum range limit is 940 m. The disadvantage of this method is the increased complexity of
making a small, high-speed, large-storage data acquisition system.
Figure 19. Multi-pulse data analysis. (A) The enlarged lower vertical axis
clearly displays the plateaus in the data. (B) The corresponding histogram
near 50 m for 3.3 µJ is displayed.
29
4.0 Scanning LADAR system
Use of fiber cantilever bending vibrations allows for a simplified means of beam scanning
compared to conventional methods with mechanical mirror or prism scanners. However, this
requires single-pulse detection (with no burst-mode or signal averaging permitted) and a means
of sensing the instantaneous fiber position. A common fiber geometry and optical detection
configuration, including the same TX/RX lens and similar MUX was used in the scanning
LADAR and static LRF systems. Similar advantages of compactness, low weight, and low cost
associated with simpler fiber construction apply to both systems.
4.1 Block diagram and system overview
A compact scanning LADAR based on a fiber-coupled, monostatic configuration which
transmits and receives through the same aperture has been demonstrated. As will be shown, the
all-fiber approach also makes it possible to implement a compact beam-scanner that uses a PZT
cantilever to vibrate a fiber tip, resulting in its large lateral movement. The combined piezo-
electric PZT stripe actuator/fiber cantilever scans both the transmit near-single-mode optical
beam from the fiber and its cladding-mode receive aperture. When the fiber tip is placed in the
focal plane of a collimating lens, this lateral translation results in scanning of the collimated
beam TX angle. If the fiber scanning process is compared to conventional bi-static systems with
polygon, galvo, or Risley-prism beam scanners, the described system offers several advantages;
30
inherent alignment of receiver field-of-view (FOV) relative to TX beam angle, small size and
weight, and power efficiency. Optical alignment of the system is maintained at all ranges since
there is no parallax between TX beam and receiver FOV for all positions of the fiber tip. During
each laser pulse, the position is measured by placing a two-dimensional silicon PSD close to the
fiber tip to determine the instantaneous tip position and resulting beam angle [13]. The Si PSD
operates in a two-photon absorption mode to detect the transmitted 1.5 µm pulses. The prototype
system collects 50,000 points per second with a 6º full scan angle capability and a 27 mm clear
aperture/40 mm focal length TX/RX lens. The system has a range measurement precision of 4.7
mm and has been tested to a range of 26 m.
A schematic block diagram of the monostatic, all-fiber LADAR system is shown in Figure
20. It consists of: (1) a MM/DCF MUX coupler with a near-single-mode core that transmits the
signal from a (2) 1.5 µm pulsed fiber laser, an inner cladding that transmits the return signal light
and couples it to (3) an APD, (4) a resonant PZT cantilever with a fiber cantilever attached to its
Figure 20. All-Fiber scanning LADAR design.
NI 4-ch A/D, D/A (PCI)
PC
Pulsed fiber Laser (1 ns, 50 kHz)
APD receiver Fiber MUX
PZT
2D-vibrating biaxial fiber cantilever
2D position sensing detector (PSD)
scan lens (TX/RX)
A
Acqiris Pulse analyzer (PCI)
(1)
(2)
(3)
(7)
(8)
(4) (6)
(5)
~ ~Ω1 Ω2
31
end, (5) a silicon 2-D PSD, (6) a scan/receive lens, (7,8) and two A/D systems on computer PCI
cards used to collect the data.
The pulsed fiber laser consists of a distributed feedback, 1.5 µm (DFB) seed-laser diode, a 2-
stage erbium-ytterbium doped fiber amplifier (EYFA), and electronics to control the EYFA
pump power and triggering of the DFB laser. The EYFA generates 2.9 µJ, 1 ns full-width half-
max (FWHM) pulses at a PRF of 50 kHz. It differs from the previously described LRF laser in
using shorter and higher peak power (3 kW) pulses. Approximately 10% of the time-averaged
output power was in the form of Amplified Spontaneous Emission (ASE). The vibrating output
fiber of the laser is spliced to a double-clad fiber of the MUX.
4.2 Vibrating fiber scanner
A key component of the system is the 2-D vibrating biaxial fiber cantilever mounted to a
PZT. PZT-driven fibers have been used extensively in scanning probe microscopy and in
endoscopes [40-45]. However, these systems typically used a symmetric fiber with 4 PZTs
arranged in a tubular shape to achieve an X-Y scan. Our beam-scanning fiber has a flattened
cladding shape, resulting in two different mechanical resonant frequencies for its thick and thin
axes. The double-cladding fiber as seen in Figure 21 consists of a 12 µm diameter near-single
mode core with 0.12 NA, a 90 µm multimode glass inner cladding with 0.22 NA, a 121 x 100
µm glass outer cladding, and an acrylate jacket (not shown in Figure 21) removed from the
vibrating tip. The circular inner cladding guides the return light for the detector, while the outer
cladding asymmetry separates the transverse vibration frequencies. This beam-scanning fiber is
32
fusion-spliced to a matched-core DCF fiber, with a circular 90 µm inner cladding, that is used to
fabricate the fiber MUX. A similar scanning method has been proposed previously by Roberts et
al [14]. Roberts et al. used a D shape fiber whereas the fiber used in these experiments is more
of a symmetrically flattened fiber. The rounded quasi-rectangular fiber shape has two lowest-
order modes of cantilever bending vibration, in perpendicular directions. The resonant
frequencies are separated sufficiently to make the medium-Q modes nearly orthogonal. A single
signal that consists of the summation of the fiber’s two resonant frequencies (Figure 22) is
applied to the PZT from the NI A/D, D/A board, creating a Lissajous scan pattern. The
frequencies are tuned to be relatively prime in order to give a low pattern repeating rate and
dense scan pattern in 0.4 s. The signal is created digitally and converted into an analog signal
from the NI A/D, D/A board.
The perpendicular-cleaved scanning fiber tip has an AR coating to minimize back reflection
Figure 21. Scanning fiber photo micrograph.
33
of the TX light into the fiber inner cladding and to the APD. This back reflection can cause
damage of the APD or signal saturation which, because of slow detector recovery, results in
reducing the minimum detection range. This signal saturation is explained in greater detail later
in the dissertation.
The polymer and acrylate coating are mechanically stripped the entire length from the PZT
mounting point to the free end. This allows a rigid epoxy bond of the glass surface of the fiber to
the PZT surface.
In the top of Figure 23, the fiber is epoxied to a single bilaminar PZT stripe actuator (APC
model #40-2040) at a 45º angle, where the cantilever PZT length is 3 cm, and the fiber cantilever
length is 1.73 cm. This method is simpler in construction and achieves similar fiber deflection
with an order of magnitude times less drive voltage when compared to a recent tubular PZT
implementation [40].
