10-40 Gb/s同軸式高速雷射模組構裝之研究

全文

(1)國立中山大學光電工程研究所博士論文. 10-40 Gb/s 同軸式高速雷射模組構裝之研究. The Study of 10-40 Gb/s High-Speed Laser Module Based on Coaxial-Type Packages. 研究生：林旻進. 撰. 指導教授：鄭木海. 博士. 共同指導教授：施天從. 博士. 中華民國九十七年一月二十九日.

(2)

(3)

(4)

(5) Acknowledgment The work presented in this dissertation could not have happened without the help and support of many people. My advisor, Dr. Wood-Hi Cheng, always reminded and guided the direction of my research as I was lost and also taught me how to organize and schedule my work. My co-advisor, Dr. Tien-Tsorng Shih, always patiently discussed and provided his professional knowledge and experience with me. Thanks for all my advisors provided the best environment to do these works. I would like to thank Dr. San-Liang Lee, Dr. Tsong-Sheng Lay, Dr. Ching-Ting Lee, Dr. Tzyy-Sheng Horng, Dr. Gong-Ru Lin, Dr. Hai-Han Lu, Dr. Kai-Ming Feng, given many invaluable comments to me in the oral examination. I want to thanks to the people gave me the support of the materials and resources for my work. Infomax optical technology corporation, ITRI OES and Dr. M. H. Mao gave valuable help and discussion for the work of the 10-Gb/s coaxial laser module. Dr. Chang-You Li and Tuan-Yu Hung from APAC Opto Electronics Inc. gave great help for the work of the 10-Gb/s BOSA module. Dr. Chieh Hu gave me valuable guidance and suggestions for the work of the 40-Gb/s optical module. I am also appreciated that Dr. Tzyy-Sheng Horng and Dr. Yi-Jen Chiu, gave me technical and measurement support without sparing for my research. Thanks for all my teammate, Pei-Hao Tseng, Chih-Ching Cheng, Kuo-Chu Liao, Kuei-Ming Chu, Tai-Fu Tseng, Jiun-Ming Chen, Jia-Neng Yang, worked hard with me for my PhD degree. Without their outstanding work, I would suffer from more difficult problems in my way. Thanks for all the stuffs in the electro-optics package lab, Chia-Ming Chang, Shin-En Chang, Pei-Ling Cheng, who were study, live together with me, and sharing their profession knowledge and experiences of life and work. Finally, I am deeply grateful for my family and my girl friend, Jennifer. Because their fully supports, I can devote myself to the study. They always take care of me all the time. They are my strongest supporters. I dedicated this dissertation to my family and my girl friend..

(6) 中文摘要：此論文係研究兼顧低成本與高性能雷射模組之構裝，研發成果可應用於高速光纖通訊、光纖到家(Fiber-To-The-Home)及被動光學網路(Passive Optical Network)。本論文已實現了「10 兆位元之同軸式雷射模組」，「10 兆位元雙向光學次模組」及「具有四個 10 兆位元通道之疏式分波多工(CWDM)雷射模組」之研究。傳統 TO-Can 基底由於結構上沒有適當設計，具有較差 RF 傳輸特性，因此訊號接腳長度與打線長度過長產生凹陷之濾波效應，此成為傳統 TO-Can 主要克服之問題。本研究首先提出同軸式雷射模組採用商業化之 TO-Can 材料，並提出具有一個內部匹配電阻 18Ω 來降低高速訊號的反射現象。研究模擬與實驗量測之小訊號結果得到不錯之一致性，並實現了 10 兆位元同軸式雷式模組可以達到 OC-192 規範之眼圖餘裕 31%。基於成本考量，研究同時提出 10 兆位元雙向光次模組(BOSA)結構採用商業化較低速 155-Mb/s 或是 1.25-Gb/s 雙向光次模組之設計，研究結果在雙向光次模組發射與接受端皆測得清晰之眼圖結果。在 10 公里單模光纖傳輸下具有 0.5 dB 之功率代價(power penalty)，與 0.9dB 之串音代價(crosstalk penalty)。本研究已經成功驗証高性能與低成本 10 兆位元雙向光次模組之雙向傳輸架構，其性能可應用於未來高速光纖到家或是被動式光網路之可行性。最後，本研究探討 4 個 10 兆位元通道之雷射模組採用現行低成本 TO-Can 雷射與 CWDM 技術，並提供一個可應用於 40-Gb/s 光纖通訊網路之解決方案。此光學模組每個通道具有 10 兆位元之操作速率，在位元誤碼率(BER)為 10-9 情況下可以傳輸距離達到 30 公里；平均系統光功率損耗約為 12dB。本研究中所提出之高性能之 40 兆位元 CWDM 模組具有低成本之架構，並可應用於 WDM-PON 之光纖到家系統中。.

(7) Abstract The goal of this dissertation is to provide a solution by using a low-cost and high-performance laser module package for the applications of high-speed optical communication, fiber-to-the-home (FTTH), and passive optical network (PON). A 10-Gb/s coaxial-type laser module, a 10-Gb/s bi-directional optical sub-assembly (BOSA) module, and a 4 channels x 10-Gb/s coarse wavelength division multiplexing (CWDM) laser module have been implemented for this study. The conventional TO-Can header suffers poor RF transmission characteristics without proper modification. The notch filter effect induced by the parasitic inductance of the long lead and wires is one of its major factors. The proposed coaxial laser module is fitted with a commercial TO-Can with an internal matching resistor of 18Ω to reduce the signal reflection. The comparison of small signal results between the theoretical and the experimental results shows good agreement. The proposed 10-Gb/s coaxial laser module implemented can achieve 31% mask margins with the OC-192 standard. For cost consideration, the structure of the proposed 10-Gb/s BOSA modules is adapted to the idea of the commercial low bit rate of 155-Mb/s or 1.25-Gb/s BOSA modules. The proposed BOSA modules show a clear opening eye diagrams at both their transmitter and receiver side. The power penalty with a 10-km SMF transmission is 0.5dB and the crosstalk penalty is 0.9dB. According to the experimental results, we have demonstrated successfully the high-performance and the low-cost of 10-Gb/s BOSA modules and verified the feasibility of the bi-directional architecture for use in the future’s high-speed FTTH or PON network applications. The 4 channel x 10-Gb/s laser modules adapted the existing low-cost TO-Can laser and the CWDM techniques provide one of the solutions for the 40-Gb/s optical communication application. The proposed optical module operating at 10-Gb/s per channel can exceed a rate.

(8) of over 30 km transmission at the bit-error-rate (BER) of 10-9, with an average system power penalty of 12 dB. The proposed high-performance 40-Gb/s CWDM module shows the low-cost possibility that ensures the application of WDM-passive optical network (WDM-PON) fiber-to-the-home (FTTH) systems..

(9) Contents Abstract Acknowledgement Contents…………………………………………………………….. I List of Figures……………………………………………………… III List of Tables……………………………………………………….. VII Chapter 1. Introduction 1.1 Background……………………………………………...…... 1.2. Motivation (Evolution of Laser Module Package)………………….. 1.2.1 DIP or Mini-DIL Package……………………………………. 1.2.2 Butterfly Package…………………………………………….. 1.2.3 TO-Can Package……………………………………………... 1.2.4 Other Modified Laser Module Package for 10 Gb/s Applications………………………………………………… 1.3 Different Solutions of the Coaxial-type Laser Modules for High-Speed…………………………………………………………... 1.4 Overview of Dissertation……………………………………………. 1.5 References……………………………………………………………. Chapter 2. 10-Gb/s Coaxial Laser Module Design and Simulation. 2.1 2.2 2.3 2.4 2.5. Equivalent Circuit Model of the High-speed Un-cooled DFB-LD….. Simulation Results of the High-speed Un-cooled DFB-LD………… Circuit Modeling of the High-speed Laser Module…………………. Simulation Results of the Coaxial Laser Module…………………… Analysis and Optimization of Present 10-Gb/s Laser Module Package……………………………………………… 2.6 References...………………………………………………………….. Chapter 3 3.1. 1 1 1 2 3 3 5 6 9 12. 15 15 19 21 29 33 41. 10-Gb/s Coaxial Laser Module Package and Results. 43. Laser module package processes……………………………………. 3.1.1 Material prepared…………………………………………….. 3.1.2 Die-Bonder System…………………………………………... 3.1.3 Wire-Bonder System………………………………………… 3.1.4 Laser Welding System………………………………………... 43 43 45 49 53. I.

