Transactions Template JOURNAL OF ENGINEERING RESEARCH AND TECHNOLOGY, VOLUME 3, ISSUE 4, DECEMBER. 2016 78 Ultra-Wideband and Low Noise Transimpedance Amplifier Jawdat Y. Abu-Taha 1 1 Department of Electrical Engineering, IUG, Gaza, Palestine, e-mail: jtaha@iugaza.edu.ps Abstract— We present a new transimpedance amplifier (TIA) design possessing an improved bandwidth. This TIA employs a parallel combination of two series resonate circuits with different resonate frequencies on the conventional regulated common gate (RGC) architecture. In the proposed TIA, we employ the capacitance degeneration and series inductive peak- ing for pole-zero elimination. We implemented the layout of proposed TIA in a 0.18-µm CMOS process, where a 100-fF photodiode is considered. Our post-simulation test results show that the TIA provides 53-dBΩ transimpedance gain and 24- pA/√𝐇𝐳 input referred noise. The designed TIA consumes 11 mW from a 1.8-V supply, and its group-delay variation is 5-ps over 13-GHz 3-dB bandwidth. Index Terms—Bandwidth, capacitance degeneration network, input-referred noise, photodiode-detecor (PD), regulated common gate (RCG) and transimpedance amplifier (TIA). I. INTRODUCTION Continuous growth in the wireless telecommunication has derived to high level of chip integration and focused research studies towards the field of high frequency applications [1]. The accelerated CMOS technology is the only candidate that can satisfy the demands for low-cost and high integration with reasonable speed for analog applications in the Giga-Hertz range [2]. Fig. 1. Block diagram of an optical reciever. The transimpedance amplifier (TIA) is the critical block in the optical communication system that converts the induced photodiode current into an amplified voltage signal to be used in the digital processing unit (Fig. 1). The bandwidth is con- sidered as the highest priority in TIA design. The challenge in TIA design lies in the large photodiode parasitic capacitance Cpd, the input node that degrades the performances of the TIA. Therefore, it is required to decrease the input parasitic effects and prior to focusing on the compromise between the bandwidth and the noise [3, 4]. There have been two commonly used topologies in designing wideband CMOS TIAs: the common gate (CG) amplifier and the shunt feedback amplifier [5]. Several bandwidth en- hancement efforts have been reported in published literature which were based on isolation of the input capacitance of the photodiode to minimize its effect on the bandwidth calcula- tion. Inductive peaking is one of the commonly used tech- niques to improve the bandwidth and decrease parasitic ca- pacitance effects [6]. Placing an inductor in a strategic loca- tion of the amplifier circuit provides a resonance with parasit- ic capacitances, which expands the bandwidth of the TIA [7- 9]. Capacitive peaking has been used for bandwidth extension by using a capacitor to control the pole locations of a feed- back amplifier [10, 11]. Multiple shunt parallel feedback is another approach for enhancing the bandwidth [12]. The ef- fect of the photodiode capacitance can be more professionally reduced from the bandwidth limitation by using regulated cascode (RGC) [3]. In this work, we have proposed a new TIA design with im- proved bandwidth. The proposed TIA is based on modification of the input part of the conventional RGC TIA architecture by using parallel arrangement of two series resonate circuits with different resonate frequencies. Capacitance