隨著通訊技術的逐年演進，新的應用對於即時傳輸巨量數據的能力也越發的要求，資料傳輸率的成長速度幾乎以每五年十倍的速率不斷的進步。以乙太網路(Ethernet)為例，從早期僅具有每秒十兆位元(10Mbps)之資料傳輸率到現在已經到達每秒十吉位元(10Gbps)的資料傳輸率，甚至還在制定中的一百吉位元(100Gbps)之乙太網路，時脈與資料恢復電路(Clock and Data Recovery, CDR)在通訊系統速度的演進上扮演著重要的角色。
本論文旨在提出一個具有每秒二十五吉位元(25Gbps)資料傳輸率的時脈與資料恢復電路(Clock and Data Recovery, CDR)設計，此電路將著眼於未來一百吉位元之乙太網路系統的應用來設計，此系統將分為四路，每路以二十五吉位元之資料傳輸率為基礎做設計。論文中介紹過去到現在時脈與資料恢復電路的演進，以及此電路設計時應用上的考量，從系統架構說明到類比部份的電路模擬再到射頻部份之電磁模擬，詳細內容將在論文中一一討論。電路設計上使用了半速率的二位元相位偵測器(Bang-Bang Phase Detector, BBPD)，相較於線性相位偵測器(Linear Type Phase Detector)，當操作頻率超過1GHz時不會有線性區間的問題。並且使用了半速率相差九十度相位差之差動時脈訊號來對輸入之不歸零(Non-Return-to-Zero, NRZ)訊號做取樣，除了能獲取相位資訊外，當相位鎖上時，時脈與資料恢復電路同時也能對輸入的不歸零訊號進行解雙工的處裡。
壓控震盪器電路(Voltage Controlled Oscillator, VCO)以LC-tank架構實現，使用頂端偏壓架構使得輸出時脈訊號的偏壓位於電源電壓(Power Supply Voltage)的一半，使得頻率變化範圍能夠最大化。最後除頻器電路(Frequency Divider, FD)則使用了電流模態邏輯閂來實現，兩組電流模態邏輯透過相互鎖定產生九十度相位差的時脈訊號，自振頻設計為12.5GHz，經過一級緩衝器後傳到相位偵測器。
本論文的時脈與資料恢復電路採用TSMC 90 nm 1P9M製程實現。壓控震盪器的中心頻率設計為25GHz，透過除頻器產生半速率且相差九十度之時脈訊號來對資料做取樣而產生領先或落後的資訊，透過一比較周期內做兩次領先或落後的比較，此電路具有較佳的相位偵測功能，輸出資料之資料率為12.5Gbps於相位鎖定時具有0.85ps(rms)的邊緣抖動範圍。此電路的操作電壓為1.2V，功耗為67.6mW，晶片尺寸包含PAD為0.777 mm2。

With the ever-growing communication technology, the more demands on the ability of transferring huge amount of data in a limited period in new applications. The progress of data transfer rate grows up with the speed of almost 10 times per 5 years. Take Ethernet for example, in the beginning the data transfer rate is only 10Mbps but now already grown up to 10Gbps, even the future 100Gbps Ethernet, which is under establish, the clock and data recovery circuits still play an important role in the progress of data transfer rate in communication systems.
This thesis aims to propose a clock and data recovery circuit with the data rate of 25Gbps, which is designed to be used in the future 100Gbps Ethernet system. The 100Gbps system is designed to be divided into four channels, each channel operates in 25Gbps data rate. The dissertation will introduce the evolution of clock and data recovery circuit and considerations including the system blocks operation and analog circuit techniques and even electrical-magnetic simulation of radio frequency circuit that should be taken into in the design of such circuits, all the topics will be discussed in the following contents. The architecture of phase detector is half-rate, Bang-Bang type, compared to the linear type counterpart, which avoids the linear range problem when the operation frequency is over 1GHz.With the input NRZ data sampled by the half-rate quarter phase differential clock signal, not only the phase information of lead or lag can be obtained but also the input data can be demutiplexed by quarter phase clock signal when the phase is locked.
The voltage controlled oscillator is realized with LC-tank architecture, with top-biased topology, the common mode voltage of the output clock signal can be half of the power supply voltage, which makes the most output frequency tuning range. Finally, the current mode logic is applied in the frequency divider circuit to make the self-oscillating frequency at 12.5GHz, the quarter phase is generated because two current mode logic latches lock each other. After passing through one-stage current mode logic buffer, the half-rate clock signal is applied to the phase detector.
This clock and data recovery circuit is realized by TSMC 90 nm 1P9M process. The center frequency of voltage controlled oscillator is designed in 25GHz, followed by frequency divider half-rate quarter phase clock signal is produced and can be used to sample input NRZ data to detect phase information. Since there are two phase detection within one comparison period, this circuit has better phase detection ability compared to ordinary counterpart. The output data rate is 12.5Gbps. When the phase is locked, the output data jitter is 0.85ps, rms. The supply voltage is 1.2 volte, power dissipation is 67.6 mW in simulation, the layout size is 0.777 m2 including PAD.

[1] Kevin Deierling, “2017 Predictions, Networking Edition”
[2] Jri Lee, “Communication Integrated Circuits”
[3] C. R. Hogge, “A Self-Correcting Clock Recovery Circuit,” IEEE J. Lightwave Tech., vol. 3, pp.1312-14, Dec. 1985.
[4] J.D.H. Alexander, “Clock Recovery From Random Binary
Signals,” Electronics Letters, vol. 11, pp. 541-542, Oct. 1975.
[5] Nikola Nedovic, “A 40–44 Gb/s 3X Oversampling CMOS
CDR/1:16 DEMUX,” IEEE. J. Solid-State Circuits, VOL. 42, NO. 12, DECEMBER 2007.
[6] Behzard Razavi, “Design of Integrated Circuits for Optical Communications, 2nd Edition”
[7] Jri Lee, Ming-Shuan Chen, and Hui-De Wang, “Design and
Comparison of Three 20-Gb/s Backplane Transceivers for
Duobinary, PAM4, and NRZ Data,” IEEE. J. Solid-State
Circuits, VOL. 43, NO. 9, SEPTEMBER 2008.
[8] Samuel Palermo, “Texas A&M_Sam Palermo_High speed Links
Circuits and Systems” lecture note 4.
[9] A. X. Widmer and P. A. Franaszek, "A DC-Balanced,
Partitioned-Block, 8B/10B Transmission Code," IBMJ. Res.
and Develop., vol. 27, pp. 440-451, Sept. 1983.
[10] M. Verbeke, P. Rombouts, X. Yin, and G. Torfs, “Inverse
Alexander phase detector,” Electron. Lett., vol. 52,
no. 23, pp. 1908–1910, Nov. 2016.
[11] J. Lee and B. Razavi, “A 40-Gb/s clock and data recovery
circuit in 0.18-um CMOS technology,” IEEE J. Solid-State
Circuits, vol. 38, no. 12, pp. 2181–2190, Dec. 2003.
[12] J. Lee, K. S. Kundert, and B. Razavi, “Analysis and
modeling of bangbang clock and data recovery circuits,”
IEEE J. Solid-State Circuits, vol. 39, no. 9, pp. 1571–
1580, Sep. 2004.