(10) 3.2 Fabrication of the Coaxial Laser Module…………………………… 3.4 Measurement Results of the Coaxial Laser Module………………… 3.5 References……………………………………………………………. Chapter 4. 10-Gb/s BOSA Module Package. 4.1 Introduction of 10-Gb/s BOSA Module……………………………... 4.2 Fabrication of 10-Gb/s TO-56 Packaged Distributed Feedback (DFB) Laser Diode………………………………………………….. 4.3 Fabrication of 10-Gb/s TO-CAN Packaged PIN-TIA Device………. 4.4 Fabrication Processes of BOSA Module…………………………….. 4.5 Measurement and Results ………………………...………………… 4.5.1 Transmitter of the BOSA module……………………………… 4.5.2 Receiver of the BOSA module………………………………… 4.6 Summary…………………………………………………………….. 4.7 References……………………………………………………………. Chapter 5. 56 58 63. 64 64 65 67 68 70 70 75 78 80. 40Gb/s Laser Module based on 10-Gb/s Per Channel. 82. 5.1 Introduction………………………………………………………….. 5.2 Fabrication of the DFB laser with collimator output………………... 5.3 Structure and Fabrication of the proposed 40-Gb/s CWDM optical module………………………………………………………………. 5.4 Measurement Results and Discussion………………………………... 82 83. 5.4.1 Performance of the DFB lasers………………………………... 5.4.2 CW property of the proposed 40-Gb/s CWDM optical module. 5.4.3 Eye Diagrams and BER performance of the 40-Gb/s CWDM module………………………………………………………… 5.5 Summary…………………………………………………………….. 5.6 References..………………………………………………………….. Chapter 6. Conclusions. 85 87 87 89 91 96 97. 98. 6.1 Conclusion.………………………………………………………….. 6.2 Discussion………………………………………….………………... 6.2.1 40-Gb/s CWDM-PON network……...………………………... 6.2.2 10-Gb/s MMF Transmission………………………………….... Vita. 98 99 99 101. 102. II.

(11) List of Figures. Chapter 1 Introduction 1.1 Different laser modules package types of (a) DIP and (b) mini-DIL packages used for optoelectronic laser modules………………………. 2 1.2 Two types of the 14 pins butterfly package, (a) type A：glass-to-metal type and (b) type B：ceramic-to-metal type……………………………. 3 1.3 The coaxial type of laser module package used for telecommunication, (a) TO-CAN and (d) TOSA module…………………………………… 4 1.4 Different types optoelectronic package of 10 Gb/s application. (a) Mini-tailed package, (b) differential signal line BTF-PKG, (c) 10 Gb/s Mini-Flat, and (d) Surface mount ceramic (MCM) package………….. 6 1.5 The conventional TO-Can laser module with silicon optical bench, (a) with V-groove and (b) with RF matching resistor and a spiral inductor. 8 1.6 (a) The side view and (b) top view of the conventional optical module with compensation element……………………………………………. 8 1.7 (a) Top view and (b) bottom view of the ceramic TO header assembly.. 9. Chapter 2 10-Gb/s Coaxial Laser Module Design and Simulation 2.1 Cross section of a ridge laser…………………………………………... 2.2 Distribution of photon density and electron density across the active layer…………………………………………………………………….. 2.3 Small-signal circuit model of the device chip………………………….. 2.4 Small-signal circuit model of the device chip………………………….. 2.5 The small signal frequency response of DFB laser diode at various bias currents. …………………………………………………………... 2.6 Different types of optoelectronics package, (a) TO-5, (b) TO-18, (c) TO-46, and (d) TO-56………………………………………………….. 2.7 Structure of TO-56 header, (a) top view and (b) cross section………… 2.8 Schematic and equivalent circuits of the conductive leads…………….. 2.9 Calculated the inductance of the lead with different length…………… 2.10 Schematic and equivalent circuits of coaxial leads…………………….. 2.11 Calculated the L-C resonance peak frequency…………………………. 2.12 A schematic diagram of the proposed module…………………………. 2.13 The equivalent circuit model of the proposed module…………………. 2.14 RF reflection characteristics of laser modules…………………………. III. 15 17 19 20 21 22 23 23 24 25 26 27 27 30.

(12) 2.15 RF transmission characteristics of laser modules with different bonded wire inductances………………………………………………………... 2.16 Small signal 3dB bandwidth of laser modules with different bonded wire inductances………………………………………………………... 2.17 RF reflection characteristics of laser modules with different lead lengths………………………………………………………………….. 2.18 RF transmission characteristics of laser modules with different lead lengths………………………………………………………………….. 2.19 Small signal 3dB bandwidth of laser modules with different lead lengths………………………………………………………………….. 2.20 The schematic drawing of the TO-56 header…………………………... 2.21 The -3dB bandwidth of the TO-56 header at different dimensions of lead length……………………………………………………………… 2.22 The -3dB bandwidth of the TO-56 header at different dimensions of lead length……………………………………………………………… 2.23 The -3dB bandwidth of the TO-56 header at different diameter of feed-through diameter…………………………………………………. 2.24 The -3dB bandwidth of the TO-56 header at different diameter of feed-through diameter…………………………………………………. 2.25 The small-signal response of the optimized TO-56 header…………….. 30 31 32 32 33 34 35 36 37 38 39. Chapter 3 10-Gb/s Coaxial Laser Module Measurement and Results 3.1 The schematic diagram of the chip resistor……………………………. 3.2 The schematic diagram of the chip resistor……………………………. 3.3 The schematic diagram of the LD submount at (a) the view of top-side and (b) back-side……………………………………………………….. 3.4 The schematic diagram of the ball lens cap……………………………. 3.5 Photo of the die-bonder system………………………………………… 3.6 Photo of (a) the viewer system and (b) the dual bond head……………. 3.7 Photo of the (a) die stage, and (b) the work stage……………………… 3.8 Photo of the TO-56 header with (a) LD submount and laser diode, (b) R submount and resistor………………………………………………... 3.9 Photo of the wire bonder system……………………………………….. 3.10 Bonding cycle of the wedge wire bonder……………………………… 3.11 Photo of the TO-56 laser module on the work holder…………………. 3.12 Photo of the (a) TO-56 header, (b) TO-Can and (d) TOSA laser module………………………………………………………………….. 3.13 Schematic diagram of the lower part…………………………………... IV. 43 44 44 45 46 46 47 49 50 52 53 54 55.

(13) 3.14 3.15 3.16 3.17 3.18 3.19. Photo of the laser welding system……………………………………... A SEM picture of TO-Can laser diode and of real module…………….. A photo of real 10Gb/s laser module…………………………………... The L-I curves of laser module at different temperatures……………… The small-signal measurement setup. The measured RF reflection characteristics of the 10Gbps DFB laser module at 25oC…………………………………………………………. 3.20 The measured RF transmission characteristics of the 10Gbps DFB laser module at 25oC…………………………………………………… 3.21 The back-to-back filtered eye diagram of 10Gbps laser module with 10Gbps Ethernet mask at (a) 25oC and (b) 85oC……………………….. 55 57 58 59 60 61 61 62. Chapter 4 10-Gb/s BOSA Module Package 4.1 A schematic diagram of a 10-Gb/s DFB laser diode packaged with a TO-56 header…………………………………………………………... 4.2 A picture of 10-Gb/s PIN-TIA packaged on TO-46 header……………. 4.3 A schematic diagram of a 10-Gb/s BOSA module…………………….. 4.4 A fabrication flow chart of BOSA module…………………………….. 4.5 L-I and I-V characteristics of the Tx at room temperature…………….. 4.6 Small-signal measurement setup of the BOSA module at Tx,…………. 4.7 The frequency response (S21) of the BOSA module Tx side at different 4.8 4.9 4.10 4.11 4.12 4.13. bias currents……………………………………………………………. The reflected (S11) coefficient of the BOSA module at Tx…………….. Transmitted eye diagrams of Tx for the (a) back-to-back connection, and (b) after a 10-km SMF transmission………………………………. BER testing results of the Tx for back-to-back connection and a 10-km fiber transmission………………………………………………………. Receiving eye diagrams of Rx for the (a) back-to-back connection, and (b) after a 10-km SMF transmission…………………………………… BER testing results of the Rx for the back-to-back connection and a 10-km fiber transmission………………………………………………. BER testing results of the Rx when the Tx is turned on and off……….. 66 67 68 68 70 71 72 72 73 74 76 77 78. Chapter 5 40Gb/s Laser Module based on 10-Gb/s Per Channel 5.1 5.2 5.3 5.4. A photo of the DFB laser module with a collimated light output……… Coupling efficiency of the optical collimator as a function of position.. A photo of the 40-Gb/s CWDM module……………………………….. A fabrication flow chart of the 40-Gb/s CWDM optical module……… V. 84 84 85 87.