The PZT response is characterized by using a function generator, and a microscope with a
digital camera (Figure 24). Figure 24 also shows a typical frequency response of a 3-cm
piezoelectric stripe from APC International. Its cantilever resonant frequency is around 330 Hz.
Figure 22. Normalized PZT excitation signal.
34
The FWHM of the response is about 40 Hz. A full PZT swing of around 0.7 mm is seen with a
voltage of 6 V. A larger deflection occurs when a higher voltage is applied.
The fiber cantilever resonant frequencies were set to be near the PZT resonance by choosing
the required fiber cantilever length. In this case, the fiber tip motion is amplified by the double
resonances of the PZT and fiber bending modes. Micrometers of PZT movement are resonantly
amplified to millimeters of fiber tip movement; typically less than 1 Vpp of PZT drive was
required in the LADAR system. The frequency response of the 1.73 cm fiber cantilever is shown
in Figure 25 for the fast and slow directions. The low-frequency mode (vibration parallel to the
Figure 23. Fiber scanning assembly and picture of scanning fiber.
ClampPZT
Piezo Stripe Actuator mounted at 45° angle. Fiber vibration planes are horizontal and vertical.Biaxial-fiber vibrating cantilever (1.73 cm) epoxied to piezo actuator.
±1 V (dual frequency) electrical stimulus to piezo from PCI card.
2D Transmissive Position Sensing Detector10 mm
Epoxy
2 mm
35
short transverse fiber axis and in blue in Figure 26) is calculated to have a resonant frequency of
290.2 Hz based on an axis length of 100 µm; the high-frequency mode (vibration parallel to the
long transverse fiber axis and in red in Figure 26) is calculated to have a resonance at 336.5 Hz
based on an axis length of 121 µm. The resonant frequencies agree well with measured resonant
frequencies of 286 Hz and 333 Hz respectively. Measured fiber FWHM resonance widths are ~8
Hz, and the damping constant in Figure 25 has been adjusted to match this. As the fiber tip
velocity increases, the damping will also increase.[40] Sources of error for the calculated
resonant frequencies include fiber uniformity, length of the epoxied fiber, and details of the
Figure 24. PZT stripe characterization.
Frequency Generator
Audio Amplifier PZT
Computer Analysis
Microscope
Camera
0
100
200
300
400
300 320 340 360 380 400Sing
le-s
ided
di
spla
cem
ent (
µm)
Frequency (Hz)
6 Vpp PZT excitation voltage
36
epoxy mounting to the PZT. The final vibrating fiber response is a combination of the PZT
cantilever and fiber cantilever response, corresponding to the convolved response of Figure 24
and Figure 25.
For this vibrating fiber PZT arrangement, an illustrative fiber tip displacement of 2 mm is
shown in Figure 23. Significantly larger amplitudes were easily achievable.
A single drive voltage consisting of the sum of two sinusoids, as shown in Figure 22, at the
fiber’s resonant frequencies is applied to the PZT. This excites both orthogonal vibrations at
their respective frequencies, creating a Lissajous scan pattern. The frequencies are fine-tuned to
give a low pattern repeating rate and dense scan pattern of 20,000 (50 kHz) points in 0.4 s as
seen in the simulated data of Figure 26. For comparison a SWIR image of the experimental
Figure 25. Biaxial fiber oscillation.
37
Lissajous scan is shown in Figure 27. The Lissajous scan is sparser in the center and denser near
Figure 26. Simulated Lissajous scan.
Figure 27. Lissajous scan pattern taken with a SWIR camera at 0.2 s
exposure (calibration patch in center).
38
the edges because of sinusoidal dwell time factors. Note also the narrow elliptical features at the
edges caused by the residual tail responses of the two modes (i.e. each mode responds weakly
and out-of-phase to excitation away from its resonant frequency). A ramped amplitude
excitation voltage was also investigated to distribute the edge dwell of the Lissajous scan more
uniformly over the entire scan area. A small vertical ellipse can be produced by applying just the
resonant low frequency as seen in Figure 28. The other axis is weakly excited as well since the
tails of the high-frequency mode reach to the first fundamental frequency.
Equation 2 shows the relation between the scan angle, 𝜃𝜃, fiber lateral tip displacement, 𝛥𝛥𝛥𝛥,
and the focal length, 𝑓𝑓, of the lens. This is also illustrated in Figure 20.
𝜃𝜃 = 𝛥𝛥𝛥𝛥𝑓𝑓
(2)
The TX/RX lens has a focal length of 40 mm and a clear aperture of 27 mm. The lens is
sufficiently large so that with a ~14º divergence of light cone emerging from the fiber core, and
Δx = 2 mm, the transmitted light is not clipped by the lens aperture (Figure 29). With these
parameters, the maximum full scan angle is 6º. A larger scan angle can be achieved with a lower
Figure 28. Model demonstrating strong single resonant frequency excitation
at 290 Hz.
39
NA fiber core and a lower F-number lens. For the LADAR data in Figure 45, a PZT drive
amplitude of 0.13 V gave a Δx of 0.32 mm, and a half scan angle θ = 0.5º.
In Figure 30, a SWIR camera was also used to take measurements of the vertical scan pattern
Figure 29. Maximum scan before clipping of the TX/RX lens occurs. The figure
2 mm
6.4 mm
4.7 mm
9.2º - Fiber TiltFiber displacement-
Total half scan displacement is 13.1 mm
Lens Focal Length ~ 40 mm
Scanning Fiber
Lens pupil plane
Clear aperture of the lens is 27 mm
Figure 30. Scan voltage versus fiber tip displacement.
y = 2.082x + 0.3558
0
0.5
1
1.5
2
2.5
3
0 0.5 1
fiber
tip
disp
lace
men
t(m
m)
Scan Voltage (V)
Fiber tip displacement
single sided fiber tipdisplacement (mm)
best linear fit
40
generated by exciting the fiber cantilevered PZT with a single frequency and a 1.5 µm source.
The resulting fiber tip displacement was then calculated using the formula shown in Equation 2.
