University of Calgary PRISM: University of Calgary's Digital Repository Graduate Studies The Vault: Electronic Theses and Dissertations 2016 CMOS Laser Diode Drivers for Supercontinuum Generation He, Yuting He, Y. (2016). CMOS Laser Diode Drivers for Supercontinuum Generation (Unpublished master's thesis). University of Calgary, Calgary, AB. doi:10.11575/PRISM/25339 http://hdl.handle.net/11023/3030 master thesis University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission. Downloaded from PRISM: https://prism.ucalgary.ca
82
Embed
CMOS Laser Diode Drivers for Supercontinuum Generation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
University of Calgary
PRISM: University of Calgary's Digital Repository
Graduate Studies The Vault: Electronic Theses and Dissertations
2016
CMOS Laser Diode Drivers for Supercontinuum
Generation
He, Yuting
He, Y. (2016). CMOS Laser Diode Drivers for Supercontinuum Generation (Unpublished master's
thesis). University of Calgary, Calgary, AB. doi:10.11575/PRISM/25339
http://hdl.handle.net/11023/3030
master thesis
University of Calgary graduate students retain copyright ownership and moral rights for their
thesis. You may use this material in any way that is permitted by the Copyright Act or through
licensing that has been assigned to the document. For uses that are not allowable under
copyright legislation or licensing, you are required to seek permission.
Downloaded from PRISM: https://prism.ucalgary.ca
UNIVERSITY OF CALGARY
CMOS Laser Diode Drivers for Supercontinuum Generation
There have been intense research efforts on developing compact and low-cost supercontinuum
generation (SCG) systems, which have various application areas including telecommunications,
spectroscopy, and optical coherence tomography. This research employs complementary metal–
oxide–semiconductor (CMOS) technology to design and implement two integrated laser diode
drivers for reducing the size and cost of SCG systems. A continuous-wave CMOS driver with a
maximum output current of 600 mA is developed for driving a laser diode in an erbium-doped
fiber amplifier (EDFA). A picosecond pulsed CMOS driver is designed and applied for gain-
switching a laser diode to produce optical pulses with a pulse width of 200 ps and a repetition rate
of 5.6 MHz. The gain-switched laser diode output is amplified by an EDFA and then launched into
a highly nonlinear fiber for SCG. The generated supercontinuum has an average power of 62 mW
and a spectral bandwidth of 806 nm.
iii
Acknowledgements
Firstly, I would like thank my supervisor Dr. Orly Yadid-Pecht for her invaluable guidance,
support and encouragement throughout my research work.
Secondly, I thank all the I2Sense lab students, post-doctoral fellows, my friends throughout
the ECE department and department staff for their help and support. Many thanks to Dr. Kartikeya
Murari, Dr. Yuhua Li, Michael Himmelfarb, Nikhil Vastarey, Pauling Cummings, Dr. J.P.E.
Hadden, Prasoon Ambalathankandy, Matthew Jackson, Dr. Arthur Spivak, Linhui Yu, Donuwan
Navaratne, Zhixing Zhao, Christopher Simon, Kathryn Simon and Shem Chenoo for their helpful
discussion and support regarding this thesis work.
Thirdly, I would like to thank my committee members Dr. Paul Barclay and Dr. Leonid
Belostotski for taking their time to serve in my committee.
Last but not least, I would like to thank the CMC Microsystems for the access to the design
tools, workshops and fabrication services through MOSIS. Special thanks to Dr. Shahriar
Mirabbasi from UBC for organizing the CMOS Electronics for Photonics training course and for
helpful discussions regarding the pulsed CMOS driver design during the course.
iv
Dedication
To my parents for their unconditional love and support!
v
Table of Contents
Abstract ............................................................................................................................... ii Acknowledgements ............................................................................................................ iii Dedication .......................................................................................................................... iv Table of Contents .................................................................................................................v List of Tables .................................................................................................................... vii List of Figures .................................................................................................................. viii List of Abbreviations and Symbols......................................................................................x
Chapter 1: Introduction ........................................................................................................1 1.1 Background and Review ............................................................................................1 1.2 Motivation and Objectives .........................................................................................4 1.3 Thesis Scope and Contributions ................................................................................6 1.4 Thesis Outlines ..........................................................................................................7
Chapter 2: High Current Continuous-Wave CMOS Laser Diode Driver ............................8 2.1 Introduction and Objectives .......................................................................................8 2.2 Characteristics of Laser Diodes .................................................................................9 2.3 Continuous-Wave CMOS Driver Circuit Design Methodology ..............................11
2.3.1 Current Reference Circuit ................................................................................12 2.3.2 Start-up Circuit ................................................................................................14 2.3.3 Current Source Circuit .....................................................................................14
2.4 Circuit Design and Simulations ...............................................................................15 2.5 Measurements ..........................................................................................................19
2.5.1 Measurement Setup .........................................................................................19 2.5.2 Measurements of the Output Current ..............................................................21 2.5.3 Measurements of the Optical Output Power ...................................................22
Chapter 4: Supercontinuum Generation in a Highly Nonlinear Fiber Using CMOS Laser Diode Drivers.......................................................................................................................45
Chapter 5: Conclusions and Future Work ..........................................................................62 5.1 Conclusions ..............................................................................................................62 5.2 Future Work .............................................................................................................63
Table 2.1: Design parameters of transistors in the proposed CMOS circuit ................................ 16
Table 2.2: Performance summary of the designed CW pumped laser diode ................................ 25
Table 3.1: Performance summary of the designed pulsed laser diode .......................................... 44
Table 4.1: Optical properties of the HNLF ................................................................................... 49
Table 4.2: Characteristics of the amplified laser diode pulses ...................................................... 53
Table 4.3: Performance summary of the seed laser module ......................................................... 57
Table 4.4: Comparison of the designed SCG system and the reference SCG system .................. 61
viii
List of Figures
Figure 1.1: Simplified SCG process in the spectral domain ........................................................... 1
Figure 2.1: Block diagram of a basic EDFA configuration ............................................................ 8