(14) 5.5 The frequency response characteristics of the DFB lasers…………….. 5.6 The eye diagram performance of the WDM TOSA module (a) ch 1-1275 nm, (b) ch 2-1350 nm, (c) ch 3-1325 nm, (b) and (d) ch 4-1300 nm………………………………………………. 5.7 L-I and I-V characteristics of the proposed 40-Gb/s module………...... 5.8 An optical emission spectrum of the 40-Gb/s module, (a) all the channels were turned on, and (b) only ch 4-1300 nm was turned on….. 5.9 Eye diagrams and BER measurement setup of the proposed 40-Gb/s optical module………………………………………………………….. 5.10 The eye diagram of the 40-Gb/s WDM module in case of BTB connection, (a)ch 1-1275 nm, (b)ch 2-1350 nm, (c) ch 3-1325 nm, and (d) ch 4-1300 nm………………………………………………..……… 5.11 The eye diagram of the 40-Gb/s WDM module in case of 10 km SMF, (a) ch 1-1275 nm, (b) ch 2-1350 nm, (c) ch 3-1325 nm, and (d) ch 4-1300 nm…………..………………………………………. 5.12 The eye diagrams of the 40-Gb/s modulen in case of 10 km SMF received by an optical receiver, (a)ch 1-1275 nm, (b)ch 2-1350 nm, (c)ch 3-1325 nm, and (d) ch 4-1300 nm……………………………….. 5.13 The BER performance of the 40-Gb/s WDM module, (a) ch 1-1275 nm, (b ) ch 2-1350 nm, (c) ch 3-1325 nm and (d) ch 4-1300 nm……... 5.14 The BER performance of the 40-Gb/s WDM module, (a) ch 1-1275 nm, (b)ch 2-1350 nm, (c)ch 3-1325 nm and (d) ch 4-1300 nm……….... 88. 88 89 90 92. 92. 93. 94 95 96. Chapter 6 Conclusions 6.1 The proposed 40-Gb/s CWDM-PON network………………………… 100 6.2 Measurement setup of the MMF transmission…………………………. 101. VI.

(15) List of Tables. Chapter 1 Introduction 1.1 Comparison different types of commercial laser module packages…. 5. Chapter 2 2.1 2.2 2.3. 10-Gb/s Coaxial Laser Module Design and Simulation Elements of the equivalent circuit in the active region……………... Elements of the equivalent circuit of a TO-Can laser………………. Dimensions of the TO-56 header……………………………………. 20 29 39. Chapter 3 3.1 3.2 3.3. 10-Gb/s Coaxial Laser Module Measurement and Results The parameters of the dual bond heads……………………………... 47 The temperature setting of the working stage………………………. 48 The parameters setting of the wedge bonder system………………... 51. Chapter 5 40Gb/s Laser Module based on 10-Gb/s Per Channel 5.1 The CW performance of the 40-Gb/s WDM module……………….. 5.2 The transmission performance of the 40-Gb/s WDM module………. VII. 91 94.

(16) Chapter 1 Introduction 1.1 Background Recently, the fiber-to-the-home (FTTH) network has become a key focus in the fast developing fiber optical communication industry. Several applications promote the increasing demands of the network’s bandwidth, such as the Internet, person-to-person communication, entertainment, broadcast, and E-learning [1.1]. The bandwidth of a single subscriber line in FTTH has gradually migrated from 155Mbps to 1.25Gbps [1.2]-[1.8]. G-PON and GE-PON networks now attract a lot of attention from most service carriers [1.9]-[1.15]. Due to the enormous demand of bandwidth from subscribers, 10Gbps solution is going to play an important role in the next generation FTTH network [1.16]. Besides, more and more high-speed commercial electronic components were developed, such as microprocessors and memory. These powerful devices need larger bandwidth to communicate with each other, and this demand stimulates that the local area network (LAN) should evolve to a larger transmission capacity [1.17]-[1.22]. At present, the high cost of current 10Gbps and higher bit rate module is one of the major factors preventing future FTTH and LAN applications. Therefore, lowering the cost of the expensive 10Gbps and higher bit rate laser module with a satisfactory performance is essential to the fiber optical communication industry.. 1.2 Motivation (Evolution of Laser Module Package) High-performance and low-cost were the development trends of the optoelectronic modules. Package of the optoelectronic device takes a large proportion of the fabrication cost. For consideration of the high-speed application, the optoelectronic devices inside the package may be affected with each other if the structure of the package was not designed properly. For consideration of the optical requirements, a high accuracy optical alignment and high reliability have to be achieved. Sealed package is commonly used to contain, protect, and 1.

(17) electrically connect the optoelectronics components. All of these considerations would increase the cost of the optoelectronic module. Several of novel packages have been developed for high-performance 10-Gb/s laser modules, such as butterfly package, mini dual-in-line (DIL) package or called DIP, standardand modified-coaxial package.. 1.2.1 DIP or Mini-DIL Package A DIP and mini-DIL package commonly used in electronics package was a rectangular housing and two parallel rows of electronics connecting pins as shown in Fig. 1.1. It can be mounted on a printed circuit board (PCB) either directly using through-hole technology, or using inexpensive sockets to easily replace the devices and to reduce the risk of the overheat damage during soldering. The disadvantage of the DIL package is poor high frequency performance due to the structure of long transmission path as transmitting high speed signals. The DIL package type of the laser module for telecom and datacom applications was usually operated below the data rate of 2Gbps.. Newport Ltd. RMT Ltd. (a) Fig. 1.1: Different laser modules package types of (a) DIP and (b) mini-DIL packages used for optoelectronic laser modules.. 2.

(18) 1.2.2 Butterfly Package Butterfly package developed for optoelectronics devices was a rectangular housing with two parallel rows of connecting pins as shown in Fig. 1.2 (a) and (b). These two types of the laser modules were 14 pins butterfly package with glass-to-metal style and ceramic-to-metal style, respectively. Figure 1.2 (a) was the glass-to-metal type butterfly package and was used for high power or pump laser modules. Figure 1.2 (b) was the ceramic-to-metal type butterfly package and was used for high-performance and high-speed laser module application. Because the inner lead inside the housing was a specifically designed for high frequency application, the high frequency characteristic of this type butterfly package could reach to 20~40GHz. The signal assignments of the electrical leads were ground-signal-ground (G-S-G). The merit of the butterfly package is large space inside the housing for placing a custom heat sink material, a thermo-electric cooler (TEC), a thermistor, a ball lens, a glass or sapphire window and an isolator. Although these components could stabilize the laser module’s performance, the package processes would be more complicated and increase the packaging cost.. RMT Ltd. Schott AG. (b). (a). Fig. 1.2: Two types of the 14 pins butterfly package, (a) type A：glass-to-metal type and (b) type B：ceramic-to-metal type.. 3.