4.3 Position sensing detector
A PSD is used to instantaneously sense the fiber tip position since changes to the PZT, fiber,
and environmental factors such as temperature and external vibrations could affect either a
model or a one-time phase calibration that attempted to predict the fiber tip location. Small
mechanically induced position artifacts were observed in a scanning fiber endoscopic system
[41] that did not use a PSD. Our silicon PSD [46], custom modified to operate in a transmissive
configuration for the 1.5 µm light, is shown in Figure 31. Since silicon has an absorption band-
gap larger than the 1.5 µm photon energy, it is highly transparent for linear one-photon
absorption in the TX and RX beams. However, the silicon sensor can detect a pulse of 1.5 µm
light because of two-photon absorption (TPA) if the intensity is sufficiently high. Britow et al.
Figure 31. 2-D position sensing silicon detector for 1.5 µm pulses.
41
characterize in detail two photon absorption in silicon at 1550 nm.[47]
Weak two-photon absorption during the ns-wide, multi-kW peak-power transmit pulses
causes sufficient electron-hole generation in silicon to generate a readily measured pulsed sensor
electrical signal for ‘x’ and ‘y’ positions. The PSD output pulse is bandwidth-limited to 400
kHz, and its shape is observed to change with excitation location (Figure 32). Some of the
ringing may be resolved by adding a 10 pF capacitor to the amplifiers. Typically PSDs are used
in a continuous mode rather than a pulsed mode as is done in this LADAR. A signal,
representing displacement, is obtained by numerically integrating the PSD response during the
first half of the 20 µs interval between pulses. This process is completed for each of the 4 PSD
signals X1 - X2, X1 + X2, Y1 – Y2, and Y1 + Y2. The x and y coordinates are then computed
using Equation 3. The position measured for each pulse is used to create the lateral display
coordinates, as in Figure 33 (B).
position 𝛥𝛥 = 𝐿𝐿2
𝑋𝑋1 − 𝑋𝑋2𝑋𝑋1 + 𝑋𝑋2
position 𝑦𝑦 = 𝐿𝐿2
𝑌𝑌1 − 𝑌𝑌2𝑌𝑌1 + 𝑌𝑌2
(3)
Figure 32. 2-D position sensor signals. (1 µs / div)
Output from X1-X2 signal
Laser on left side of PSD
Laser on center side of PSD
Laser on right side of PSD
Filtered pulse shape changes with location:
Integrating first 50 points sampled at 0.2 us gives a reliable integrated signal level.
Fiber
42
Instantaneous sensing of fiber position for every laser pulse by the PSD makes the system
insensitive to the effects of external vibrations.
The scanning fiber tip is positioned close to the PSD surface. The signal on the PSD is
derived from Equation 4, where I(r) is the spatially dependent intensity, P is the peak power and
w(z) is the mode waist of the Gaussian beam at the PSD (z is the on-axis distance from the fiber
tip to the PSD silicon sensor element).[48]
𝐼𝐼 (𝑟𝑟) =2𝑃𝑃
𝜋𝜋𝑤𝑤(𝑧𝑧)2𝑒𝑒−2𝑟𝑟2 𝑤𝑤(𝑧𝑧)2⁄ (4)
∫ 𝐼𝐼2(𝑟𝑟)𝑑𝑑𝐴𝐴 =𝑃𝑃2
𝜋𝜋𝑤𝑤(𝑧𝑧)2
(5)
Figure 33. Lissajous scan pattern for a (A) 200 ms camera exposure and (B)
a normalized computer display of the PSD signals in a 400 ms interval.
Normalized a.u.
Nor
mal
ized
a.u
.
(A) (B)
43
Signal ∝Peak Power2
Fiber Distance from PSD2 (6)
The two-photon absorption depends on the intensity, which (in addition to the spatial
dependence) also has a time dependence. This can be accounted for as above by multiplying I(r)
by 𝑓𝑓(𝑡𝑡) = 1√2𝜋𝜋𝜎𝜎
e−t2/2𝜎𝜎2. This implies narrowing the pulse and maintaining the same energy
results in increased two-photon absorption.
In Figure 34, a slope of 2.02 (red line) is seen compared to a theoretical slope of 2.0 (dashed
Figure 34. Data taken on PSD signal strength by varying the peak power of
the laser—a result of changing the EYFA pump current.
44
blue-line). Figure 34 has a one-decade vertical change versus a half-decade horizontal change;
therefore the dashed line between scale limits has a log-scale slope of 2, i.e. a P2 signal
dependence. This is the simplest model for 2-photon absorption. The data of Figure 34 are not
sufficient to prove the presence of a 2-photon process, but the best linear fit of their logarithmic
values shows them to be consistent with that model.
In addition to reflection from the TX fiber facet, the reflection of the TX beam by the PSD
can also be a significant contribution to the To pulse. The PSD structure contains several
surfaces, with each contributing to the total reflection. Individual PSD surface reflectivities were
measured by focusing an optical probe beam onto each surface. The reflectivity of the silicon
has the highest reflectivity of any of the surfaces of the PSD (Figure 35 and Figure 36). The
surfaces in Figure 35 (B) going from top to bottom are the two sides of the glass window, the
(A) (B)
Figure 35. (A) On-axis total reflectivity measurement of the silicon in the PSD
with a 1560 +/- 10 nm source. (B) On-axis reflectivity of each surface of the
PSD. The scale ranges from 0% to 2% reflectivity.
45
two sides of the PSD, and the two sides of the other glass window surrounding the silicon PSD.
This data is also shown on an image of the PSD in Figure 36. The initial reflectivity of bare
silicon before AR coating is close to 33%. Our measured silicon reflectivities of 1.90% and
1.71% for the two sides of the silicon is a considerable reduction but could be improved by
optimization of the AR coating; reflectivity of less than 0.25% has been demonstrated [49].
The reflections from the two Si surfaces, combined with those from window surfaces,
qualitatively agree with a separate experiment that measured a 3% total on-axis reflectivity from
the PSD.
4.4 Data acquisition
Figure 36. Side view of the PSD structure and interface reflectivities of
individual surfaces at 1.55 µm.
10mm
.55 mm Glass
.25 mm Air Gap
.35 mm Silicon
.25 mm Air Gap
.55 mm Glass
0.15%0.24%
1.71%
1.9%0.12%
0.24%
fiber
46
A somewhat simplified diagram of the system signal paths is shown in Figure 37. The heart
of the capture system is the AP 240 (Acqiris, now Agilent) pulse analyzer. The AP 240 (after
receiving its own external trigger) sends a trigger to the DFB seed laser of the EYFA and
calculates the time-of-flight from its output trigger to the peak of the return pulse. Using the
AP240 output trigger (rather than a trigger from the SRS pulse generator) to the DFB laser
reduces the internal timing jitter on the measured time-of-flight. The PC uses a National
Instruments (NI) Analog/Digital (A/D), Digital/Analog (D/A) board to excite the PZT vibration
and to capture pulsed analog data from the PSD. The acquisition system is also dependent on a
stable-timebase temperature-compensated crystal oscillator (TCXO) to synchronize the 10 MHz
Figure 37. Block diagram of system signal paths. Data from optical sensor
inputs is in red.