Figure 2.2: Typical output light versus injection current (L-I) curve of laser diodes ................... 10
Figure 2.3: Typical forward voltage versus injection current (V-I) curve of laser diodes ........... 10
Figure 2.4: Block diagram of the CW laser diode driver circuit design methodology ................. 11
Figure 2.5: Circuit schematic of the proposed CMOS laser diode driver ..................................... 11
Figure 2.6: Simulation results of transistors’ operation regin versus the resistance ..................... 17
Figure 2.7: Comparison of modelling current output and simulation current output at the resistance of the off-chip resistor range from 1100 Ω to 1950 Ω. ........................................ 18
Figure 2.8: Die micrograph of the fabricated CW laser diode driver ........................................... 19
Figure 2.9: Package of a complete driver circuit with two identical CMOS dies; Note that the designed driver circuit is only one small part of the whole CMOS die, there are unrelated circuits shared on the same die. ............................................................................................ 20
Figure 2.10: (a) Host PCB with a driver package and four identical potentiometers which controls three current output ports; (b)Laser diode package mounted on a PCB ................. 20
Figure 2.11: Comparison of the output current between the measurement result, simulation result and mathematical modelling result; Note that the resistance at x-axis represents the resistance of each one potentiometer on the PCB in the measurement. ............................... 21
Figure 2.12: Measured data of optical output power with respect to the injection current, and the fitting curve based on measured data .............................................................................. 22
Figure 2.13: Relation curve of the optical output power of the laser diode and the resistance of potentiometers ....................................................................................................................... 23
Figure 2.14: Stability test of the optical output power over a 400 minute period. ....................... 24
Figure 3.1: Evolution of the photon and carrier density during a gain-switching cycle [37] ....... 27
Figure 3.2: Block diagram of the proposed pulsed laser driver circuit design ............................. 28
Figure 3.3: Design methodology for generating pulse waves ....................................................... 29
Figure 3.4: Circuit design schematic of the VCRO ...................................................................... 30
Figure 3.5: Schematic of the VCDL circuit .................................................................................. 31
ix
Figure 3.6: Timing sequence diagram of the VCDL operation .................................................... 32
Figure 3.7: Design schematic of the XOR circuit ......................................................................... 33
Figure 3.8: Negative power supply operation of the laser diode .................................................. 34
Figure 3.9: Post-layout transient simulation of the proposed CMOS laser driver: (a) the output current waveform; (b) a sample current pulse. ..................................................................... 36
Figure 3.10: Output plots when tuning (a) repetition rate and (b) pulse width in simulations ..... 37
Figure 3.11: (a) Die micrograph and (b) PCB layout of the pulsed laser source .......................... 38
Figure 3.12: (a) Laser pulse waveform and (b) one Gaussian fitted laser pulse ........................... 39
Figure 3.13: Optical spectrum of the pulsed laser diode output ................................................... 40
Figure 3.14: Measured results of tuning the optical output pulses’ repetition rate (a) by adjusting control voltage Vvar and (b) by adjusting control voltage Vctr ............................... 42
Figure 3.15: Measured results of tuning the optical output pulses’ pulse width (a) by adjusting control voltage Vvar2 and (b) by adjusting control voltage Vb .............................................. 42
Figure 4.1: Supercontinuum generation system design block diagram ........................................ 45
Figure 4.2: EDFA amplification mechanism based on an energy-level diagram of erbium ions . 46
Figure 4.3: Block diagram of the EDFA module setup ................................................................ 47
Figure 4.4: Spectrum of the output light from the EDFA monitor output .................................... 48
Figure 4.5: Experimental setup diagram of the SCG system ........................................................ 53
Figure 4.6: Spectrum of supercontinuum output .......................................................................... 54
Figure 4.7: Supercontinuum output evolution at different pump power level .............................. 55
Figure 4.8: Temporal profile of the seed laser pulse, the EDFA monitor pulse and the supercontinuum pulse ........................................................................................................... 56
Figure 4.9: Output optical pulses from the commercial seed laser module: the left figure shows the pulse waveform; the right figure shows a single pulse shape. ........................................ 58
Figure 4.10: (a) Spectrum of optical pulses output from the seed laser module; (b) spectrum of the amplified pulses output from the monitor of EDFA ....................................................... 58
Figure 4.11: SCG results from two different systems .................................................................. 60
x
List of Abbreviations and Symbols
Abbreviation Definition ASE Amplified Spontaneous Emission CBL Current-Balanced Logic CDS Cadence Design System CFP Ceramic Flat Package CMOS Complementary Metal-Oxide-Semiconductor CW Continuous-Wave DFB Distributed Feedback DWDM Dense Wavelength-division Multiplexing EDA Electronic Design Automation EDFA Erbium-doped Fiber Amplifier FC Fiber Connector FS Fusion Splicing FWHM Full Width at Half Maximum FWM Four-Wave Mixing HNLF Highly Nonlinear Fiber MI Modulation Instability NMOS N-channel Metal-Oxide-Semiconductor OCT Optical Coherence Tomography OSA Optical Spectrum Analyzer PCB Printed Circuit Board PCF Photonic Crystal Fiber PDK Process Design Kit PMOS P-channel Metal-Oxide-Semiconductor RMS Root Mean Square SCG Supercontinuum Generation SMF Single-Mode Fibers SRS Stimulated Raman Scattering TEC Thermo-Electric Cooler VCDL Voltage-Controlled Delay Line VCRO Voltage-Controlled Ring Oscillator XOR Exclusive–OR ZDW Zero Dispersion Wavelength
xi
Symbol Definition Ith Threshold injection current of laser diodes ROUT Off-chip resistor IREF Reference current IO Output current 𝜇𝜇𝑛𝑛𝐶𝐶𝑜𝑜𝑜𝑜 Transconductance parameter of CMOS transistors Mi Transistor number i L Length of CMOS transistors Vth Threshold voltage Vgs Gate source voltage of CMOS transistors K An integer KI An integer L Optical output power I Injection current N Number of delay stages 𝑓𝑓𝑜𝑜𝑜𝑜𝑜𝑜 Oscillation frequency td Delay time of each delay stage CL Load capacitance Vosc Oscillation voltage amplitude Ictrl Control current Vvar Varactor control voltage Vctr Transistors’ current control voltage Vb Control voltage Rin Input impedance Vf Forward voltage If Forward current 𝑡𝑡𝐹𝐹𝐹𝐹𝐹𝐹𝑀𝑀 Duration time of FWHM a1, b1, c1 Coefficients of the Gaussian pulse Ppeak Peak power Pavg Average power fRR Repetition rate Ep Pulse Energy ω Frequency γ Nonlinear coefficient
1
Chapter 1: Introduction
1.1 Background and Review
A supercontinuum is a special type of light with high intensity, broad spectral bandwidth
and a high degree of spatial coherence. It is considered to be the combination of a lamp with
broadband spectrum and a laser with high brightness. This kind of light does not exist in nature. It
only occurs when high intensity laser light interacts with nonlinear media. The nonlinear process
that broadens the spectrum of the laser light to produce the supercontinuum is called
supercontinuum generation (SCG). A simplified SCG process in the spectral domain is illustrated
in Figure 1.1.