(19) 1.2.3 TO-Can Package Transistor outline (TO)-Can package type laser modules were widely used in different field, such as optical storage, lighting, and optical communication. Figure 1.3 (a) and (b) show the photograph of a standard TO-Can package laser module and TOSA module, respectively. The standard TO-Can package laser module with four electrical connection pins was consisted of a laser diode, a monitor diode and a ball lens cap, as shown in Fig. 1.3 (a). Sometimes an isolator would be packaged inside to prevent the light from reflecting into the laser diode. The standard TO-Can laser module was tightly jointed with a metal lower part and welding with a metal upper part by laser welding technology, as shown in Fig. 1.3 (b). The advantages of this type are low-cost, easy fabrication, small volume, and low connection pins. Because few components were necessary for packaging, this could decrease the package cost and simplify the fabrication processes.. Upper part. Ball lens cap. Lower part. (a). (b). Fig. 1.3: The coaxial type of laser module package, (a) TO-CAN and (b) TOSA module.. The comparisons of the different types commercial laser module packages are listed in Table I. Although the package types of butterfly and Mini-DIL could provide a stable and high performance light source output, the disadvantages of the butterfly and Mini-DIL packages were expensive fabrication cost than the TO-Can type. For the Mini-DIL and 4.

(20) butterfly package, they have larger volume than TO-Can package to place the TEC, thermistor, saddle and others components. But, these additional components would cause the complex package processes that were difficult to fully automatic fabrication. In contrast, the transmission speed of the coaxial type TO-Can package was the important issue and needed to be solved urgently. Until to now, several researchers were focused on the high-speed coaxial type TO-Can package [1.23]-[1.29]. This will be described in the next section.. Table 1.1: Comparison different types of commercial laser module packages. Package Type. Mini-DIL. Butterfly. TO-Can. Data Rate. 155Mbps~2.5Gbps. >10Gbps. 155Mbps~10Gbps. LD to Fiber Alignment. Passive. Active. Active/Passive. 0.25~2. 0.25. 1~2. Components Assembly. Automated flip-chip. Largely manual. Fully automated. Material Cost. Medium. High. Low. Package Cost. Low. High. Lowest. Dimension. Medium. Large. Smallest. PIN output. 6~12. 8~14. 2~4. High frequency characteristic. medium. good. poor. Alignment Tolerance (µm). 1.2.4 Other Modified Laser Module Package for 10 Gb/s Applications There are several novel optoelectronic package types for 10 Gb/s or more high speed applications developed by Kyocera Inc, as shown in Fig. 1.4. Figure 1.4 (a) and (b) were the mini-tailed type and differential type package with large space for placing optoelectronic devices. Fig. 1.4 (c) and (d) were mini-flat type and surface mount ceramic type package with. 5.

(21) a specific transmission design inside the housing and flat electrical leads output for attaching onto the PCB. These novel package types were properly impedance controlled at 25 or 50 ohms to reduce the reflection signal. The inner electric signal leads inside the hosing case were designed on the ceramic material base. This design could reduce the transmission loss at high frequency operation.. (b). (a). Kyocera Inc.. Kyocera Inc. (c). (d). Kyocera Inc.. Kyocera Inc.. Fig.1.4: Different types optoelectronic package of 10 Gb/s application. (a) Mini-tailed package, (b) differential signal line BTF-PKG, (c) 10 Gb/s Mini-Flat, and (d) Surface mount ceramic (MCM) package.. However, all of these packages have complicated design, customized components, and specialized fabrication process that make their cost too high. It will be highly beneficial to develop a low cost module with existing cheap materials and process equipments. In order to reducing the package cost and improving the performance, we would like to analysis the commercial low-cost TO-Can material. 6.

(22) 1.3 Different Solutions of the Coaxial-type Laser Modules for High-Speed The advantages of the coaxial-type laser modules were compact, low I/O pins output, and easy for automatic fabrication. It is well suitable for the receptacle-type module. However, an increasing parasitic inductance with signal lead length would severely degrade the performance of the laser module as the transmission rate up to 10-Gb/s. Several researchers are focused on the laser module based on the TO-structure for high-speed applications. They try to increase transmission bandwidth and decrease the RF return loss of the optical communication module. The optical module comprising of a commercial TO-56 header and a silicon optical bench (SiOB) sub-mount connecting with the electric signal leads as shown in Fig. 1.5. Figure 1.5 (a) shows a conventional TO-Can package with a laser diode was bonded on the SiOB with a V-groove design. A photo-diode convert the receiving light into a photo current was bonded on the base of the TO-56 header. The sub-mount was formed by materials of AlN or SiC. The TO-56 header was formed by the CuW, KOVAR, and iron. However, the optical module was difficultly used for high-speed transmission over 10-Gb/s due to the use of the bonding wire from the leads to the sub-mount and the leads to the anode of the laser diode. Further, the dimension of the SiOB sub-mount was large in size and the signal propagation distance from the laser diode to lead was too long. Figure 1.5 (a) and (b) were the similar structure. The characteristic impedance of the transmission line and ground plane designed on the SiOB was 25 or 50 ohm, respectively. A RF matching resistor was disposed on the transmission line. The transmission line was connected with the anode of the laser diode through a bonding wire. On the other end of the transmission line, it was electrically connected with the electrical lead by solder material. The ground plane was directly connected with cathode of the laser diode by die-bonding technique and electrically connected with the TO-56 header. A spiral inductor was also connected with the anode of the laser diode by the boding wire. From the consideration of 7.

(23) heat generation, it was caused from the DC current flow through the matching resistor and degraded the performance of the laser module at high DC current operation. Then, the spiral inductor could separate the DC current and RF signal from different electrical lead, such as the RF lead and the DC lead. The RF matching resistor and spiral inductor could be monolithically integrated on the SiOB by thin-film processes.. Laser diode RF matching resistor Spiral inductor Monitor-diode. Stem SiOB Laser diode Monitor-diode TO-56 header Lead ` (a). (b). Fig. 1.5: The conventional TO-Can laser module with silicon optical bench, (a) with V-groove and (b) with RF matching resistor and a spiral inductor. [1.30]. Cap TO-56 header Compensation element Signal lead Dummy lead. LD cathode Dummy LD anode MD anode. (a) (b) Fig. 1.6: (a) The side view and (b) top view of the conventional optical module with compensation element. [1.31] 8.

(24) Figure 1.6 shows the side view and bottom view of the conventional optical module with a compensation element. The laser diode and photo-detector was bonded on the TO-56 headr and protected inside a metal cap. The compensation element based on a ceramic material was consisted of a thin-film resistor or a chip resistor and it was connected to the signal and the dummy lead. The compensation element also have predetermined resistor element and predetermined inductance element. This type of the TO-Can laser module can be externally impedance matching by the compensation element to improve the transmission bandwidth. For the conventional TO-Can structure, the electrical impedance of the glass/metal feedthru was difficulty to precisely control, because the limit of the lead diameter, dielectric value of the sealing glass, and difficult in controlling the position of the lead with respect to the thru hole in the header base. Figure 1.7 shows an improvement TO-56 header assembly with a rectangular ceramic platform perpendicular bisecting the base of the TO header. The optoelectronic devices and passive component were disposed on the ceramic platform. The ceramic platform with higher thermal conductivity could solve the temperature problem and high density of the electrical input/output could be designed on both sides of the ceramic platform. On the contrary, this type of ceramic TO header could easily control the electrical impedance of the transmission line on the ceramic platform.. (a). (b). Fig. 1-7: (a) Top view and (b) bottom view of the ceramic TO header assembly. [1.32] 9.

(25) 1.4 Overview of Dissertation In this dissertation, we demonstrate that the 10Gbps coaxial DFB laser module packaging can not only achieve the high performance but also can keep low cost by employing the conventional TO-Can processes. The laser module has a built-in matching resistor to reduce the resonant phenomenon. In order to optimize the module’s performance, detailed equivalent circuit model is employed to investigate both the DFB laser diode and the coaxial package comprehensively. This study makes it possible to fabricate the 10Gbps coaxial laser modules in low cost while still maintaining the high performance by just using the existing low-cost TO-Can package technology. Previously, the high-performance 10Gbps coaxial laser modules have only been available by using complicated design, customized components, and specialized fabrication process that lead to high packaging cost. The subsequent content of this thesis is organized as follows. Chapter 2 describes the equivalent circuit model of 10-Gb/s DFB laser diode. The comparisons of the theoretical and experimental results show a good agreement in frequency response. The fabrication procedures of the 10-Gb/s laser module are also described in chapter 2. Chapter 3 presents the equivalent circuit model of proposed coaxial package and the theoretical study result. The measurement results of 10Gbps coaxial DFB laser modules are also described in Chapter 3. This uncooled 10Gbps laser module operates at a high temperature up to 105oC and maintains an eye mask margin above 28% in the full operational temperature range to meet the stringent requirements of 10Gbps Ethernet for long reach applications. Chapter 4 describes the application of the 10-Gb/s BOSA modules by applying the 10-Gb/s coaxial laser module. A modulation bandwidth of 11.86GHz and an OC-192 eye diagram of 19% mask margin are obtained from the transmitter side. After 10-km single-mode-fiber (SMF) transmission, the mask margin for OC-192 eye diagram decreases 10.