degeneration and series inductive peaking networks are used for pole-zero elim- ination to improve the bandwidth. The paper is organized as follows: in Section II we present an overview of the traditional RCG input stag. The concept of modified RCG input stage and the analysis of the architecture of parallel arrangement of two series resonate circuits with different resonate frequencies are introduced. We present the capacitance degeneration architecture and proposed TIA de- sign in Section III and Section IV, respectively. Finally, we present the noise analysis in Section V, demonstrative simula- tion results in Section VI, and the conclusions in Section VII. II. REGULATED COMMON GATE (RCG) INPUT STAGE A. Conventional RCG Input Stage Among all the building blocks in an optical communication system, the TIA is the one of the most critical blocks in re- ceiver design. It is a well-known fact that RGC input configu- ration can attain better isolation within the large photodiode capacitance 𝐶 by local feedback topology. Fig.2 shows the schematic diagram of the conventional RGC with a PD, which converts the incoming optical signal to a small signal rent 𝐼 . The common-source (CS) amplifier consists of 𝑀 Photodiode Transimpedance Amplifier High gain Amplifier Clock & Data Recovery De-Multiplexer Data Decoding Digital Signal Processing Receiver Jawdat Abu_Taha / Ultra-Wideband and Low Noise Transimpedance Amplifier(2016) 79 and 𝑅 operates as a local feedback technique and regulates the CG. As a result of the small-signal analysis, the input resistance of the RGC circuit is given by [13] and [14].   Dmm RCGi Rgg Z 12 , 1 1   , (1) where 𝑔 and 𝑔 are the transconducatnce of 𝑀 and 𝑀 respectively. It is clearly seen that the input resistance de- creased because the transconductance 𝐺 is (1 + 𝑔 𝑅 ) times larger than that of CG amplifier input stage, where 1 + 𝑔 𝑅 is the DC gain of the local feedback. Therefore, RGC stage acts as a buffer between the PD and the TIA stage and decreases the effect of the photodiode capacitance 𝐶 [14]. DD M1 bias Rs Buffer Zi,RCG Output voltage M2 Cpd I pd RD Fig. 2. Regulated common gate (RCG) TIA. B. Modified RCG Input Stage In the design of ultra-wideband TIAs, the wideband input stage plays very critical role. The design methodology of the narrow-band TIA is our first focus followed by demonstration of how to extend its input bandwidth. CgsRs L I i Zi gm 1 Rs VG L io Ii Zi (a) (b) I pd Cpd Fig. 3. The input part of a narrowband RCG TIA. Fig. 3 shows the input part of a typical narrowband TIA to- pology. The RGC TIA topology improves the bandwidth limi- tation due to the input pole that consists of the gate-source capacitance (𝐶 ) and the input resistance (𝑍 ). Nevertheless, the large parasitic capacitance of photodiode 𝐶 still reduces both the bandwidth and the noise performance of the TIA. The series inductive peaking technique is used to overcome this problem. The inductor L is placed between 𝐶 and of 𝐶 , which creates an inductive π network [14]. The expres- sion used to analyze the performance of the current transfer function is derived from the small-signal mode circuit shown in Fig.3. (b).   1 1 1 1 1)( 1 0 2 0 3 0 23                       s m ks k m ks CCsRLCsCRCsI I gspdgsgspdpd i (2) where R = (1 𝑔 ⁄ )//𝑅 , 𝑘 = , 𝑚 = ( ) and the cutoff frequency 𝜔 = ( ) . Inductive-peaking technique provides significant bandwidth extension ratio (BWER) by selecting different values for variables 𝑘 and 𝑚 [15]. M1-1 Rs1 VG M1-2 Rs2 L1 L2 io i i Zi Fig. 4. The input part of RCG TIA with two input branches. To improve the input-bandwidth, we use a parallel combina- tion of two series resonate circuits with different resonate fre- quencies, as shown in Fig. 4. The input impedance is given by 21 // ZZZ i  (3) where . 