λ λ0 λ λ0 λ1 λ2
Nonlinear Process
Laser Light Spectrum Supercontinuum Spectrum
Figure 1.1: Simplified SCG process in the spectral domain
SCG was first observed by Alfano and Shapiro back in 1970s [1]. They focused powerful
picosecond pulses into a bulk BK7 glass to generate a white light supercontinuum covering a
spectral range of 400 nm to 700 nm. Since then, SCG has been observed when intense picosecond
or femtosecond laser light is incident on various nonlinear materials which can be solid, liquid and
gaseous [2], [3]. SCG in optical fibers was first observed in 1976 by Lin and Stolen [4]. They used
2
nanosecond laser pulses at kilowatt peak power levels to pump the conventional silica fiber in
order to generate a supercontinuum with spectral bandwidth of 180 nm and spectrum centered at
530 nm. Many subsequent research efforts have been made for SCG in standard single-mode
optical fibers [5], [6]. As the nonlinearity of standard fibers is relatively low, it requires pump
lasers with high peak power levels to generate significant spectral broadening in standard fibers.
However, laser light with a high peak power can damage the silica fibers, which limits the power
level and consequently the spectrum bandwidth of the supercontinuum.
The advent of photonic crystal fiber (PCF) greatly improved SCG technology. PCFs are
produced by the cladding of an optical fiber incorporated with photonic crystals that are dielectric
periodic structures on the scale of a wavelength of light. The PCF has the advantage of high
nonlinearity as well as designable dispersion, which enhance nonlinear effects and obtain much
broader spectra than the standard fibers [7]. The first octave spanning SCG was accomplished with
a PCF pumped by nanojoule pulses from a Ti:Sapphire laser in the year 2000 [8].
The successful SCG designs with PCFs have inspired similar work using optical tapered
fibers [9] or highly nonlinear fibers (HNLFs) [10]. Both the optical tapered fibers and HNLFs
present comparable enhanced nonlinearity, which makes them good alternative to PCFs. Optical
tapered fibers are produced by gently stretching optical fibers to a thin core diameter. Optical
tapered fibers have designable dispersion, a controllable tapering process and a reduced fiber
diameter, all of which enhance the SCG performance. SCG in a tapered fiber with a diameter of 2
µm and a length of 90 mm was reported to have a broad spectrum output of more than two octaves
(370 -1545 nm) [9].
HNLFs are produced by fabricating optical fibers with a narrow core and high material
doping level, which reduces the effective core area and thus enhances the nonlinearity. Compared
3
to the PCF and tapered fibers, HNLFs are the easiest to fabricate but exhibit the lowest
nonlinearity. As HNLFs are single-mode optical fibers, they can be coupled with standard single-
mode fibers (SMF) at a low coupling loss, which is an advantage of HNLFs based all fiber SCG
systems. SCG in a 200-meter-long HNLF pumped by a femtosecond fiber laser with 110 fs pulses
at 1550 nm was reported to have a spectrum spanning from 1100 nm to 2100 nm [10].
SCG in optical fibers has found numerous novel applications in the field of
telecommunication [11], optical frequency metrology [12], [13], optical coherence tomography
(OCT) [14] and spectroscopy [15]. For example, in the field of telecommunications, SCG systems
can be used in dense wavelength-division multiplexing (DWDM) systems. One can use optical
filters to slice the supercontinuum spectrum, so that thousands of single wavelengths of laser light
can be obtained and applied to a number of transmission channels. This approach can realize a
high transmission rate with only one light source.
4
1.2 Motivation and Objectives
Although SCG systems have great potential to be applied in these cutting-edge areas, most
commercial SCG systems nowadays are just used as a laboratory tool. Current commercial SCG
systems are still limited in performance by the availability of suitable wavelength ranges and
limited by size, cost, and power. There is high demand for the development of compact, low-cost
and reliable SCG systems that are practical and accessible for use in various application areas [16].
As the supercontinuum is generated when laser light interacts with a nonlinear medium, SCG
systems in general contain two components, which are a laser light source and a nonlinear medium.
Numerous experimental results using different lasers and nonlinear optical media for SCG have
been reported. PCFs, optical tapered fibers, HNLFs, dispersion-shifted fibers [17], and even silicon
waveguides [18] have proven to be successful nonlinear optical media. Mode-locked lasers [19],
A block diagram of a CW CMOS laser diode driver circuit design methodology is shown in
Figure 2.4. The driver circuit is a current source. There are three sub-circuits including a start-up
circuit, a current reference circuit and a current source circuit. The start-up circuit ensures that the
current reference circuit is turned on. The current source circuit amplifies the reference current
generated from the current reference circuit. Based on this methodology, a CMOS laser diode
driver circuit is proposed and shown in Figure 2.5.
Figure 2.4: Block diagram of the CW laser diode driver circuit design methodology
M6
VDD
M5
M7
M3 M4 M0
M1 M2
RO UT
Laser Diode
IREF IO
Start-up CurrentReference
CurrentSource
Figure 2.5: Circuit schematic of the proposed CMOS laser diode driver
Start-up Circuit Current Reference CircuitTurn On
Current Source CircuitAmplified
12
2.3.1 Current Reference Circuit
This proposed current reference circuit shown in the center of Figure 2.5 is based on the
structure of beta-multiplier voltage reference [31]. Transistors M1-M4 are self-biased to operate in
the saturation region. The p-channel metal-oxide-semiconductor (PMOS) current mirror M3 and
M4 force the same current through each leg of the circuit. An off-chip resistor ROUT has been placed
between the source of M2 and ground. The size of M2 is made larger than that of Ml so that the
difference in the gate to source voltage of Ml and M2 is dropped across ROUT. Through adjusting
the resistance of the off-chip resistor, the reference current will be adjusted accordingly.