PC
Pulsed fiber Laser (1ns, 50kHz)
Gated APD receiver
Fiber MUX
PZT
2D-vibrating biaxial fiber cantilever
Transmitting 2D position sensing detector (PSD)
scan lens (TX/RX)
A
AP 240-Acqiris Pulse analyzer (PCI)
NI 4-ch A/D (PCI)
2-freq D/A sinusoids
SRS Pulse Generator
sum & difference position outputs
triggers
LADAR return pulse output signal
APD gate
trigger
47
and 20 MHz time-bases and 50 kHz master trigger rate on the NI board, AP 240, and the
Stanford Research Systems (SRS) pulse generator (Figure 38).
The AP 240 has the ability to delay acquiring data, as can be seen in the bottom pulse of the
acquisition system temporal diagram (Figure 39). This delay feature allows transients and
ringing from T0 reflections to be ignored for a short period.
The Acqiris AP240 pulse analyzer performs an analog to digital conversion of the RX pulse
signal in order to calculate range from the pulse time-of-flight. It samples at 2 GSPS, operating
in a peak finding mode that uses interpolated data every 31.25 ps (0.5 ns / 16). The purpose of
Figure 38. LADAR acquisition system synchronization. The thicker arrows
appear in Figure 37.
48
locking the time-bases, and using the AP240 to generate the DFB trigger is to ensure that we can
eliminate 0.5-ns system jitter, and achieve timing precision of the detected pulses accurate to
31.25 ps. The equivalent range precision is 𝑐𝑐∆𝑡𝑡2
= 4.69mm. The timebase of the NI A/D, D/A
card is also locked to ensure that its internal 50 kHz signal processing is sychronous with the
master pulse rate. However, any residual jitter less than a few ns has no impact on the PSD
signal processing or the analog PZT output voltage generation.
Figure 39. Sequence acquisition timing diagram.
49
4.5 Results
4.5.1 Beam analysis
The LADAR system was tested and optimized via several steps. Exciting the fiber vibration
along mainly a single axis allows the system to be easily optimized using the APD oscilloscope
signal instead of the full image acquisition system. The LADAR system TX lens was adjusted
for collimation by using a shear plate to monitor the phase-front, as can be seen by the horizontal
fringes in Figure 40.
Figure 40. Horizontal fringes using a shear plate confirm collimation for on-
axis.
50
To characterize temporal resolution, a tilted target board was placed 7.9 m away from the
LADAR. The board was large enough to encompass the entire angular scan of the LADAR.
Small patches of reflective tape were placed at the center and the edge of the board; due to the
tilt of the board, these generated two time-separated LADAR returns; one corresponding to the
center of the scan and the other from the extreme of the scan angle. The time separation between
the two returns was 3.6 ns, as shown in an oscilloscope trace in Figure 41 , thereby
demonstrating that the system has sufficient temporal resolution.
Figure 41. Oscilloscope return pulses (100 mV/vertical division and 2 ns/
horizontal division) from 7.9 m, separated by 3.6 ns.
51
To model the shape of the laser spot in the far field of the TX lens, a Fourier beam-
propagation model was created to predict the spot intensity distribution as a function of scan
angle. The model first used differential equation analysis of the fiber cantilever to determine
fiber-tip angle as a function of fiber-tip lateral distance from the optical axis. These fiber tip
positions and angles were then used in a Zemax sequential model of the TX collimating lens to
generate an optical phase map of the field exiting the lens for each position. The optical field
(including the phase distortion) exiting the TX lens was then Fourier transformed to find the far-
field TX beam intensity distribution.
Modeled laser spot shapes corresponding to the center of the scan (fiber tip on axis) and
when the fiber tip is displaced approximately 1.5 mm from center are shown in Figure 42.
Captured images of the laser spots are also shown for comparison. Multiple beam spots appear
in the upper left image since the spots were in the camera’s field of view while the fiber was
scanning.
The model and experiment show that the aberrations increase in the laser spot at the edges of
the LADAR scan. It appears that the aberrations are dominated by spherical aberration
(displacement of the fiber tip from the lens focal surface) and coma (off-axis location of the
optical source).[50] In the future these effects might be reduced by using a customized lens
design.
An important measure of RX lens performance is the fraction of power incident on the lens
that is coupled into the 90 um RX fiber. A non-sequential model of the LADAR system (Figure
43) was created in Zemax to examine how well a single lens, set for best collimation, focuses a
52
diffuse reflection back into the 90 µm fiber. CNC-polished aspheres such as Asphericon a30-
Figure 42. Images (at range of 8 m) and models (at range of 100 m) used in
a field curvature analysis with the laser focused at the center (upper tiles) and
the edge of the scan (lower tiles). Due to experimental limitations, the edge-
of-scan image was taken for 1.2 mm offset rather than 1.5 mm offset used in
the model.
Dis
tanc
e (a.
u.)
Dis
tanc
e (a.
u.)
Distance (a.u.)
Distance (a.u.)
mm
53
26hpx, Asphericon a25-20hpx, Asphericon a50-40hpx, and a Lightpath GRADIUM GBX-30-40
lens were investigated and can be chosen depending on the scan angle needed. The GBX-30-40
lens provided a good tradeoff between scan angle, compact size, and small TX/RX spot size.
As shown in Figure 44, the model shows that the curved envelope of the scanning fiber tip
bends in the opposite direction to the GBX-30-40 focal field curvature. Figure 44 was used to
show for a 2 mm fiber offset and 9.2º fiber tip angle, the focus moves away from the fiber by 195
µm (separation between red and green paths in Figure 44).
The fiber tip is bending in the opposite direction of the field curvature by 134 µm at the same
point, further defocusing the return LADAR light. Zemax models suggest that using a meniscus
lens with the GBX-30-40 lens corrects some of the field curvature. Off axis coupling at a 2 mm
fiber offset increases coupling by 25% according to the model. Another viable option is to use a
Figure 43. Non-sequential Zemax LADAR model incorporating fiber offset,
and tilt for system optimization. Rays change color as they pass into a new
medium.
TX beam
RX lightfrom diffuse target
TX/RX fiber
54
staircase lens to correct the field curvature. [51]
The loss parameters for the scanning LADAR are similar as the monostatic rangefinder
shown in Table 2, except for an additional off-axis efficiency loss and a coupling loss due to the
mismatched NA of the scanning fiber tip (0.22) and the TX/RX lens of 0.33. This mismatch
results in a coupling ratio of 0.222
0.332= 0.44.