1 1 222 2223 222 jgsjj jjjgsjigsj jgsjj j j RC LRCLRCL j RC R Z jjjj         (4) 𝐶 , 𝐿 and 𝑅 are the gate –source capacitance , serial induc- tor and equivalent input resistance of transistor 𝑀 , respec- tively (𝑗 = 1, 2). Jawdat Abu_Taha / Ultra-Wideband and Low Noise Transimpedance Amplifier(2016) 80 In (4), one should note that, if the reactive elements are accu- rately selected, then the input impedance become purely resis- tive. Moreover when the gate of 𝑀 and 𝑀 have the same bias voltages, 𝑀 and 𝑀 have identical cutoff fre- quency 𝜔 . As a result, the circuit can realize a wide band- width. III. THE CAPACITANCE DEGENERATION Modification of RGC input stage can be augmented through the possibility of achieving a broadband frequency response through the increment of the effective transconductance 𝐺 of the circuit at high frequencies [5],[16]. To emphasize more on the above stated point, we can compensate the dominant pole of the overall circuit with a zero, which can be reached through capacitive degeneration configuration [14]. DD Mx RS RD CS vin vout v Fig.5. Configuration of capacitive degeneration. For the capacitive degenenation toplogy shown in Fig.5, the transconductance equivalency is calculated as [17]. sm ss ss sm m m Rg CR s CsR Rg g G      1 1 1 1 (5) which introduces a zero (𝑧 ) at (𝑅 𝐶 ) and a pole at (1 + 𝑔 𝑅 ) 𝑅 𝐶 ⁄ . The dominant pole can be compensated by the zero. As a result, the bandwidth is limited by the second lowest pole of the circuit. The propsed capacitive degenartion toplogy shown in Fig.6 is employed to provide capacitive and resisitive degeneration. Therefore, extra gain and bandwidth enhancement can be achived at the same time. DD M3 RD3 DD M4 RD4 R C Z  ZA 1  ZA11  Fig.6. The proposed configuration of capacitive degeneration. The transconductance equivalency of half of the circuit in Fig.5 is expressed as   ssm m m CsRRg sRCg G    2 1 1 (6) Note that in (6), the transconductance introduces a zero (𝑧 ) at (𝑅 𝐶 ) and brings an additional pole 𝑝 at (1 + 𝑔 𝑅 2⁄ ) 𝑅 𝐶 ⁄ . The dominant pole 𝑝 = appears at the drain node. If 𝑅 𝐶 = 𝑅 𝐶 , then the zero 𝑧 cancels the pole 𝑝 , therefore the bandwidth is extended to the se- cond pole of the circuit 𝑝 = (1 + 𝑔 𝑅 2⁄ ) 𝑅 𝐶 ⁄ . In pole-zero elimination technique, if the zero is moved to a lower frequency (𝐶 large), the frequency response displays a source peaking so that the capacitor should be small to avoid the gain peaking. This is an important advantage of the intended circuit stems from the variation of the amplifier’s input impendance and thus the proceeding stage load is seen. IV. THE PROPOSED TIA We present the proposed wideband TIA based on RCG in Fig. 7. The modification of the input network of RCG TIA pro- vides better enhancement of bandwidth and decreases the input-referred noise current. A 100-fF photodiode capacitor is Jawdat Abu_Taha / Ultra-Wideband and Low Noise Transimpedance Amplifier(2016) 81 used at the input of the TIA. The gain stage is composed of two common source amplifiers with capacitance degeneration technique. Referreing to Miller theorem [18], the shunt impedance 𝑍 = (𝑅//(1/sC)) connected between the drain nodes of 𝑀 and 𝑀 can be separated into a couple of grounded