As the PMOS M3 current mirrors the PMOS M4, the current I1 passing through M3 and M1
is the same as the reference current IREF passing through M4 and M2, which is expressed as
The gate source voltage of M1, Vgs1 equals the sum of the gate source voltage of M2, Vgs2, and the
voltage across the resistor ROUT, which is shown as
𝑉𝑉𝑔𝑔𝑜𝑜1 = 𝑉𝑉𝑔𝑔𝑜𝑜2 + 𝐼𝐼𝑅𝑅𝑅𝑅𝐹𝐹𝑅𝑅𝑂𝑂𝑂𝑂𝑂𝑂. (2.2)
As all transistors are operated in the saturation region, the current I1 and IREF can be described as
(neglecting the channel-length modulation)
𝐼𝐼1 = 𝜇𝜇𝑛𝑛𝐶𝐶𝑜𝑜𝑜𝑜𝐹𝐹12𝐿𝐿1
(𝑉𝑉𝑔𝑔𝑜𝑜1 − 𝑉𝑉𝑡𝑡ℎ1)2
𝐼𝐼𝑅𝑅𝑅𝑅𝐹𝐹 = 𝜇𝜇𝑛𝑛𝐶𝐶𝑜𝑜𝑜𝑜𝐹𝐹22𝐿𝐿2
(𝑉𝑉𝑔𝑔𝑜𝑜2 − 𝑉𝑉𝑡𝑡ℎ2)2, (2.3)
where 𝜇𝜇𝑛𝑛𝐶𝐶𝑜𝑜𝑜𝑜 is the transconductance parameter of CMOS transistors, W1 and L1 are the width and
length of the transistor M1, Vth1 is M1’s threshold voltage, W2 and L2 are the width and length of
the transistor M2 and Vth2 is M2’s threshold voltage.
Based on Equation (2.3), Vgs1 and Vgs2 can be rewritten as
𝐼𝐼1 = 𝐼𝐼𝑅𝑅𝑅𝑅𝐹𝐹. (2.1)
13
𝑉𝑉𝑔𝑔𝑜𝑜1 = 2𝐼𝐼1𝛽𝛽1
+𝑉𝑉𝑡𝑡ℎ1
𝑉𝑉𝑔𝑔𝑜𝑜2 = 2𝐼𝐼𝑅𝑅𝑅𝑅𝑅𝑅𝛽𝛽2
+𝑉𝑉𝑡𝑡ℎ2,
where β1 = µnCoxW1L1
,β2 = µnCoxW2L2
.
(2.4)
Neglecting the body effect from M2, for the same CMOS technology, the two threshold voltages
Vth1 and Vth2 should be equal
𝑉𝑉𝑡𝑡ℎ1 = 𝑉𝑉𝑡𝑡ℎ2. (2.5)
By substituting Equation (2.1) (2.4) (2.5), Equation (2.2) can be rewritten
2𝐼𝐼𝑅𝑅𝑅𝑅𝑅𝑅
𝛽𝛽1= 2𝐼𝐼𝑅𝑅𝑅𝑅𝑅𝑅
𝛽𝛽2+ 𝐼𝐼𝑅𝑅𝑅𝑅𝐹𝐹𝑅𝑅𝑂𝑂𝑂𝑂𝑂𝑂. (2.6)
The reference current can be derived by solving Equation (2.6)
𝐼𝐼𝑅𝑅𝑅𝑅𝐹𝐹 = (1 − 1√𝐾𝐾
)2 2𝛽𝛽1𝑅𝑅𝑂𝑂𝑂𝑂𝑂𝑂2
where 𝐾𝐾 = (𝐹𝐹1𝐿𝐿1
)/(𝐹𝐹2𝐿𝐿2
). (2.7)
Equation (2.7) shows that the reference current is only dependent on the resistance of the off-chip
resistor and the device parameters of transistors. The device parameters of CMOS transistors are
fixed after fabrication of the design. The reference current can only be tuned by adjusting the
resistance of the off-chip resistor.
14
2.3.2 Start-up Circuit
When solving Equation (2.6), there is another scenario except the answer indicated in
Equation (2.7). That is the zero reference current scenario. In this scenario, the circuit is completely
off even after the power supply is on. This is possible because all transistors are self-biased.
To eliminate the zero reference current scenario, a start-up circuit shown in the left part of
Figure 2.5 is introduced into the design. This start-up circuit is able to initiate the current reference
circuit from a dead (zero current) operating point to its normal operating point [32]. When the
current reference circuit is at a dead operating point, the start-up circuit sets the drain voltage of
the transistor M1, and initiates M1 to draw current. Once the start-up transistor M7 provide a current
path between the supply voltage and ground, the transistors M1, M2, M3, and M4 operate normally
and the reference current will reach the desired amount. After the current reference circuit is turned
on, the gate source voltage of M7 drops to below the threshold voltage and no current flows through
M7. Thus, the start-up circuit has no impact on the value of the reference current.
2.3.3 Current Source Circuit
The current source circuit is shown in the right part of Figure 2.5. It is a single PMOS with
a bias voltage from the reference current circuit. This current source configuration extends the
output voltage to accommodate the required forward voltage of the laser diode.
The gate bias voltage of transistor M0 is connected to the gate bias voltage of transistor M4.