(26) to 11%. For the receiver, an OC-192 eye diagram of 31% mask margin is obtained under back-to-back connections. The mask margin maintains at 29% after a 10-km SMF transmission. The measured crosstalk penalty is 0.9dB at the receiver side. These results indicate that the BOSA module is capable of a 10-Gb/s bi-directional transmission. Chapter 5 describes another application of the 10-Gb/s coaxial laser modules. A 4channels x 10-Gb/s CWDM laser module are described. The fabrication and characteristic of the 4 channels CWDM laser module are also presented. The results of the 40-Gb/s optical module showed that the output optical power was above -1 dBm per channel and the system power penalty was 12 dB. The transmission distance with a single-mode fiber reached more than 30 km at a bit-error-rate of 10-9. This proposed high-performance 40-Gb/s CWDM optical module demonstrates not only the feasibility of a 30 km transmission, but also shows the low-cost possibility of ensuring the application of WDM-passive optical network (WDM-PON) fiber-to-the-home (FTTH) systems. Chapter 6 draws the conclusions and the final remarks.. 11.

(27) 1.5 Reference [1.1]. [1.2] [1.3]. [1.4]. [1.5] [1.6]. [1.7]. [1.8] [1.9]. [1.10] [1.11]. [1.12]. J.M. Pedersen, T.P. Knudsen, O.B. Madsen, “Reliability demands in FTTH access networks,” in Proc. Advanced Communication Technology Conf., 2005, pp. 1202-1207. H. Shinohara, “Broadband access in Japan: rapidly growing FTTH market,” IEEE Communication Magazine, Vol.43, pp. 72-78, Sept. 2005. W.K. Park, Sung.I. Nam, C.S. Choi, Y.K. Jeong, and K.R. Park, “An implementation of FTTH based home gateway supporting various services,” in Proc. Consumer Electronics Conf., Jan. 2006, pp. 63-64. M. Nakamura, H. Ueda, S. Makino, T. Yokotani, and K. Oshima, “Proposal of networking by PON technologies for full and Ethernet services in FTTx,” J. Lightwave Technol., vol.22, no. 11, pp. 2631-2640, Nov. 2004. H. Hayashida, M. Yasunaga, T. Ema, and K. Nakazawa, “Reducing costs for first one mile FTTH lines” in Proc. of OFC 2005, vol.3, Mar. 2005. J.J. Yoo, J.D. Park, T.Y. Kim, H.H. Yun, and B.W. Kim, “WDM-PON platform development”, in Proc. of Advanced Communication Technology 2005, vol. 1, pp. 714-716, 2005. Y.H. Kim, Y.D. Bae, E.H. Lee, I. Kim, Y.C. Bang, J.K. Lee, Y. Oh, and D.H. Jang, “InGaAsP SSC LD for Low-cost Uncooled FTTH Module with Bandwidth over 4GHz,” in Proc. of Indium Phosphide and Related Materials, pp. 543-546, 2005. N. Gagnon, A. Girard, and M. Keblance, “Considerations and Recommendations for In-Service Out-of-Band Testing on Live FTTH Networks,” R. Luo, T.G. Ning, T.J. Li, L.B. Cai, F. Qiu, S.S. Jian, and J.J. Xu, “FTTH-A Promising Broadband Technology,” in Proc. of Communication, Circuits and Systems, vol. 1, p.p 609-612, May 2005. S. Kallukka and P. Raatikainen, “Link Utilization and Comparison of EPON and GPON Access Network Cost,” IEEE Globecom 2005, pp. 301-305, 2004. R.P. Davey, P. Healey, I. Hope, P. Watkinson, D.B. Payne, O. Marmur, J. Ruhmann, Y. Zuiderveld, “DWDM reach extension of a GPON to 135 km,” J. Lightwave Technol., vol. 24 , no. 1, pp. 29-31, 2006. E. Hugues-Salas, R. Razavi, T.J. Quinlan, M.P. Thakur, S.D. Walker, “A 2.5 Gb/s. Edge-Detecting Burst-Mode Receiver for GPON Access Networks,” in Proc. Optical Fiber Communication Conf., p.p 1-3, March 2007. [1.13] C.H. Yu, and D.U. Li, “A 2.5 Gb/s CMOS Burst-Mode Limiting Amplifier for GPON System,” in Proc. Circuits and Systems Conf., p.p 2538-2541, May 2007. [1.14] X.Z Qiu, Y.C. Yi, P. Ossieur, S. Verschuere, D. Verhulst, B. De Mulder, W. Chen, J. Bauwelinck, T. De Ridder, B. Baekelandt, C. Melange, J. Vandewege, “High 12.

(28) [1.15]. [1.16]. [1.17]. [1.18]. [1.19]. [1.20]. [1.21] [1.22] [1.23]. Performance Burst-Mode Upstream Transmission for Next Generation PONs,” in Proc. Optical Fiber Communication Conf., p.p 1-3, Oct. 2006. X.Z. Qiu, P. Ossieur, J. Bauwelinck, Y. Yi, D. Verhulst, J. Vandewege, B. De Vos, P. Solina, “High Performance Burst-Mode Upstream Transmission for Next Generation PONs,” J. Lightwave Technol., vol. 22 , no. 11, pp. 2498-2508, 2004. X.Z. Qiu, Y.C. Yi, P. Ossieur, S. Verschuere, D. Verhulst, B. De Mulder, W. Chen, J. Bauwelinck, T. De RIdder, B. Baekelandt, C. Melange, and J. Vandewege, “High Performance Burst-Mode Upstream Transmission for Next Generation PONs,” in Proc. Optical Fiber Communication and Optoelectronic Exposition Conf., p.p 1-3, Oct. 2006. N. Kashima, “Dynamic properties of FP-LD transmitters using side-mode injection locking for LANs and WDM-PONs,” J. Lightwave Technol., vol. 24 , no. 8, pp. 3045-3058, 2006. C. Schubert, R.H. Derksen, M. Moller, R. Ludwig,C.J. Weiske, J. Lutz, S. Ferber, A. Kirstadter, G. Lehmann, and C. Schmidt-Langhorst, “Integrated 100-Gb/s ETDM Receiver,” J. Lightwave Technol., vol. 25 , no. 1, pp. 122-130, Jan. 2007. J.P. Turkiewicz, E. Tangdiongga, G.D. Khoe, H. de Waardt, W. Schairer, H. Rohde, G. Lehmann, E.S.R. Sikora, Y. R. Zhou, A. Lord, and D. Payne, “Field trial of 160 Gb/s OTDM add/drop node in a link of 275 km deployed fiber,” in Proc. of OFC 2004, PDP1. S. Vorbeck, R. Leppla, W. Weiershausen, M. Schneiders, and E. Lach, “Long-haul field transmission experiment of 8 x 170 Gb/s over 421 km installed legacy SSMF fibre infrastructure,” presented at the ECOC 2005, We3.2.1. F. Koyama, “Recent Advances of VSCEL Photonics,” J. Lightwave Technol., vol. 24 , no. 12, pp. 4502 - 4513, Dec. 2006. E. Lach, and K. Schuh, “Recent Adcances in Ultrahigh Bit Rate ETDM Transmission Systems,” J. Lightwave Technol., vol. 24 , no. 12, pp. 4455 - 4467, Dec. 2006. K. Sakai, H. Aruga, S.I. Takagi, M. Kawano, M. Negishi, Y. Kondoh, and S.I. Kaneko,. “1.3µm uncooled DFB laser-diode module with a coupled differential feed for 10Gb/s Ethernet applications,” J. Lightwave Technol., vol. 22 , no. 2, pp. 574-581, 2004. [1.24] O. Stier, O. Reznik, F. Meyer-Guldner, R. Scholz, G. Wenger, N. Iwanowski, and M. Weigert, “Transmitter component for 10.5 Gbps at 1310 nm with receptacle and 50 ohm flexboard,” J. Lightwave Technol., vol. 5, no. 10, pp. 1403-1411, 1987. [1.25] M. Winter, R. Hauffe, and A. Kilian, “Simplified optical coupling and alignment scheme for cost effective 10Gb/s TOSA modules based on edge emitters hermetically packaged in micro-machined silicon structures,” in Proc. of OFC 2005, OThU8. [1.26] F. Mederer, I. Ecker, R. Michalzik, G. Steinle, H. Riechert, K.J. Ebeling, B. Lunitz, J. Moisel, and D. Wiedenmann, “VSCEL Transmitters For 10-Gibabit Ethernet: 1.3um 13.