Multiple modeling approaches using Zemax’s lens file for the GBX-30-40 predicted the
return focused spot to be less than 90 µm. These models and a wavefront measurement showed
less than 6 waves of distortion across the GBX-30-40 lens. However, a careful measurement at 3
Figure 44. Sequential Zemax model demonstrating lens field curvature
effects by focusing diffuse light onto a curved plane.
Enlarged
GBX-30-40 Gradium lens
Fiber tip envelope
Current Focus
55
different distances found the on-axis coupling efficiency to be ~40% through a 90 µm pinhole.
The reasons for this discrepancy will require further investigation. When the coupling loss from
the NA mismatch and the effect of excessive spot size are combined, a total coupling of 18% is
predicted for the RX light into the RX fiber.
4.5.2 3-D target measurements
Monostatic scanning LADAR data was taken in the laboratory and an indoor tunnel. A
pedestal target with five discrete levels, each 1 cm thick, was constructed from squares of foam
board. As shown in Figure 45, the uppermost target square measured 2.5 x 2.5 cm and the
lowest measured 12.5 x 12.5 cm. Figure 45 shows the point cloud LADAR return from the
Figure 45. 3-D Scanning (Monostatic) LADAR point-cloud output at 8 m with
gated out background.
Pedestal target of 2.5 x 2.5 x 1 cm thick, 5.0 x 5.0 x 1 cm, …
3D Scanning Ladar Display (0.4 s frame) at 8 m (False color indicates range difference)
Time slices with 31 ps (0.47 cm rangeseparation)
1°Illumination
On-axis view
1°
56
target placed at 8 m. It can be seen that the system can resolve a clear image of the pedestal
target with a 0.4 s frame using first returns. The PSD data provides the lateral coordinates, and
the measured pulse delay gives the longitudinal (z) coordinate. The data points are also false-
color coded by the delay. For these data, a PZT drive amplitude of 0.13 V gave a Δx of 0.32
mm, and a half scan angle θ = 0.5º.
Data for targets at 26 m range are shown in Figure 46. The center figure is a user-rotatable
(during and after the acquisition) projection of the 3-D (x, y, z) point cloud for each frame; the
right part uses 2-D PSD data for each frame (with fixed orientation), false-color coded by the
range for each point. At longer target ranges and LADAR scan angles, added delay is apparent
in the corners of the image compared to the center of the scan in Figure 46. This is attributed to
changes in the interpolated pulse peak location from variations in a sloping baseline and signal
strength. The former is caused by angular variation of the T0 reflection from the PSD; the latter
Figure 46. 3-D Scanning LADAR point cloud at 26 m. The left pedestal is the
same used in the previous figure and the right pedestal consists of four
square foam slices, ranging from 7.6 cm (3”) to 30.5 cm (12”) wide, each 1 cm
thick.
57
is caused by larger lens aberration with increasing scan angle. The resulting range error in
Figure 46 is less than 5 cm. A constant-fraction threshold circuit, reducing the T0, or improving
the anti-reflection coatings on the PSD would alleviate this problem.
The edges of the scan pattern with some PZT-mounted fibers are not orthogonal, like the
images in Figure 46 and Figure 47 (B). It is hypothesized that this results from coupling of the
vibrational modes to a higher frequency torsional mode of the fiber. Such coupling could be
induced by the PZT vibration forces misaligned from the center of the fiber, or by asymmetry in
the epoxied fiber attachment. A generalized coupled-mode model with the fiber vibrational
modes coupled by a small perturbation gives results consistent with those in Figure 47 (B) and
Figure 48. Some mounted fibers have shown scan edges that are more orthogonal, as seen in
Figure 47 (A). Note that the outer shape of the scan pattern does not affect the shape of scanned
Figure 47. Lissajous scan pattern taken with a SWIR camera for two different
mounted scanning fibers.
(A) (B)
58
objects; square objects will appear square (see Figure 46).
Figure 48. Lissajous scan pattern for a (A) 400 ms camera exposure and (B)
a normalized computer display of the PSD signals in a 400 ms interval.
(A) (B)
59
5.0 Detector and optical reflection considerations
5.1 Zero-time-delay (T0) pulse return
One of the characteristics of a monostatic LRF or LADAR system is a large zero-time-delay
(T0) pulse return caused by residual reflections in the transmit beam optical path. This back
reflection could cause optical damage to the APD and signal saturation which, because of slow
detector recovery, results in a large minimum detection range.
One source of the T0 signal is the reflection of the core light from the AR coated output face
of the DCF fiber. If the fiber face is perpendicular to the fiber axis, all of the reflected light
couples back into the core, and does not couple into the MM pigtail of the APD. However, if the
face angle deviates from perpendicular, some reflected light couples into the DCF cladding and
the MM fiber, generating a T0 pulse.
The fraction of reflected light that couples into the DCF cladding as a function of this
deviation angle was calculated. Although only the fundamental LP01 mode is launched from the
fiber laser into the DCF core, the core of this DCF fiber also supported a second-order LP11
mode. Therefore, this calculation also includes the coupling of the reflected light from LP01 to
LP11. Figure 49 shows the calculated fractions (normalized by unity reflection from the face) of
incident LP01 power coupling back into the DCF cladding, LP01, and LP11 core modes. Only the
cladding-mode reflections are assumed to couple through the MUX into the receiver. The
intensity profiles for LP01 and LP11 modes are also shown for reference. For an angle deviation
60
of 0.1º, achievable by fiber-cleaving or polishing methods and guaranteed by our vendor, the
coupling fraction is predicted to be 4 x 10-5. For an AR coating reflectivity of 0.5%, this would
result in 2 x 10-7 of the incident LP01 power coupling into the DCF cladding. For a 50% DCF-to-
MM fiber transfer ratio, 1 x 10-7 of the laser power would couple into the APD. As indicated
below, the actual T0 powers were significantly larger than this fraction.
The rangefinder experimental results were obtained using a CMC Electronics 200 MHz APD
receiver. The initial T0 return pulse from the end of the output fiber was found to cause a large
saturation of the APD and an offset in the transimpedance amplifier (TIA) output signal, making
Figure 49. The normalized effect of cleave angle versus reflected light.