This mirrors the current along M0 and M4. Since M4 operates under the saturation region, M0 does
in the same mode. Thus, the output current Io is
𝐼𝐼𝑂𝑂 = 𝜇𝜇𝑛𝑛𝐶𝐶𝑜𝑜𝑜𝑜𝐹𝐹02𝐿𝐿0
(𝑉𝑉𝑔𝑔𝑜𝑜0 − 𝑉𝑉𝑡𝑡ℎ0)2. (2.8)
15
Since the gate-source voltage Vgs0 and the threshold voltage Vth0 of the transistor M0 are the
same as the gate-source voltage and the threshold voltage of the transistor M4, the amplification
ratio KI of the output current over the reference current can be expressed as
𝐾𝐾𝐼𝐼 = 𝐼𝐼𝑂𝑂𝐼𝐼𝑅𝑅𝑅𝑅𝑅𝑅
= (𝐹𝐹0𝐿𝐿0
)/(𝐹𝐹1𝐿𝐿1
). (2.9)
By substituting IREF with the result from Equation (2.7), the output current can be expressed as
𝐼𝐼𝑂𝑂 = 𝐾𝐾𝐼𝐼(1 − 1√𝐾𝐾
)2 2𝛽𝛽1𝑅𝑅𝑂𝑂𝑂𝑂𝑂𝑂2
. (2.10)
Equation (2.10) shows that the output current has the same output characteristics as the
reference current. The tunable output current is achieved by adjusting the resistance of the off-chip
resistor. Equation (2.10) also shows that the output current is independent of the supply voltage. It
is desirable for the current source to be insensitive of the supply voltage, as most power supplies
fluctuate. By utilizing the self-biasing technique, power supply sensitivity can be greatly reduced.
Self-biasing technique means the biasing voltages especially gate biasing voltages are not
connected to the supply power directly. All transistors in this proposed design are self-biased.
Thus, the output is insensitive to the power supply voltage.
2.4 Circuit Design and Simulations
CMOS circuits are often designed with electronic design automation (EDA) software.
Cadence Design System (CDS) is a ubiquitous commercial tool for CMOS circuit schematic
design, simulation, layout design and verification. To start a CMOS circuit design with CDS, an
industry-standard CMOS process with a foundry-certified process design kit (PDK) needs to be
determined. As the drain voltage of M0 requires a minimum 2 V output voltage to accommodate a
2 V forward voltage of the laser diode, a 0.35 µm CMOS process with a high transistor breakdown
voltage of 3.6 V is selected for designing the laser diode driver circuit.
16
To have a maximum output power of 350 mW from the pump laser, the laser diode driver
circuit needs to inject 600 mA CW current to the pump laser, according to the L-I curve of the
pump laser. This required output current is quite high for the CMOS technology due to the metal
electromigration issue. For each CMOS process, the amount of current carried on a metal wire or
bus is limited. A metal wire or bus carrying too much current causes a change in the metal
dimensions, spots of higher resistance and eventually failure [31]. This is termed as the metal
electromigration effect. The current density limit for the 0.35 µm CMOS process is 1.6 mA/µm.
To avoid the electromigration effect, the output current is distributed to 12 identical metal wires
on chip. Each metal wire has a width of 33 µm. This setup guarantees that the driver circuit can
deliver 600 mA current without experiencing the electromigration effect. These 12 metal wires are
connected together as a bus wire on a PCB, which allows a much higher current density.
The proposed CMOS circuit was designed and simulated with CDS. In order to create a large
tuning range of the output current, the size of the transistor M2 is made four times larger than the
transistor M1. As increasing the channel length can reduce the body effect, transistors of the
reference current circuit are set to a fairly large 3 µm channel length. Detailed design parameters
of transistors are listed in the Table 2.1.
Table 2.1: Design parameters of transistors in the proposed CMOS circuit
Transistor Parameter Value Transistor Parameter Value
M1 L=3 µm, W=10 µm M5 L=0.5 µm, W=20 µm
M2 L=3 µm, W=40 µm M6 L=5 µm, W=20 µm
M3 L=3 µm, W=60 µm M7 L=0.5 µm, W=20 µm
M4 L=3 µm, W=60 µm M0 L=1 µm, W=100 µm *216
17
For the transistor M0, the total width is 21.6 mm. A multi-finger structure is implemented in
the design. The multi-finger structure allows multiple identical transistors with short width
connected in parallel to replace a single transistor with long width. This technique reduces the gate
resistance and the circuit’s physical size. The supply voltage for this driver circuit is 3.3 V and the
laser diode is modeled as a –2 V voltage source in simulations based on the datasheet.
The model of the output current in Equation (2.10) is based on the fact that all transistors
M1, M2, M3, M4 and M0 are in the saturation region. With the help of simulations in the CDS, the
relation between the resistance and transistors’ operation region is shown in Figure 2.6. Transistors
M1 and M4 can only operate in the saturation region. Reg0, Reg2 and Reg3 represent the operation
region of transistor M0, M2 and M3, respectively. In this simulation, Y = 1 means the transistor is
in the triode region and Y = 2 means the transistor is in the saturation region.
Figure 2.6: Simulation results of transistors’ operation regin versus the resistance
1000 1200 1400 1600 1800 20000-Cutoff
1-Triode
2-Saturation
3-Subthreshold
4-Breakdown
Rout (Ω)
Y (O
pera
tion
Reg
ions
)
Reg0Reg2Reg3
18
The plotted result shows that all transistors are operating in saturation region (Y=2) when
resistance is higher than 1100 Ω. Under this condition, the relation between the output current and
the resistance as illustrated in Equation (2.10) is valid. By substituting design parameters K=4,
KI=1080 and β1=7.1667·10-4, the obtained output current is
𝐼𝐼𝑂𝑂 = 1.88371∙105 𝑅𝑅𝑂𝑂𝑂𝑂𝑂𝑂2
. (2.11)
This expression of output current is obtained based on mathematically modelling behaviors
of transistors. To verify the mathematical model, a comparison between the modelling output and
the CDS simulation output is made in Figure 2.7. The comparison shows that the modelling current
output agrees with the simulation current output. In the range from 1300 Ω to 1400 Ω, the
modelling one has an accurate fit to the simulation one.
Figure 2.7: Comparison of modelling current output and simulation current output at the resistance of the off-chip resistor range from 1100 Ω to 1950 Ω.
Due to the size restriction of the fabrication process, the design has to be divided into two
identical CMOS dies. Each CMOS die provides half of the total current output, therefore each one
has six identical current outputs. The die micrograph is shown in Figure 2.8. As the complete driver
circuit is connecting two of these dies in parallel, two dies are placed into one package. A complete
driver with the package is shown as Figure 2.9. The package is mounted on a custom-designed
PCB, as shown in Figure 2.10 (a). Four identical off-chip potentiometers are also connected on the
PCB for tuning the output, as each potentiometer controls three current output ports. This driver
circuit with the host PCB can output the driving current to the laser diode with a mount PCB,
which is shown in Figure 2.10 (b).