(29) [1.27]. [1.28]. [1.29]. [1.30] [1.31] [1.32]. Wavelength VSCELs For Metroplitan Area Networks And TO-Packaged 850nm Wavelength VCSELs For Data Transmission Over Multimode Fibers And Optical Backplane Waveguides,” in Proc. of ECTC 2002, pp. 2-11, May 2002. A. Ebberg, F. Auracher and B. Borchert, “10Gbit/s transmission using directly modulated uncooled MQW ridge waveguide DFB lasers in TO package,” Electronics Letters, vol. 36, no. 17, Aug. 2000. A. Ebberg, R. Bauknecht, M. Bittner, M. Grumm, and M. Bitter, “High performance optical receiver module for 10Gbit/s applications with low cost potential,” Electronics Letters, vol. 36, no. 8, April 2000. O. Stier, D. Reznik, F. Meyer-Guldner, R. Scholz, G. Wenger, N. Iwanowski, and M. Weigert, “Transmitter Component for 10.5 Gbps at 1310mnm with Receptacle and 50 Ohm Flexboard,” in Proc. of ECTC 2004, pp. 1036-1041, May 2004. Y. C. Keh, and M .K. Park, “High Speed TO-CAN Based Optical Module,” US Patent 20040126066A1 S. H. Lee, and S Shi, , “Optical Telecommunication Module,” US Pattern 6721513B2 R. K. Rosenberg, G. Giaretta, S. Schiaffino, R. J. Hofmeistoer, “Ceramic Header Assembly,” US Pattern 006586678B1.. 14.

(30) Chapter 2 10-Gb/s Coaxial Laser Module Design and Simulation The structure of 10-Gb/s coaxial laser module is influenced by the limitations of the package structure and optoelectronics devices. In this chapter, the equivalent circuit model will be developed at first. It was divided into two parts, the active region of the laser diode and the structure of the laser diode package. Then, the package process of the coaxial laser module will also be introduced. Finally, the simulation results of the laser module will be shown and discussed.. 2.1 Equivalent Circuit Model of the High-speed Un-cooled DFB-LD The high frequency characteristics of the semiconductor lasers are strongly depends on the structure of the devices. Figure 2.1 shows the cross section of the ridge laser diode with InGaAsP ridge waveguide which was used to derive the circuit modeling of the active region in the laser diode [2.1]-[2.3]. Although the actual structure of the ridge laser we used was AlInGaAs heterojunction, the circuit model of the laser diode was almost the same. From the figure 2.1, the parameters of the RSR and RSS are ridge resistance and the submount resistance under the active region. The parameter of the RQ is the leakage resistance thorough the cladding layer.. Fig. 2.1: Cross section of a ridge laser. [2.1] 15.

(31) The capacitance parameters of the CN and CL are metal-insulator semiconductor capacitance and space-charge capacitance, respectively. The substrate resistance RSUB is in series with CL. The heterojunction diode at the quaternary p-n interface is expressed as DL. The relaxation oscillations of the laser diode were strongly influenced with the dynamic response of the active region. The resonance peaks of the relaxation oscillations were associated with the rate equations. The circuit model of the active region could derive from the single-mode rate equations. The rate equation of the photon density S(x) and electron density N(x) are written as [2.4] dS ( x) S ( x) N ( x) = G ( x) S ( x) − +β τp τn dt. (2.1). L2eff d 2 N ( x) dN ( x) J ( x) N ( x) = − − G ( x) S ( x) + ⋅ τn τn dt qd dx 2. (2.2). where τ p is the photon lifetime, τ n is the electron lifetime, β is the spontaneous emission coefficient, J(x) is the injected current density, q is the electronic charge, d is the active layer thickness, Leff is the effective carrier diffusion length. The optical gain G(x) is given by. G ( x) = Γy p{N ( x) − N g }. (2.3). where Γy is the optical confinement factor normal to the junction plane, p is a constant, N g is the electron density for zero gain. The junction voltage Vj(x) is given by N ( x) = N e exp{qV j ( x) / ηkT }. (2.4). where Ne is the equilibrium electron density, k is Boltzmann’s constant, T is the temperature is degree Kelvin. It is assume the photon S(x) and electron N(x) density are given by [2.2] S ( x) = 2S 0 cos 2 (πx / W ). (2.5). N ( x) = N 0 − N 1 cos(2πx / W ). (2.6) 16.

(32) where S0 is the average photon density, N0 is the average electron density, N1 is the deviation of carrier density from a uniform distribution, and W is the width of the active layer. These distributions are illustrated in Fig. 2.2.. Fig. 2.2: Distribution of photon density and electron density across the active layer. [2.2]. J(x) was assumed a constant across the active layer due to the narrow strip device with lateral carrier confinement. By quasi-static approximation, equation (2.5) and (2.6) substituted in equation (2.1) and (2.2), the rate equation were simplified as [2.3] dS 0 N N = {G ( N 0 )[1 − εS 0 ] − 1 / τ p }S 0 + β 0 + β 0 dt τn τn. (2.7). dN 0 I N = − 0 − G ( N 0 )[1 − εS 0 ]S 0 α τn dt. (2.8). Comparing with single-mode rate equation, the only difference [1-εS0] is a scaling factor in the gain term. This scaling factor describes equivalent gain saturation. The laser with non-uniform electron density and lateral carrier diffusion is equivalent to a laser in which there is no lateral diffusion, but which instead exhibits gain saturation. The average junction voltage is obtained by substituting (2.6) in (2.4). For a narrow-strip device, the maximum variation of Vj(x) across the active layer is few mille-volts. Thus, the average junction voltage 17.

(33) Vj0 can be written as 2kT ln( N 0 / N e ) q. V jo =. (2.9). The time-varying variables split into dc and ac components as follows. I = I 0 + ie jωt. (2.10). N 0 = N 00 + n0 e jωt. (2.11). V j 0 = V j 00 + V j e jωt. (2.12). G = G0 + ge jωt. (2.13). S 0 = S 00 + se jωt. (2.14). I = I 0 + ie jωt. (2.15). Equation (2.10), (2.11), (2.12), (2.13), (2.14), and (2.15) substituted into equation (2.7), (2.8) and (2.9) and written as [2.4] vi = i s ( Rs1 + Rs 2 + jωLs ). (2.16). i = v j (1 / R1 + jωC t ) + i s. (2.17). i s = αG0 (1 − 2εS 00 ) s ≅ αG0 s. (2.18). R1 = Rd [1 + γτ n S 00 (1 − εS 00 )] −1 ≅ Rd (1 + γτ n S 00 ) −1. (2.19). Rd = 2τ n kT / qαN 00. (2.20). C t = C d + C sc. (2.21). C d = τ n / Rd. (2.22). where. Ls =. Rd G0 (1 − 2εS 00 )[ β + γτ n S 00 (1 − εS 00 )]. Rs1 = εG0 S 00 Ls Rs 2 =. (2.23) (2.24). β N 00 Ls τ n S 00. (2.25). 18.