impedanc- es. If A is the voltage gain between the two terminals of 𝑍 in Fig.5, then the equivalent split impedances are −(𝐴 − 1)𝑍 and [(1 − 1 𝐴⁄ )𝑍] . These impedances produce zeros with 𝑅 and 𝑅 to make perfect cancellation of the poles at the drain of 𝑀 and 𝑀 . Therefore, the bandwidth is further improved [14]. The source follower consisting of 𝑀 and 𝑅 as a buffer is used to evade affecting the frequency response of the TIA due to input parasitic capacitances of the succeeding stage in the receiver system namely, the Limiting Amplifier (LA). V. NOISE ANALYSIS In the proposed TIA of Fig. 6, we consider the thermal noise generated by the active devices (𝑀 , 𝑀 and 𝑀 ) and thermal noises of resistors (𝑅 ,𝑅 , 𝑅 and 𝑅 ). The flicker noise (1 𝑓⁄ ) is ignored because it is not dominant in MOS transistor. The noise contribution of 𝑅 is neglected due to the parasitic capacitance in parallel with 𝑅 which makes its noise impact non dominant. The noise analysis is performed based on the noise model shown in Fig.7. The thermal noise in MOS transistor is modeled by a noise current source between the drain and source terminals with spectral noise of [19] mBdn gTki 4 2 ,  , (7) where 𝑘 is the Boltzmann's constant (J/ºK), T is the absolute temperature (ºK) and γ is the complex function of transistor parameters and bias conduction. The equivalent input noise current spectral density can be given as       , 11 2 2, 2 21, 2 11, 22 2 2 22 1 222 , 2 , MnMnMnpd pdpdRninn iiiC CLCLxii      (8) Fig.8. Modified RCG TIA with noise sources. where 𝑖 , ̅̅̅̅̅ is the thermal noise of the resistors given by                                             2 2 2 1 2 2 2 21 2 , 11 1 1 11 4 SS B m B m SS Rn RR R g R g RR kTi  (9) and              Bm gd Bm Rg Cgs C Rg LL x 2 2 2 2 212 11 1  (10)   . 1 14 2,1, 14 2 2 2 2 2, 2 1 1 2 1,                            B m B m Mn im D jm jMn R g R gkT i j g R gkT i   (11) For half circuit of the capacitive degeneration stage, the input noise current spectral density (𝑖 , ̅̅̅ ̅̅ ̅̅ ̅̅ ̅) is given by Fig.7. Schematic of proposed RCG TIA. M1-1 Rs1 M1-2 Rs2 L1 L2 DD M3 Rs3 RD3 L3 M2 Cpd I pd DD RB DD RD1 DD M4 Rs4 Output voltage DD M5 Rs5 RD4 R C G2 CGs2 g m2 D1-2 V gs2 RB CpdI pd RB L2 CGs1-1 g m1-1 D1-1 V gs1-1 RD1 RD1Cgs1-2 g m1-2 D1-2 V gs1-2 RS1 RS2 L1 D2 D1 Jawdat Abu_Taha / Ultra-Wideband and Low Noise Transimpedance Amplifier(2016) 82 4,3, 1 1 1 4 2 , , ,0 2 , 2 , 2 2 , 2 2 ,                                  j R g Rg R g CR CkT i eqsj mi eqsjmj Dj jd eqsjeqsj eqDj HCDEGn    (12) where 𝐶 , and 𝐶 , are the equivalent parasitic capaci- tance at drain and source nodes respectively, 𝑅 , is the equivalent resistance at source node and 𝑔 , is the zero-bias drain conductance. Because the main deliberation is given to the modified RCG input stage and the capacitive degeneration stage, the contribution noise of the buffer is ignored, As shown in (7), the resistors noise is the main noise at low fre- quencies and the impact noise of 𝑀 and 𝑀 becomes domi- nant at high frequencies. Note that the input noise current reduces appreciably at high frequencies using 𝐿 and 𝐿 at the input of the TIA. Furthermore the minimum noise can be realized by boosting the transconductance 𝑔 . VI. SIMULATION RESULTS We performed simulation analysis of the proposed TIA circuit using Cadence tools. Simulations are done by utilizing RF transistor model based on 0.18µm HV CMOS technology with a 1.8-V single supply and a 100-fF photodiode capaci- tance. Fig.9 shows the layout of