When the CMOS driver circuit provides the driving current to the laser diode, the laser diode
outputs laser light. The output current from the CMOS driver circuit is measured by an ammeter
placed between the driver circuit and the laser diode. The power of the output light is detected by
an optical power meter (Newport 1916-R). The supply voltage VDD to the driver circuit is 3.3 V.
GND Io1 Io2 Io3 Io4 Io5 Io6
VDD
VDD
VDD VDDRout1 Rout2
W =
2 m
m
Figure 2.8: Die micrograph of the fabricated CW laser diode driver
20
Figure 2.9: Package of a complete driver circuit with two identical CMOS dies; Note that the designed driver circuit is only one small part of the whole CMOS die, there are unrelated circuits shared on the same die.
Figure 2.10: (a) Host PCB with a driver package and four identical potentiometers which controls three current output ports; (b)Laser diode package mounted on a PCB
21
2.5.2 Measurements of the Output Current
By tuning the four off-chip potentiometers with the same pace from 1100 Ω to 1950 Ω, a
wide range of the output current from 200 mA to 600 mA is realized. This result is compared to
the results from the simulation and the modelling. The comparison result is shown in Figure 2.11.
Results indicate that the measurement agrees to the modelling better than the simulation. From a
resistance of 1250 Ω to 1450 Ω, results from three different approaches have good agreement.
Figure 2.11: Comparison of the output current between the measurement result, simulation result and mathematical modelling result; Note that the resistance at x-axis represents the resistance of each one potentiometer on the PCB in the measurement.
Simulation results show that the repetition rate and the pulse FWHM of the driver circuit’s
output current pulses can be tuned by adjusting control voltages. In measurement, the control
voltages Vctr, Vvar, Vb and Vvar2 can be adjusted by adjusting the potentiometers on the PCB. The
repetition rate of optical output pulses is related with the control voltages Vctr and Vvar, while the
pulse width is related with control voltages Vb and Vvar2. Figure 3.14 shows measured results of
tuning the optical output pulses’ repetition rate by adjusting the control voltage Vvar and Vctr. When
the Vvar is adjusted from -2.5 V to -0.5 V, the output pulses’ repetition rate is increased from
5.6 MHz to 10.4 MHz, as shown in Figure 3.14 (a). This measured result agrees with the simulation
result. When the Vctr is adjusted from -1.5 V to 0.9 V, the repetition rate is increased from 5.6 MHz
to 12.6 MHz, as shown in Figure 3.14 (b). However, when the Vctr is above -0.9 V, the amplitude
of the optical pulses starts decreasing. The maximum repetition rate with a stable output
performance is 13.2 MHz, when the Vctr is -0.9 V and Vvar is -2.1 V. Thus, the tuning range of the
repetition rate is from 5.6 MHz to 13.2 MHz.
Figure 3.15 shows measured results of tuning the pulse width of optical output pulses by
adjusting control voltages Vb and Vvar2. The minimum pulse width occurs when Vvar2 equals 2.1 V
and Vb equals 2.5 V. The tuning range of the pulse width is not wide in measurement. When the
pulse width is increased to above 350 ps, ultrashort pulse trains appear next to the main pulse, and
the output characteristics become unclear.
42
Figure 3.14: Measured results of tuning the optical output pulses’ repetition rate (a) by adjusting control voltage Vvar and (b) by adjusting control voltage Vctr
Figure 3.15: Measured results of tuning the optical output pulses’ pulse width (a) by adjusting control voltage Vvar2 and (b) by adjusting control voltage Vb
-2.5 -2 -1.5 -1 -0.55.5
6
6.5
7
7.5
8
8.5
9
9.5
10
10.5
Control Voltage Vvar (V)
Rep
etitio
n R
ate
(MH
z)
-1.6 -1.4 -1.2 -1 -0.85
6
7
8
9
10
11
12
13
Control Voltage Vctr (V)R
epet
ition
Rat
e (M
Hz)
(a) (b)
-2.5 -2 -1.5180
200
220
240
260
280
300
320
Control Voltage Vvar2 (V)
Pul
se W
idth
FW
HM
(ps)
-2.6 -2.4 -2.2 -2 -1.8200
205
210
215
220
225
230
235
Control Voltage Vb (V)
Pul
se W
idth
FW
HM
(ps)
(a) (b)
43
3.5.4 Analysis
There are certain differences between measurement and simulation in this design. First of
all, the supply voltage (VSS_33) in the measurement setup is higher than in the simulation.
Reasons for this difference could be that the laser diode equivalent circuit model is not accurate
enough or circuit connections in the measurement suffer the power transmission/reflection loss.
Although the supply voltage (3.65 V) is higher than the thick oxide transistor’s breakdown voltage
(3.6 V), it does not cause breakdown in the transistor due to the laser diode load circuit that is
connected with the transistor in series. The laser diode load circuit acts like a voltage dividing
circuit and also limits the in-rush current to the transistor when being powered on.
Secondly, the measured laser pulse width is very close to the simulation current pulse width.
Theoretically, the measured laser pulse width should be smaller than the simulated current pulse
width due to the gain-switching mechanism. However, as the parasitic capacitance from the die
package, the PCB and the laser diode causes a delay on the pulse rise time and fall time, the
switching current pulse width gets larger than the simulated one. A combination of these two
effects makes the measured pulse width close to the simulation current pulse width.
44
3.6 Summary
This work presents a compact picosecond pulsed laser source by gain-switching a DFB laser
diode with a CMOS integrated circuit. The CMOS circuit has a 0.3 mm2 die area. A commercial
DFB laser diode is assembled with the CMOS driver circuit and measurements on the laser light
output have been conducted. A performance summary of this laser source is given in Table 3.1.
This compact picosecond pulsed laser source can be used in SCG as the seed laser source.