(34) The small-signal circuit model of the laser follows directly from equation (2.16) and (2.17), and it is shown in Fig. 2.3. The parameter of Cs and Rs are modeled as the contact capacitance and resistance. The small-signal photon density s is proportional to the current is from equation (2.18). The current is can be used as the optical output intensity or IM response, and the output voltage vs across Rs2 can be used as a measure of optical intensity.. Fig. 2.3: Small-signal circuit model of the device chip. [2.4]. 2.2 Simulation Results of the High-speed Un-cooled DFB-LD According to the previous equivalent circuit model, we built the equivalent circuit model for the high-speed uncooled DFB LD, as shown in Fig. 2.4. The equivalent circuit model of the active region in the ridge waveguide laser was based on the single-mode rate equations. In the active region, the photon storage and the charge storage were modeled by the inductance L2 and the diffusion capacitance C4, respectively. The capacitance of C5 was the space-charge capacitance in the active layer. The relaxation oscillation frequency was contributed major by C4 and C5. The inductance L2 represented the energy transfer from carriers and photons. The R6 and C7 represented the series resistance of the laser and the parasitic capacitance from the other area, respectively. The C7 was smaller compared with the parasitic capacitance in the active layer. The related parameters values were listed in Table 2.1. In order to evaluate the validity of these parameters used in the equivalent model, the values of the related parameters in the equivalent circuit model were adapted by curve fitting method. The calculated results 19.

(35) of the DFB LD frequency response were compared with the measured results, as shown in Fig. 2.5. The comparison of small signal results between the theoretical and the experiment results shows good agreement. The frequency peak of the DFB LD due to the relaxation oscillation effect were observed at 8.2, 9.5, and 11.8GHz, as the operation current was set at 30, 40, and 50mA, respectively. The measured -3dB bandwidth of the DFB LD frequency response were 12.2, 14.2, and 16.6 GHz. The results indicated the DFB LD used in the laser module with an excellent performance was suitable for the 10-Gb/s application. The small-signal frequency response was calculated by Advanced Design System (ADS) software.. Fig. 2.4: Small-signal circuit model of the device chip.. Table 2.1: Elements of the equivalent circuit in the active region. Element Unit. Element Unit. Value. Value. R1. mΩ. 140. C5. pF. 10. L2. pH. 5. R6. Ω. 5. R3. Ω. 1. C7. pF. 0.2. C4. pF. 220. 20.

(36) 5. Response (dB). , ,. Measured Simulation. 0. 50mA. -5. 40mA. -10 30mA. -15 0. 2. 4. 6. 8. 10. 12. 14. 16. 18. 20. 22. Frequency (GHz) Fig. 2.5: The small signal frequency response of DFB laser diode at various bias currents.. 2.3 Circuit Modeling of the High-speed Laser Module There are several optoelectronic package types of the TO-Can materials for different applications, such as TO-5, TO-18, TO-46, and TO-56. The profile view reveals a very important functional difference between TO-5, TO-18, TO-46 and TO-56 as shown in Fig. 2.6 (a), (b) and (d), respectively [2.5]. TO-5 and TO-18 have a hollow head and are filled with glass. TO-46 and TO-56 have a solid head and are made out of Kovar or Alloy42. The glass material was only filled between the lead and the head used for insulation. These packages types of TO-5, TO-18, and TO46 have two different diameters, plateau diameter and outside diameter. Usually, these package types were used for the optical receiver applications. Figure 2.6 (d) shows the profile-view of the TO-56 type and it has only one outside diameter. It is one of the commercial TO-Can materials and commonly used for optical transmitter package. In general, the conventional TO-Can header suffers poor RF transmission characteristics without proper modification. The notch filter effect induced by 21.

(37) the parasitic inductance of the long lead and wires is one of the major factors. The resonant phenomenon due to the impedance mismatch is also the key issue. Here, we analyzed the structure of the commercial TO-Can header and developed the equivalent circuit model.. outside plateau. glass. (a). glass. (b). Kovar/Alloy42. Kovar/Alloy42. (c). (d). Fig. 2.6: Different types of optoelectronics package, (a) TO-5, (b) TO-18, (c) TO-46, and (d) TO-56. [2.5]. Figure 2.7 shows the structure of TO-56 header we adapted in the study at the top-view and the profile-view. The diameter and thickness of the TO-56 header are 5.6mm and 1.2mm, respectively. The space of 1.3mm x 1.0mm on the stem lead was used for placing the optoelectronic devices. The stem lead was physically contact and electrical short with the header. It was used for improving the thermal conduction caused by the laser diode. The 22.

(38) diameter of the lead is 0.45mm and is surrounded by glass seal (εs=5.5) for insulating from the header. The laser diode with a submount was bonded on the stem lead and a monitor diode was bonder on the center of the header.. 5.6mm Stem lead. 1.0mm Stem lead. 1.3mm Header. 1.2mm Lead Glass seal. Lead. Position of Monitor Diode Lead (a). (b). Fig. 2.7: Structure of TO-56 header, (a) top view and (b) cross section.. Based on the structure of the commercial TO-56 header shown as Fig. 2.7(b), the lead was divided into three parts, the lead above the header, the lead below the header, and the feed-thru lead. The lead above and below the header could be taken as a round lead. The equivalent circuit of these leads could be roughly modeled as a conductive leads as shown in Fig. 2.8 and value of the inductance was estimated by equation (2.26) [2.6].. r. ι. L. Fig. 2.8: Schematic and equivalent circuits of the conductive leads. [2.6]. 23.

(39) L = 2µ s l{ln. l + l2 + r2 r r − 1 + ( ) 2 + } × 10 − 7 [H] r l l. (2.26). where the relative permeability μs was unity, l is the length of the lead, r is the radius of the lead. According to equation (2.26), the lead above the header is about 1.0 mm and the estimated value of the inductance was about 0.2628 nH. The lead below the header could be cut as short as 2.0 mm, and the estimated value of the inductance was about 0.7575 nH. Figure 2.9 shows the estimated inductance value of different lead length by equation (2.26). As increasing the length of the lead, the calculated inductance of the lead was increased. The increased inductance of the lead would limit the bandwidth of the lead. In order to ensure a maximum bandwidth of the laser module, lead length should be designed as short as possible.. 10. Inductance (nH). 8 6 4 2 0 0. 2. 4 6 8 Length of Lead (mm). 10. Fig. 2.9: Calculated the inductance of the lead with different length.. The third part of the feed-thru lead was surrounded by glass seal and inside the header. It can be taken as a coaxial lead. The schematic diagram and the equivalent circuit of the coaxial lead were shown in Fig. 2.10. The coaxial lead could be roughly modeled as a series inductance and a shunt capacitance. Equation (2.27), (2.28) and (2.29) was derived from the basic transmission line theory [2.7].. 24.

(40) L. C. Fig. 2.10: Schematic and equivalent circuits of coaxial leads [2.7].. C=. 2πε 0 ε s ×l ln(b a ). L = z 02 C = [. Z0 =. [F]. (2.27). µ0 ln(b a )] ⋅ l [H] 2π. (2.28). µ0 c ⋅ ⋅ ln(b a) [Ω] 2π εs. (2.29). where ε0 and μ0 are the dielectric constant and permeability in free space, respectively. The parameters of ‘a’ and ‘b’ are defined as the inner and outer diameter of the conductivity. The feed-thru lead inside the header surrounded by glass could be taken as a coaxial lead as shown in Fig. 2.10. The equivalent circuit of the lead could be estimated by equation (2.27), (2.28) and (2.29). A resonance peak was caused by the parasitic capacitance C and inductance L and it could be roughly described as equation (2.30) [2.8]. According to this equation, the L-C resonance frequency was calculated and shown in Fig. 2.11. For example, the wire to wire inductance of the gold wire with 100 mm long is commonly taken to be 0.1 nH/mm, and Cs of 2.53 pF produced an L-C resonance frequency peak at 10 GHz. From equation (2.30), we could deduce the maximum parasitic capacitance C caused from the package structure.. fr =. 1 2π LwC s. [Hz]. (2.30). 25.