the proposed TIA with 147µm × 230µm of area cost. The frequency responses of the conventional RCG and the proposed TIAs are presented in Fig.10. The RCG TIA provides a bandwidth of 3.5 GHz, whereas the bandwidth of proposed TIA extends up to 13 GHz. Transimpedance gains of the conventional RCG and proposed TIAs are 47.7 dB and 53.2 dB, respectively. While the total power consumption of the conventional RCG TIA is 5 mW, the proposed TIA consumes only 11 mW. Fig.9. The layout of the proposed TIA Fig.11 shows the simulation results of input noise current spectral densities of the RCG and the proposed TIA. As shown, the proposed TIA has less input referred noise current than the RGC configuration. It shows an average input noise current spectral density below 24pA/√Hz within the band- width. Fig. 12 shows the group-delay variation with frequency. As shown, the proposed TIA provides smaller group-delay varia- tion than the RGC configuration. Fig. 10. Frequency response results of the TIA The TIA has a minimum group delay of 4 ps, increases to 14 ps within the bandwidth of 13 GHz. This small variation means that output signal will not suffer from distortion as RGC TIA. Fig.11. Spectral density of the input noise curren The transient response of the TIA is shown in Fig.13 at differ- ent process corners. The width of the input current pulse is 10 ps with a rise/fall time of 1 ps and the peak-to-peak current is 50 µA. The simulation result shows that, at different process Fig.12. The group delay variation of TIA. Jawdat Abu_Taha / Ultra-Wideband and Low Noise Transimpedance Amplifier(2016) 83 corners, the output swing variations is very small. This de- picts that the transient response of the TIA is fast enough even for small inpt current. Fig.13. Transient response of the TIA at different process corners. Table 1. Performance summery and comparison with the other works using (180nm CMOS) technology Parameter [20] [16] [21] [22] [23] This Work Gain (dB ) 59 53 54 62.3 51 53.2 BW (GHz) 8.6 8 7 9 30.5 13 CPD (fF) 150 250 300 150 50 100 Power (mW) 18 13.5 18.6 108 60.1 11 Noise 𝑝𝐴/√𝐻𝑧 25 18 18 - 55.7 24 Table 1 shows a comparison of the proposed TIA performance with other works. It can be seen that the noise of the proposed TIA is smaller than the other TIA configurations, where the active inductors have been used. In addition, the power con- sumption is comparatively smaller than the other TIA circuits. VII. CONCLUSION The proposed TIA design improves the performance of the RGC TIA. Use of parallel combination of two series resonate circuits with different resonate frequencies improves the bandwidth and minimizes the equivalent input noise current density of RCG TIA. The capacitance degeneration and series inductive peaking networks are used for pole-zero elimina- tion. The proposed design is implemented in a 0.18-µm CMOS process in the presence of a 100fF photodiode capaci- tance. It is observed that the TIA achieves a -3-dB bandwidth at 13GHz and transimpedance gain of 53.2 dB. The input referred noise current spectral density is 24pA/√Hz and the average group-delay variation is 5 ps over the 3-dB bandwidth and the TIA consumes 11 mW from a 1.8 V supply. Simula- tion results show that the TIA displays a broadband flat re- sponse, provides an ultra-low noise performance, and hence it is proficient for applications in optical transceivers. REFERENCES [1] M. Gupta, U. Singh, R. Srivastava, Bandwidth Extension of High Compliance Current Mirror by Using Compensation