61
it difficult to have a short minimum range for the present version of the rangefinder system. The
receiver requires 200 ns to recover from T0, and exhibits a large offset due to a slower negative-
going ramp from the broad-band Amplified Spontaneous Emission (ASE) power, that builds-up
in the fiber laser between the pulses, and is reflected into the APD by the fiber output end. In
Figure 18, the return pulse of -42 mV is shown. The pulse sits on top of a -78 mV offset from
the receiver baseline voltage of 850 mV measured in the absence of an optical signal (laser off).
Detected ASE build-up can be suppressed by fiber-laser improvements and a narrow-band
spectral filter before the receiver.
In the LADAR experiment, the fiber tip cleave tilt guaranteed by the vendor was less than 1º.
For a fiber face AR coating reflectivity of 0.5%, based on the plot in Figure 49, this would result
in 3.6 x 10-4 of the incident LP01 power coupling into the DCF cladding; therefore, for a 50%
DCF-to-MM fiber transfer ratio, 1.8 x 10-4 of the laser power would couple into the APD. In this
experiment, 1.0 x 10-4 of the TX (non-ASE) power is measured at the APD. Another possible
contribution to T0 is from leakage of TX light from the fiber core into the DCF cladding at fusion
splices between the isolator and DCF; this light will be reflected by the AR-coated DCF fiber
face and will propagate back towards the APD. Yet another contribution to T0 is TX light
reflected by the PSD, as discussed below. It should be possible to achieve a significant reduction
of the total reflected power by further minimization of these effects and optimization of fiber
face angle.
The PSD can contribute a significant amount of power towards T0 when it is close to the fiber
tip. Figure 50 shows the reflected TX power as a function of the fiber tip-to-PSD distance,
showing that the further the PSD is from the fiber, the less T0 reflection will occur. For the
62
LADAR result shown in Figure 46 the PSD was 2 mm from the fiber tip, and the power reflected
by the PSD, indicated by an X, was on par with the power reflected from the AR coated fiber tip
(corresponding to reflected power in Figure 50 for large PSD-to-fiber separations).
The unused DCF side of the MUX is coated with a light absorbing carbon and petroleum
striping paste for the last 1 cm before the fiber tip. An index matching fluid is also placed on the
fiber tip to reduce stray T0 reflections caused by light leaking from the MUX or from fiber
splices.
5.2 Detector
Figure 50. Varying the PSD to scanning fiber distance effects the T0 pulse
due to PSD reflections.
0
10
20
30
40
50
60
70
80
0 2 4 6 8 10
CW P
ower
(µW
)
Fiber Distance from PSD (mm)
Received T0 power vs. PSD distance
X
63
APD receivers investigated were Spectrolab (75 µm), Voxtel (200 µm), CMC Electronics
(200 µm), and Princeton Lightwave (200 µm). The rangefinder system used the Princeton
Lightwave receiver and the LADAR system used the CMC electronics receiver due to the
detector’s individual recovery performance.
Several Spectrolab APD and trans-impedance amplifier (TIA) dies were investigated on
custom PCBs. Placing a 1 kΩ resistor in between the APD and the TIA limited the current
sufficiently so that the APD would not be overloaded or suffer optical damage. Gating the APD
bias, by switching 5 V over ns time scales to get into a low gain region where the APD would
not be damaged, proved to be difficult; transients from such switching were picked up by the
TIA and amplified, causing ringing that was troublesome to dampen within 50 ns, for a small
minimum range.
A Princeton Lightwave PLA-661 200 µm APD with a 100 MHz TIA was used for the
scanning LADAR system. It uses a limiter capacitor between the TIA and APD to lower the TIA
saturation and allows for quicker pulse recovery. The rangefinder system used a CMC
Electronics P264-339769-101 InGaAs APD, which has a 200 µm diameter active area and 200
MHz bandwidth. Some of the CMC electronics APD designs incorporate a variable TIA gain for
T0 recovery.
An off-the-shelf fiber coupled Voxtel APD receiver was considered. The fiber being used in
the receiver is Nufern’s MM-S105/125-22A (105 µm, NA = 0.22). The receiver also consists of
a 75 µm InGaAlAs APD detector which uses a MAXIM 3277 TIA, post stage amplifier, and a
single-stage thermo-electric cooler. The amplified detector bandwidth is about 2 GHz, which is
sufficient to resolve the 1 ns pulses from the LADAR’s pulsed fiber laser. Optics inside the
64
detector focus the light from the 105 µm fiber onto the 75 µm active region. However, the T0
recovery was not sufficient to be used in either system
Other detector technologies were also considered. A Geiger Mode APD would not work
well due to after-pulsing that gives a dead-time of ~1 µs after the T0 signal. [52] A PIN detector
was not considered due to its lower optical signal sensitivity.
Since neither one of the off-the-shelf 200 µm dimeter APDs used in the systems were fiber
coupled, fiber-to-APD coupling was investigated. Ball, drum, hemispherical, and conical lenses
were compared, but I chose a GRIN lens (Go!Photon PCH-180-022-156.1) due to its availability,
ease of modeling, and performance for the fiber to detector coupling. A Spectrolab APD was
chosen for modeling due to its low capacitance (corresponding to a high-bandwidth) and large
size compared to other APDs. Zemax predicted a 95% fiber-to-APD coupling efficiency (Figure
51) from a 200 µm output fiber of NA = 0.3, or 98% efficiency from a 200 µm fiber of NA =
0.22, into a Spectrolab back illuminated, AR coated, 125 µm, InGaAlAs APD. Even higher
efficiencies are predicted for the 200 µm APDs used with a 200 µm fiber.
Figure 51. High NA, high bandwidth, 125 µm back illuminated detector with
95% estimated fiber-to-APD coupling efficiency.
2.617 mm 3.799 mm
200 µm, NA=0.3 Fiber
0.3 mm
APDGRIN Lens
65
6.0 Conclusion
A monostatic fiber-based LRF was demonstrated with single-pulse ranging to 50 m. It was
calculated that if this system used 1000 waveform-averaged pulses (within a 20 ms interval), the
maximum range limit would increase to 940 m for an 11% reflectivity target.
The work presented here also was the first implementation of a novel, compact, monostatic,
all-fiber scanned LADAR system. It used a combination of several unique components and
techniques: an all-fiber MUX for separating transmit/receive signals, a monostatic all-fiber
optical transmitter-receiver, an all-fiber PZT LADAR beam scanner, and a transmissive silicon
PSD beam tracker that utilized 2-photon absorption.