Table 3.1: Performance summary of the designed pulsed laser diode
For comparison, a reference SCG system using a commercially available nanosecond seed
laser module produced by Opeak (DFB131101) is tested for SCG. The setup of the reference SCG
system is the same as with Figure 4.5, except that the reference SCG system uses a seed laser
module to replace the DFB laser diode and the CMOS driver circuit. The seed laser module
contains a 1550.8 nm DFB laser diode and an electronic control circuit. The module is able to
output optical pulses with 10 ns pulse width (shortest available) and 10 kHz repetition rate under
a 5 V power supply. The temporal waveform characteristics are shown in Figure 4.9. The spectrum
characteristics of the output pulses are shown in Figure 4.10 (a). The average output power of this
seed laser module is 1.1 µW. The performance is summarized in Table 4.3.
Table 4.3: Performance summary of the seed laser module
Parameter Value Parameter Value
Pulse Width 10 ns Pulse Energy 110 pJ
Repetition Rate 10 kHz Duty Cycle 1:10000
Average Power 1.1 µW Center Wavelength 1550.8 nm
Peak Power 10.3 mW Size (L · W) 14 cm · 9 cm
58
Figure 4.9: Output optical pulses from the commercial seed laser module: the left figure shows the pulse waveform; the right figure shows a single pulse shape.
Figure 4.10: (a) Spectrum of optical pulses output from the seed laser module; (b) spectrum of the amplified pulses output from the monitor of EDFA
-100 0 100 200-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Time (µs)
Volta
ge (V
)
99.99 100 100.01 100.02 100.03-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Time (µs)
Volta
ge (V
)
1550 1550.5 1551 1551.5-70
-65
-60
-55
-50
-45
-40
-35
-30
Wavelength (nm)
Spec
tral in
tens
ity (d
Bm)
1520 1540 1560 1580-70
-65
-60
-55
-50
-45
-40
-35
-30
Wavelength (nm)
Spec
tral in
tens
ity (d
Bm)
59
Compared to the designed DFB laser diode with CMOS driver circuit, this seed laser module
has a lower repetition rate and much higher pulse energy. The designed DFB laser diode has the
advantage of much shorter pulse width over the commercial seed laser.
The seed laser pulses are amplified by the EDFA with the maximum pump power. The
spectrum of the monitor output from the EDFA is shown in Figure 4.10(b). As the average output
power of the monitor port is 5.9 µW, the pump pulses launched into the HNLF have an expected
average power of 59 mW, a pulse energy of 5.9 µJ, and a peak power of 554 W. Although these
pump pulses have almost the same average power as the designed DFB laser diode after
amplification, the peak power of this laser module is 9.7 times higher than the designed one. The
difference of the duty cycle between these two designs results in the peak power difference.
The spectrum at the HNLF output is measured by the OSA, shown as a red trace in
Figure 4.11. The average power of this supercontinuum is 51 mW, achieving 86.4% power
conversion efficiency. The spectrum bandwidth is 892 nm at –40 dB level, spanning from
1258 nm to 2150 nm. A 5 dB spectrum flatness covers almost the full bandwidth except the pump
wavelength region. The nonlinear mechanism leading to the SCG is similar with the picosecond
pulses pumping regime. The wavelength shifting is also initiated by the MI and enhanced by the
SRS and the FWM effects.
Compared to the reference SCG system, the SCG system using the designed DFB laser diode
has a similar supercontinuum spectrum but a high peak at the pump wavelength. The high peak
reflects that wavelength shifting is not efficient enough due to a low peak power of pump pulses.
60
Figure 4.11: SCG results from two different systems
4.6 Summary
A SCG system based on an HNLF pumped by amplified diode-laser pulses is demonstrated.
The designed DFB laser diode with the CMOS driver circuit is employed to provide diode-laser
pulses. The SCG achieves a flat supercontinuum spectrum with 806 nm bandwidth (-40 dB level)
at a low peak power and a 91.2% power conversion efficiency at 62 mW output power. For
comparison, the designed DFB laser diode is replaced by a commercially available seed laser
module in the SCG system. A comparison table between the designed SCG system and the
reference SCG system is shown in Table 4.4.
1200 1400 1600 1800 2000 2200 2400-60
-50
-40
-30
-20
-10
0
10
20
Wavelength (nm)
Spe
ctra
l int
ensi
ty (d
Bm
)
Supercontinuum with proposed DFB LaserSupercontinuum with seed laser module
51 mW 62 mW
61
This comparison shows that the designed DFB laser diode is a good candidate for the seed
laser in picosecond level SCG systems. Additionally, the designed DFB laser diode is attractive
due to reduced complexity, compact size and low cost compared to commercial seed laser modules.
Table 4.4: Comparison of the designed SCG system and the reference SCG system
Supercontinuum Parameter The Designed SCG System The Reference SCG System
Average Power 62 mW 51 mW
Spectrum Bandwidth (-40 dB) 806 nm 892 nm
Spectral Density 0.08 mW/nm 0.05 mW/nm
Pulse Duration ~200 ps ~ 10 ns
Repetition Rate 5.6 MHz 10 kHz
Pulse Energy 11.1 nJ 5.1 µJ
62
Chapter 5: Conclusions and Future Work
5.1 Conclusions
This thesis provides experimental evidence to support that CMOS laser diode drivers can be
used in supercontinuum generation (SCG) systems. SCG systems with CMOS laser diode drivers
have the advantage of compact size and low cost compared to the traditional fiber lasers and the
mode-locked lasers based SCG systems. In this thesis, two different design of CMOS laser diode
drivers are presented.
In Chapter 2, a high-current continuous-wave (CW) CMOS laser diode driver was designed
for driving pump lasers in SCG systems. Based on the design requirements, a circuit design
methodology was proposed. Circuit simulations were conducted in Cadence Design System (CDS)
to predict the driver’s performance. Experimental measurements on both the driver’s current
output and the pump laser’s optical output confirmed that the driver could provide up to 600 mA,
which corresponded to an optical output power up to 350 mW from a 975 nm laser diode. The
driver features tunable output, long-term stable operation and reduced sensitivity to power supply
voltage variations.
In Chapter 3, a picosecond pulsed laser diode driver was designed for gain-switching a
distributed feedback (DFB) laser diode. The driver design implemented CMOS analog logic
circuits to generate picosecond level current pulses. Simulations showed that the driver circuit can
provide current pulses with a pulse width of 200 ps, a repetition rate of 5.8 MHz, and a peak current
of 80 mA. Experimental measurements of the DFB laser diode output showed that the output
optical pulses had a pulse width of 200 ps at FWHM, a peak power of 5.9 mW and a repetition
rate of 5.6 MHz. The average output power of the pulsed DFB laser diode is 7.0 µW and the pulse
energy is 1.25 pJ. The driver design also features a tunable repetition rate output and a tunable
63
pulse width output. The size of the designed package, including the laser and the driver together,
is 7 cm length and 4 cm width. This compact picosecond pulsed laser diode is a suitable seed laser
for SCG systems.