(41) 10. Inductance (nH). 1 GHz. 1 0.1. 5 GHz 10 GHz 50 GHz. 0.01 100 GHz. 1E-3 0.1. 1 10 Capacitance (nF). 100. Fig. 2.11: Calculated the L-C resonance peak frequency.. In order to easy measure, the coaxial laser module was soldered and attached to a FR4 evaluation board with a SMA connector. Figure 2.12 shows a schematic diagram of the proposed laser module. To reduce the resonant phenomenon, we inserted a matching resistor between lead and laser diode. Since we intended to design an uncooled and simple laser module, the impedance of the TO-Can device was designed to be 25 ohm. There were two sections of the bonded wires inside the TO-Can package. One of the bonded wires was used to connect the lead and the matching resistor; while the other bonded wire was connected the matching resistor and the laser diode. The equivalent circuit of the proposed laser module was shown in Fig. 2.13 [2.9]-[2.15]. The L18 in the equivalent circuit represented the first wire and the second wire that was modeled by inductance L16. The built-in matching resistor was modeled by resistor R9 with the induced parasitic capacitance of C15.. 26.

(42) SMA Connector. TO-Can laser Signal Line. Ground Laser Diode & submount Matching resistor & submount Fig. 2.12: A schematic diagram of the proposed module.. TL2 MLIN TL2. Term1. Term Term1. L7. L5. L L7. L L5. C7. C C7 C16. L17. L18. L L17 C C16. C17. L L18 C C17. C15. R9. L16. R R9. L L16. C C15. C18 C1 5 R12. C C18. R R12. Laser Diode. Fig. 2.13: The equivalent circuit model of the proposed module.. In this work, the signal lead was divided into three sections. The lead outside the TO-Can used to solder with PCB was modeled by L7, the feed-through lead surrounded by the glass seal was modeled by inductance L5, and the lead inside the TO-Can used to bond wire connecting the laser was modeled by inductance L17. There were some parasitic capacitances in the equivalent circuit. The C16 and C17 represented the parasitic capacitances induced by the lead and TO-Can header. Since we soldered the lead with microstrip line 27.

(43) together, the capacitance C7 and inductance L7 were inserted to represent the parasitic effect of the microstrip line. The C18 and R12 represented the induced parasitic capacitance and resistance, respectively. Since we intended to use the commercial TO-Can header, the C16, L17, C17, L18 were fixed. However, there were two controllable parasitic parameters to improve the performance of the laser module. The parasitic inductance due to the bonded wires L16 and L18 could be controlled. The bonded wire L18 was easily reduced, since we could use the ribbon or multi-wires to reduce the parasitic inductance. Compared with the previous the bonded wire connecting the lead and the matching resistor, the wire bonded L16 on the laser to the matching resistor was more critical. The bond pad on the laser diode was usually small about 100 x 100μm and the wire bonding was difficult to maintain at a low parasitic inductance. The second controllable parasitic inductance was L7. We could cut the lead into different length to control the parasitic inductance. Although a short lead has the benefit of low parasitic inductance, but it is unpractical to use. By combining with the equivalent circuit of laser diode, we could calculate the RF characteristics of the laser and could find out the reflection and transmission coefficients. The bias condition of laser diode was chosen at 40mA which was the normal operation situation. According to the analysis of the signal lead as described before, rough values of the L7, C7, L5, C16, L17 were calculated from equation (2.26) ~ (2.29). Then, the accuracy values of these parameters were decided by comparing the simulation results with the experimental results. The equivalent circuit model values of the matching resistor R9 and C15 was decided by the vendor’s specification [2.16]. The equivalent circuit values of C18 and R12 were the submount’s parasitic [2.6], [2.11]. The parameters used for simulation were listed in Table 2.2.. 28.

(44) Table 2.2: Elements of the equivalent circuit of a TO-Can laser. Element. Unit. Value. Element. Unit. Value. C18. pF. 3. C17. pF. 0.1. R12. Ω. 2. L17. nH. 0.185. L16. nH. 0.3. C16. pF. 0.465. R9. Ω. 18. L5. nH. 1.1. C15. pF. 0.1. C7. pF. 0.35. L18. nH. 0.3. L7. nH. 0.95. 2.4 Simulation Results of the Coaxial Laser Module In order to optimize the performance of the laser module, two controllable parameters were calculated from the circuit model of the laser module, one is the inductance of the bonding wire, the other is the signal lead length. Fig. 2.14 shows the simulation results of the RF reflection performance with various bonded wires. The reflection was serious when the parasitic inductance of the bonded wire was larger than 0.3 nH. The reflection was controlled under -7 dB at a frequency of 10 GHz as the induced parasitic inductance of bonded wire was reduced to 0.3 nH. Due to the resonant effect, the reflection at lower frequency was slightly increased for a smaller parasitic inductance. But it would not degrade the RF characteristics of the device obviously. The theoretical results of the small signal response and the 3-dB bandwidth of the proposed TO-Can laser at different bonded wire inductance are shown in Figs. 2.15 and 2.16, respectively. When the bonded wire inductance was 0.3 nH, the 3-dB bandwidth was as high as 9.73 GHz. The 3-dB bandwidth saturated at about 10 GHz for a smaller bonded wire inductance.. 29.

(45) Reflection (dB). 0. -10. -20. 1.5 nH 1 nH 0.5 nH 0.3 nH 0.2 nH. -30. -40 0. 2. 4. 6. 8. 10. 12. 14. 16. 18. 20. Frequency (GHz) Fig. 2.14: RF reflection characteristics of laser modules.. Transmission (dB). 0 0.2 nH 0.3 nH 0.5 nH 1 nH 1.5 nH. -10 -20 -30 -40 -50 -60 0. 2. 4. 6. 8. 10. 12. 14. 16. 18. Frequency (GHz) Fig. 2.15: RF transmission characteristics of laser modules with different bonded wire inductances.. 30. 20.

(46) 3dB Bandwidth (GHz). 12 10 8 6 4 2 0. 0.2. 0.4. 0.6. 0.8. 1. 1.2. 1.4. 1.6. Bond Wire Inductance (nH) Fig. 2.16: Small signal 3dB bandwidth of laser modules with different bonded wire inductances.. Figures 2.17, 2.18, and 2.19 show the simulation results of the RF performance with various lead lengths. The longer lead length would result a higher reflection, as showed in Fig. 2.17. We should keep the lead length shorter than 2 mm to maintain the reflection under -7 dB at 10 GHz. Such a short lead length might induce a low production yield when the device soldered with PCB together. According to multi-source agreement (MSA) of 10-Gb/s laser module, the module should be attached with a perpendicular flexible PCB. It would be easier to keep the lead length short when adopt the through-hole soldering process. The simulation result of RF transmission characteristics and the 3-dB bandwidth for different lead length are shown in Figs. 2.18 and 2.19, respectively. Lead length influenced the transmission characteristic significantly. The 3 dB bandwidth saturated at about 10 GHz.. 31.

(47) 0. Reflection (dB). -5. 1 mm 2 mm 3 mm 4mm 5mm. -10 -15 -20 -25 -30 0. 2. 4. 6. 8. 10. 12. 14. 16. 18. 20. Frequency (GHz) Fig. 2.17: RF reflection characteristics of laser modules with different lead lengths.. 0 1 mm 2 mm 3 mm 4mm 5mm. Transmission (dB). -10 -20 -30 -40 -50 -60 -70 0. 2. 4. 6. 8. 10. 12. 14. 16. 18. 20. Frequency (GHz) Fig. 2.18: RF transmission characteristics of laser modules with different lead lengths.. 32.

(48) 3dB Bandwidth (GHz). 12. 10. 8. 6. 4 0. 1. 2. 3. 4. 5. 6. Length of Lead (mm) Fig. 2.19: Small signal 3-dB bandwidth of laser modules with different lead lengths.. According to the simulation results, the 10 Gbps transmission was shown to be feasible by employing the conventional TO-Can architecture. The key issues of coaxial laser modules were listed following: 1. A matching resistor should be inserted to minimize the impedance mismatch. 2. The bonded wire inductance should be controlled under 0.3 nH to prevent the degradation of the reflection and 3-dB bandwidth. 3. The lead length should be controlled shorter than 2 mm to prevent the degradation of the reflection and 3-dB bandwidth.. 2.5 Analysis and Optimization of Present 10-Gb/s Laser Module Package Both TO-56 and TO-46 header are widely used for edge emitting and vertical emitting in laser packages. In order to adopt the edge emitting laser diode, the TO-56 header is used in this work. The equivalent circuit model of the 10Gbps laser module packaged by. 33.