There are many system tradeoffs in the scanning LADAR system. Some of these tradeoffs
are summarized in Table 3. A larger scan angle can be achieved with a lower NA fiber core,
which gives a smaller spot at the lens, thereby reducing beam clipping. However, this will also
reduce the source spot of the TX collimated beam, resulting in a higher beam divergence and a
larger spot on the target. A larger scan angle can also be achieved by having a shorter focal-
length lens. Lower F-number (higher NA) lenses tend to have more aberrations and thus the RX
focal spot will be larger and will require a fiber with a larger (inner) cladding diameter. The lens
aperture should be as wide as possible to maximize capture of the return light, but consistent
with the lens NA not exceeding the RX fiber cladding NA. The LADAR frame time can be
halved by doubling the laser repetition rate (thus doubling the laser average power to maintain
the same pulse energy) as seen in the simulations of Figure 52. The frame time can also be
66
reduced by decreasing the excitation voltage to the vibrating fiber scanner (covering a smaller
area and requiring fewer pulses) at the expense of decreased angular coverage.
We are continuing to investigate time-gating higher-bandwidth detectors for higher
resolution and faster T0 recovery times. We are also making component improvements that will
reduce the T0 signal and allow for a shorter minimum range. The 3-D images could be even
clearer under operational conditions due to multiple look angles. This would allow further image
manipulation using resampling and gridding. [53-55]
System performance variable
Required system/component changes
Increase maximum range - Fiber laser with higher pulse energy/peak power - APD with improved sensitivity. - Increase MUX RX power transfer efficiency by using a larger fiber coupled-APD. - Improve RX collection efficiency by decreasing lens aberrations and field curvature with added lens elements.
Shorter minimum range Reduce T0 pulse by: Lower loss fiber splices Improved AR coatings on TX/RX fiber tip and PSD Improved APD T0 recovery
Decrease signal processor size, weight, power & cost
Use an embedded high-speed National Semiconductor A/D and microcontroller.
Increase frame rate - Increase PRF and average laser power, maintaining same pulse energy. - Decrease fiber cantilever length to increase fiber resonant frequencies, and compensate for decreased scan amplitude by increasing the PZT drive voltage. - Reduce scan angle at the expense of lower angular coverage. - Reduce number of points collected at the expense of lower angular resolution.
Increase angular scan - Increase PZT drive voltage - Shorter focal length, lower F-number RX/TX lens
Table 3. Future work to increase system performance.
67
Figure 52. Simulations demonstrating the frame time can be halved and
angular resolution maintained by doubling the laser pulse rate and average
power.
68
Bibliography:
1. P. F. McManamon, "Review of ladar: a historic, yet emerging, sensor technology with
rich phenomenology," Optical Engineering 51, 060901-060901-060913 (2012).
2. Laser Radar: Progress and Opportunities in Active Electro-Optical Sensing (The
National Academies Press, 2014).
3. D. C. Carmer and L. M. Peterson, "Laser radar in robotics," Proceedings of the IEEE 84,
299-320 (1996).
4. P. F. McManamon, G. Kamerman, and M. Huffaker, "A history of laser radar in the
United States," in SPIE Defense, Security, and Sensing, (International Society for Optics
and Photonics, 2010), 76840T-76840T-76811.
5. J. Savage, W. Harrington, R. A. McKinley, H. Burns, S. Braddom, and Z. Szoboszlay,
"3D-LZ helicopter ladar imaging system," in SPIE Defense, Security, and Sensing,
(International Society for Optics and Photonics, 2010), 768407-768407-768409.
6. W. C. Stone, M. Juberts, N. Dagalakis, J. Stone, J. Gorman, P. J. Bond, and A. L.
Bement, "Performance analysis of next-generation LADAR for manufacturing,
construction, and mobility," (2004).
7. W. C. Stone, M. Juberts, N. Dagalakis, J. Stone, J. Gorman, P. J. Bond, and A. L.
Bement, "Performance analysis of next-generation LADAR for manufacturing,
construction, and mobility," Report NISTIR 7117. (May 2004).
69
8. C. Ye and J. Borenstein, "Characterization of a 2D laser scanner for mobile robot
obstacle negotiation," in Robotics and Automation, 2002. Proceedings. ICRA'02. IEEE
International Conference on, (IEEE, 2002), 2512-2518.
9. B. L. Stann, J. F. Dammann, M. M. Giza, P.-S. Jian, W. B. Lawler, H. M. Nguyen, and L.
C. Sadler, "MEMS-scanned ladar sensor for small ground robots," in SPIE Defense,
Security, and Sensing, (International Society for Optics and Photonics, 2010), 76841E-
76841E-76812.
10. M. A. Powers, B. L. Stann, and M. M. Giza, "Agile beam laser radar using computational
imaging for robotic perception," in 2015), 94650E-94650E-94611.
11. A. Schneider, Z. La Celle, A. Lacaze, K. Murphy, M. Del Giorno, and R. Close, "Sensor
study for high speed autonomous operations," in 2015), 949408-949408-949415.
12. L. Goldberg and S. Chinn, "Compact monostatic optical transmitter and receiver," (US
Patent # 8,730,456, May 20, 2014).
13. S. Chinn and L. Goldberg, "Compact fiber-based scanning laser detection and ranging
system," (US Patent # 8,946,637, Feb. 3, 2015).
14. D. A. Roberts and R. R. Syms, "1D and 2D laser line scan generation using a fiber optic
resonant scanner," in Symposium on Applied Photonics, (International Society for Optics
and Photonics, 2000), 62-73.
15. B. Pal, "Fabrication and modeling of fused biconical tapered fiber couplers," Fiber and
integrated optics 22, 97-117 (2003).
16. G. Z. Wang, K. A. Murphy, and R. O. Claus, "Effect of external index of refraction on
multimode fiber couplers," Applied optics 34, 8289-8293 (1995).
70
17. K. Ogawa, "Simplified Theory of the Multimode Fiber Coupler," Bell System Technical
Journal 56(1977).
18. S. Lemire-Renaud, M. Rivard, M. Strupler, D. Morneau, F. Verpillat, X. Daxhelet, N.
Godbout, and C. Boudoux, "Double-clad fiber coupler for endoscopy," Optics express 18,
9755-9764 (2010).
19. C. Cryan and C. Hussey, "Annealed twists for stable tuned fused polished couplers,"
Electronics Letters 29, 53-54 (1993).
20. C. E. Chryssou, "Theoretical analysis of tapering fused silica optical fibers using a carbon
dioxide laser," Optical Engineering 38, 1645-1649 (1999).
21. T. E. Dimmick, G. Kakarantzas, T. A. Birks, and P. S. J. Russell, "Carbon dioxide laser
fabrication of fused-fiber couplers and tapers," Applied Optics 38, 6845-6848 (1999).