Chapter 4 demonstrated an HNLF based SCG system with the designed picosecond DFB
laser diode. The supercontinuum was generated by using amplified picosecond optical pulses to
induce nonlinear effects in the HNLF. The system consists of the designed picosecond DFB laser
diode, a commercial EDFA module and the HNLF. The supercontinuum output of the designed
system has an average power of 62 mW, a spectral bandwidth of 806 nm, a pulse width of 210 ps
and a repetition rate of 5.6 MHz. For comparison, a reference SCG system using a nanosecond
pulsed commercial seed laser module was tested for supercontinuum generation. With the
commercial seed laser module, the supercontinuum output has a spectral bandwidth of 892 nm and
an average output power of 51 mW. Compared to this commercial module, the designed DFB laser
diode has the advantages of compacter size, lower cost, higher output power and shorter pulse
width. The designed SCG system has potential applications in OCT, spectroscopy and DWDM.
5.2 Future Work
Future work will be focusing on improving performance of the CMOS driver circuits and
developing a compact package for the SCG system.
In the design of the CW CMOS laser diode driver circuit, the CMOS chip size, the
potentiometer setup and the PCB assembly can be further optimized. The CMOS chip size can be
reduced by fabricating a new chip with all circuit parts integrated together. The number of
potentiometers on the PCB board can be reduced to one if the CMOS circuit is fully integrated.
The driver circuit and the laser diode can be assembled into one host PCB to make them a more
compact optoelectronic device.
64
The output of the designed picosecond pulsed DFB laser diode is sensitive to the noise from
the power supply connectors and the surroundings. A new PCB with an improved circuit grounding
and noise filtering setup can be designed to improve the circuit stability performance.
The designed SCG system uses a commercial EDFA module, which meets the requirement
for the purpose of pulse amplification. However, the module is not cost-effective and is difficult
to be integrated with other components in the system. It is necessary to develop a custom-made
EDFA with the designed pump laser. A custom-made EDFA will reduce the cost and size and also
increase the system flexibility.
The designed SCG system has not been fully packaged as a device. The largest discrete
component in this system is the 10-meter-long HNLF, which has a circular shape with an 8 cm
diameter. The device package will be built around the HNLF. The expected full package dimension
will be 10 cm length, 10 cm width and 5 cm height with a power supply input port and a
supercontinuum output port.
Silicon photonics technology has been an intense research interest recently, as it allows
optical devices to be made using standard semiconductor fabrication. This technology provides a
new platform for integration between the optical devices and CMOS circuits. It has already been
approved that the CMOS-compatible silicon wire waveguides can be used as a nonlinear medium
for SCG [18]. This report shows the opportunity for designed CMOS laser diode driver circuits.
In the future, these CMOS circuits can be integrated with silicon waveguides on the same platform
as lab-on-a-chip devices. This will be a revolutionary improvement on the design of compact and
low-cost SCG systems.
65
References
[1] R. R. Alfano and S. L. Shapiro, “Emission in the region 4000 to 7000 via four-photon
coupling in glass,” Phys. Rev. Lett., vol. 24, no. 11, pp. 584–587, 1970.
[2] R. R. Alfano, The supercontinuum laser source (Second Edition): Fundamentals with
updated references. Springer, 2006.
[3] P. B. Corkum, C. Rolland, and T. Srinivasan-Rao, “Supercontinuum generation in gases,”
Phys. Rev. Lett., vol. 57, no. 18, pp. 2268–2271, 1986.
[4] C. Lin and R. H. Stolen, “New nanosecond continuum for excited-state spectroscopy,” Appl.
Phys. Lett., vol. 28, no. 4, pp. 216–218, 1976.
[5] R. H. Stolen, C. Lee, and R. K. Jain, “Development of the stimulated Raman spectrum in
single-mode silica fibers,” J. Opt. Soc. Am. B, vol. 1, no. 4, p. 652, 1984.
[6] I. Ilev, H. Kumagai, K. Toyoda, and I. Koprinkov, “Highly efficient wideband continuum
generation in a single-mode optical fiber by powerful broadband laser pumping.,” Appl.
Opt., vol. 35, no. 15, pp. 2548–53, 1996.
[7] J. M. Dudley and J. R. Taylor, Supercontinuum generation in optical fibers. Cambridge
University Press, 2010.
[8] J. K. Ranka, R. S. Windeler, and a J. Stentz, “Visible continuum generation in air-silica
microstructure optical fibers with anomalous dispersion at 800 nm.,” Opt. Lett., vol. 25, no.
1, pp. 25–27, 2000.
[9] T. a Birks, W. J. Wadsworth, and P. S. Russell, “Supercontinuum generation in tapered
fibers.,” Opt. Lett., vol. 25, no. 19, pp. 1415–1417, 2000.
[10] N. Nishizawa and T. Goto, “Widely broadened super continuum generation using highly
nonlinear dispersion shifted fibers and femtosecond fiber laser,” Japanese J. Appl. Physics,
66
Part 2 Lett., vol. 40, no. 4 B, 2001.
[11] H. Takara, T. Ohara, K. Mori, K. Sato, E. Yamada, Y. Inoue, T. Shibata, M. Abe, T.
Morioka, and K.-I. Sato, “More than 1000 channel optical frequency chain generation from
single supercontinuum source with 12.5 GHz channel spacing,” Electron. Lett., vol. 36, no.
25, p. 2089, 2000.
[12] T. Udem, R. Holzwarth, and T. W. Hänsch, “Optical frequency metrology,” Nature, vol.
416, no. 6877, pp. 233–237, 2002.
[13] S. Diddams, D. Jones, J. Ye, S. Cundiff, J. Hall, J. Ranka, R. Windeler, R. Holzwarth, T.
Udem, and T. Hänsch, “Direct Link between Microwave and Optical Frequencies with a