Methods, Publisher, City, 2014. [2] S. Ravikumar, Circuit architectures for high speed CMOS clock and data recovery circuits, Publisher, City, 2015. [3] P. Sung Min, Y. Hoi-Jun, 1.25-Gb/s regulated cascode CMOS transimpedance amplifier for Gigabit Ethernet applications, Publisher, City, 2004. [4] M.N. Ahmed, Trnsimpedance Amplifier (TIA) Design for 400Gb/s Optical Fiber Communications, in: Dept. of Electrical Eng., Virginia Polytechnic Institute and State Unive., Blacksburg , VA., 2013. [5] R. Behzad, Design of Integrated Circuits for Optical Communications, 2nd Edition, Wiley, 2012 [6] C. Li, S. Palermo, A Low-Power 26-GHz Transformer-Based Regulated Cascode SiGe BiCMOS Transimpedance Amplifier, Publisher, City, 2013. [7] S.S. Mohan, M.D.M. Hershenson, S.P. Boyd, T.H. Lee, Bandwidth extension in CMOS with optimized on-chip inductors, Publisher, City, 2000. [8] E. Sackinger, W.C. Fischer, A 3 GHz, 32 dB CMOS limiting amplifier for SONET OC-48 receivers, in: Solid-State Circuits Conference, 2000. Digest of Technical Papers. ISSCC. 2000 IEEE International, 2000, pp. 158-159. [9] K. Sung-Eun, S. Seong-Jun, P. Sung Min, Y. Hoi-Jun, CMOS optical receiver chipset for gigabit Ethernet applications, in: Circuits and Systems, 2003. ISCAS '03. Proceedings of the 2003 International Symposium on, 2003, pp. I-29-I-32 vol.21. [10] F.C.a.Y. Chan, Bandwidth Enhancement of TIA by a Capacitive- peaking Design, Publisher, City, 1999. [11] K. Chin-Wei, C.-C. Hsiao, Y. Shih-Cheng, Y.-J. Chan, 2 Gbit/s transimpedance amplifier fabricated by 0.35 μm CMOS technologies, Publisher, City, 2001. [12] O. Momeni, H. Hashemi, E. Afshari, A 10-Gb/s Inductorless Transimpedance Amplifier, Publisher, City, 2010. [13] R.H. Mekky, P.V. Cicek, M.N. El-Gamal, Ultra low-power low-noise transimpedance amplifier for MEMS-based reference oscillators, in: Electronics, Circuits, and Systems (ICECS), 2013 IEEE 20th International Conference on, 2013, pp. 345-348. [14] S. Qiwei, M. Luhong, X. Sheng, K. Yuzhuo, Novel pre-equalization transimpedance amplifier for 10 Gb/s optical interconnects, Publisher, City, 2015. [15] J.K.a.J.F. Buckwalter, Bandwidth Enhancement with Low Group- Delay Variation for a 40-Gb/s Transimpedance Amplifier Publisher, City, 2010. [16] L. Zhenghao, Y. Kiat Seng, M. Jianguo, A.V. Do, L. Wei Meng, C. Xueying, Broad-Band Design Techniques for Transimpedance Amplifiers, Publisher, City, 2007. [17] M.S. AL-JUAID, DESIGN OF VARIABLE GAIN TRANSIMPEDANCE PREAMPLIFIER, in, 1998. [18] A.S. Sedra , K. C. Smith, Microelectronic Circuits, Oxford University Press 2014. [19] F. Levinzon, Comparison of 1/f Noise and Thermal Noise in JFETs and MOSFETs, in: Piezoelectric Accelerometers with Integral Electronics, Springer, 2015, pp. 93-106. [20] C.-Y. Wang, C.-S. Wang, C.-K. Wang, An 18-mW two-stage CMOS transimpedance amplifier for 10 Gb/s optical application, in: Solid-State Circuits Conference, 2007. ASSCC'07. IEEE Asian, IEEE, 2007, pp. 412- 415. Jawdat Abu_Taha / Ultra-Wideband and Low Noise Transimpedance Amplifier(2016) 84 [21] Z. Lu, K.S. Yeo, W.M. Lim, M.A. Do, C.C. Boon, Design of a CMOS broadband transimpedance amplifier with active feedback, Publisher, City, 2010. [22] A.K. Petersen, K. Kiziloglu, Y. Ty, F. Williams, Jr., M.R. Sandor, Front-end CMOS chipset for 10 Gb/s communication, in: Radio Frequency Integrated Circuits (RFIC) Symposium, 2002 IEEE, 2002, pp. 93-96. [23] J. Jun-De, S.S.H. Hsu, 40-Gb/s Transimpedance Amplifier in 0.18µm CMOS Technology, Publisher, City, 2008.