22. T. Ozeki and B. Kawasaki, "Efficient power coupling using taper-ended multimode
optical fibres," Electronics Letters 12, 607-608 (1976).
23. E. G. Rawson and A. B. Nafarrate, "Star couplers using fused biconically tapered
multimode fibres," Electronics Letters 14, 274-275 (1978).
24. A. Robertson, "Fused fiber components for high-power fiber lasers," in Proc. SPIE,
2004), 147-154.
25. C. Rowe, D. Mortimore, I. Wilkinson, and N. Achurch, "Fabrication of Fused Fibre
Devices," (WO Patent 1,991,003,436, 1991).
26. D. B. Mortimore, "Optical fused couplers," (Google Patents, 1989).
27. K. Tajima, M. Tateda, and M. Ohashi, "Viscosity of GeO2 - doped silica glasses,"
Lightwave Technology, Journal of 12, 411-414 (1994).
71
28. J. M. Ward, D. G. OShea, B. J. Shortt, M. J. Morrissey, K. Deasy, and S. G. Nic
Chormaic, "Heat-and-pull rig for fiber taper fabrication," Review of scientific
instruments 77, 083105-083105-083105 (2006).
29. A. D. McLachlan and F. P. Meyer, "Temperature dependence of the extinction coefficient
of fused silica for CO2 laser wavelengths," Applied optics 26, 1728-1731 (1987).
30. B. Wang and E. Mies, "Review of fabrication techniques for fused fiber components for
fiber lasers," in SPIE LASE: Lasers and Applications in Science and Engineering,
(International Society for Optics and Photonics, 2009), 71950A-71950A-71911.
31. D. R. Moore, V. J. Tekippe, R. Fechner, and S. Bernhardt, "Recent advances in fused
fiber optic coupler technology," in Video Communications and Fiber Optic Networks,
(International Society for Optics and Photonics, 1993), 28-42.
32. J. W. Goodman, Speckle phenomena in optics: theory and applications (Roberts and
Company Publishers, 2007).
33. P. Rugeland, C. Sterner, and W. Margulis, "Monolithic interferometers using Gemini
fiber," Photonics Technology Letters, IEEE 23, 1001-1003 (2011).
34. A. Loriette, "Monolithic Interferometer in twin optical fibers," (2007).
35. C. Alegria and M. Zervas, "Nondestructive coupler characterization technique," Journal
of lightwave technology 20, 1034 (2002).
36. J. Leach, S. R. Chinn, and L. Goldberg, "Monostatic All-Fiber Rangefinder System," in
CLEO: Science and Innovations, (Optical Society of America, 2015), STh3O. 1.
37. J. H. Leach, S. R. Chinn, L. Goldberg, and S. A. Mathews, "Monostatic all-fiber
rangefinder system," Applied Optics 54, 7687-7692 (2015).
72
38. J. Mindly, W. Barnes, R. Laming, P. Morkel, J. Townsend, S. Grubb, and D. Payne,
"Diode-Array Pumping of Er3+/Yt> 3+ Co-Doped Fiber Lasers and Amplifiers," IEEE
Photonics Technology Letters 5(1993).
39. "ACAM GP22" (2015), retrieved http://www.pmt-fl.com/tdc/tdc-gp22.php.
40. Z. Li, Z. Yang, and L. Fu, "Scanning properties of a resonant fiber-optic piezoelectric
scanner," Review of Scientific Instruments 82, 123707-123707-123705 (2011).
41. L. Huo, J. Xi, Y. Wu, and X. Li, "Forward-viewing resonant fiber-optic scanning
endoscope of appropriate scanning speed for 3D OCT imaging," Optics express 18,
14375-14384 (2010).
42. X. Liu, M. J. Cobb, Y. Chen, M. B. Kimmey, and X. Li, "Rapid-scanning forward-
imaging miniature endoscope for real-time optical coherence tomography," Optics letters
29, 1763-1765 (2004).
43. S. Moheimani, "Invited review article: Accurate and fast nanopositioning with
piezoelectric tube scanners: Emerging trends and future challenges," Review of Scientific
Instruments 79, 071101-071101-071111 (2008).
44. S. Moon, S.-W. Lee, M. Rubinstein, B. J. Wong, and Z. Chen, "Semi-resonant operation
of a fiber-cantilever piezotube scanner for stable optical coherence tomography
endoscope imaging," Optics express 18, 21183 (2010).
45. M. Mokhtar and R. Syms, "Tailored fibre waveguides for precise two-axis Lissajous
scanning," Optics Express 23, 20804-20811 (2015).
46. "SiTek S2-03162D silicon PSD" (2015), retrieved http://www.sitek.se.
73
47. A. D. Bristow, N. Rotenberg, and H. M. Van Driel, "Two-photon absorption and Kerr
coefficients of silicon for 850–2200," Applied physics letters 90, 191104-191104-191103
(2007).
48. E. Saleh and M. Teich, Fundamentals of Photonics (2007), p. 79.
49. "AccuCoat AR Coating on Silicon Wafer", retrieved
http://www.accucoatinc.com/ac_155.html.
50. M. Born and E. Wolf, Principles of optics: electromagnetic theory of propagation,
interference and diffraction of light, 4th ed., Ch. 9 (1970).
51. J. M. Sasian and R. A. Chipman, "Staircase lens: a binary and diffractive field curvature
corrector," Applied optics 32, 60-66 (1993).
52. M. A. Itzler, X. Jiang, R. Ben-Michael, B. Nyman, and K. Slomkowski, "Geiger-mode
APD single photon detectors," in Optical Fiber Communication Conference, (Optical
Society of America, 2008).
53. P. Cho, H. Anderson, R. Hatch, and P. Ramaswami, "Real-time 3D ladar imaging," in
Defense and Security Symposium, (SPIE, 2006), 62350G-62350G-62312.
54. O. Kreylos, G. Bawden, and L. Kellogg, "Immersive Visualization and Analysis of
LiDAR Data," in Advances in Visual Computing, G. Bebis, R. Boyle, B. Parvin, D.
Koracin, P. Remagnino, F. Porikli, J. Peters, J. Klosowski, L. Arns, Y. Chun, T.-M.
Rhyne, and L. Monroe, eds. (Springer Berlin Heidelberg, 2008), pp. 846-855.
55. A. N. Vasile, L. J. Skelly, M. E. O’Brien, D. G. Fouche, R. M. Marino, R. Knowlton, M.
J. Khan, and R. M. Heinrichs, "Advanced Coincidence Processing of 3D Laser Radar
Data," in Advances in Visual Computing (Springer, 2012), pp. 382-393.