Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Equalization/Compensation of Transmission Media Channel (copper or fiber) EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 1 Optical Receiver Block Diagram OE TIA ≈ -18 dBm ≈ 10 µA EECS 270C / Winter 2016 LA ≈ 10 mV p-p EQ CDR DMUX ≈ 400 mV p-p Prof. M. Green / UC Irvine 2 Copper Cable Model 4-foot cable Copper Cable ( ) H w » e -La w 15-foot cable Where: L is the cable length a is a cable-dependent characteristic EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 2 Effect of Copper on Broadband Data waveform EECS 270C / Winter 2016 eye diagram Prof. M. Green / UC Irvine 3 Adaptive Analog Equalizer for Copper Implemented in Jazz Semiconductor SiGe BiCMOS process: • 120 GHz fT npn • 0.35 µm CMOS EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 4 Equalizer Block Diagram EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 5 Analog Equalizer Concept (1) Simple linear circuit (normalized to 1Hz): 1 V2 V1 V3 +0.5 -0.5 C1 1s simple channel model EECS 270C / Winter 2016 1 1 1 1×V1 bandpass filter Prof. M. Green / UC Irvine a ×V1 ( ) 1- a ×V2 1 combined flat response + peaked response 6 Analog Equalizer Concept (2) 1 V2 V1 V3 +0.5 -0.5 1s V1 EECS 270C / Winter 2016 C1 1 1 1×V1 1 a ×V1 (1- a) ×V 1 2 V2 Prof. M. Green / UC Irvine 7 Analog Equalizer Concept (3) Equalized output pulses: Rise time = voltage swing/slew rate V3 Rise time nearly constant over different channels! EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 8 Feedforward Path Vout EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 9 Equalizer Frequency Response Veq Vin (dB) Vcontrol f (Hz) EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 10 ISI & Transition Time VFFE teq = 45ps PW = 108ps 0.3 teq = 60ps PW = 100ps 0 -0.3 2.4 teq = 75ps PW = 86ps 2.5 2.6 2.7 2.8 t (ns) • Simulations indicate that ISI correlates strongly with FFE transition time teq. • Optimum teq is observed to be 60 ps. • Nonlinearities affect pulse shape, but not location of zero crossings. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 11 Slicer Restores full logic levels EECS 270C / Winter 2016 Exhibits controlled transition time Prof. M. Green / UC Irvine 12 Feedback Path ò EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 13 Transition Time Detector DC characteristic: V+ VS VVS Transient Characteristic: ISS V+ -V- V+ -V- CSS (b) (a) VS Rectification & filtering done in a single stage. (b) (a) EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 14 t Integrator A0 1 H(s) = » 1+ st int A0 st int ( A0 = gm 1 ro1 || ro2 t int = EECS 270C / Winter 2016 Prof. M. Green / UC Irvine ) CL gm 1 15 Detector + Integrator From FFE tFFE From Slicer tslicer= 60ps FFE transition Time tFFE Vcontrol (mV) 90ps 60 slope detector slope detector 40 75ps 20 60ps 0 ò -20 -40 45ps -60 15ps 0 10 _ + Vcontrol EECS 270C / Winter 2016 20 30 40 50 t (ns) Prof. M. Green / UC Irvine 16 System Analysis tslicer detector + Kd å Vcontrol _ integrator feedforward equalizer H(s) Keq teq detector Kd t eq t slicer = H (s ) » Keq = 1.5 ps/mV Kd = 2.5 mV/ps K d K eq H (s ) 1 + K d K eq H (s ) 1 st int EECS 270C / Winter 2016 t eq t slicer = 1 1+ s t int K d K eq Prof. M. Green / UC Irvine int = 75ns tadapt = tint = 20ns Kd Keq 17 Measurement Setup EQ inputs Die under test 231 PRBS signal applied to cable EECS 270C / Winter 2016 EQ outputs Prof. M. Green / UC Irvine 18 Measured Eye Diagrams EQ input EQ output 4-foot RU256 cable (-5 dB atten. @ 5 GHz) 4.0 ps rms jitter 15-foot RU256 cable (-15 dB atten. @ 5 GHz) 3.9 ps rms jitter EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 19 Summary of Measured Performance Supply voltage 3.3 V Power Dissipation 350 mW (155 mW not including output driver) Die Size 0.81mm X 0.87mm Output Swing 490 mV single-ended p-p Random Jitter 4.0 ps rms (4-foot cable) 3.9 ps rms (15-foot cable) Presented at ISSCC Feb. 2004 EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 20 Equalization vs. Compensation Equalization is accomplished by inverting the transfer function of the channel. Compensation is accomplished only by canceling the ISI at each unit interval. Electronic Dispersion Compensation (EDC) refers to the electronics that accomplishes compensation of copper or optical transmission media. EDC is becoming especially critical as bit rates increase on legacy equipment (e.g., backplane, optical connectors, optical fiber). EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 21 Pre-Cursor/Post-Cursor ISI Input pulse (no ISI): 0 T cursor pre-cursor ISI post-cursor ISI Output pulse: 0 T EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 22 Feedforward Equalization (FFE) Idea: To cancel ISI, subtract a weighted & delayed version of the pulse: output pulse delayed by T: d-1 d0 output pulse: æ d ö Xç- -1 ÷ è d0 ø Result with 0 pre-cursor ISI: EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 23 Feedforward Equalization (2) Din (t) T Din (t -T) Time domain: Dout (t) = Din (t) - a1 ×Din (t -T ) a1 a1 = _ + d-1 d0 Dou t (t) 1+ a12 Frequency domain: H(s) = 1- a1 × e -sT H( jw ) = 1- a1 × e - jwT ( = 1- a1 × cos wT - j sin wT 1- a1 ) Þ| H( jw ) |2 = (1+ a12 ) - 2a1 ×cos wT EECS 270C / Winter 2016 p Prof. M. Green / UC Irvine T 24 Feedforward Equalization (3) N-tap FFE structure: Din (t) T a0 T a1 T a2 an Dou t (t) FFE can cancel both pre- and post-cursor distortion. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 25 Feedforward Equalization (4) 3-tap summing circuit: negative coefficient R _ R Vout + Din+ (k) Din- (k) V0 Din+ (k -1) Din- (k -1) Din+ (k - 2) V1 Din- (k - 2) V2 ISS Coefficients set by gm of each differential pair. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 26 Feedforward Equalization (5) Fractional spacing: p T 1-tap T-spaced FFE frequency response 2p T 1-tap T/2-spaced FFE frequency response EECS 270C / Winter 2016 5-tap T-spaced FFE eye diagram 5-tap T/2-spaced FFE eye diagram Prof. M. Green / UC Irvine 27 Adaptation (1) Assume original sequence Din(k) is known. Define error signal e(k) as: ^ ^ e(k) º Dout (k) - Dout (k) where Dout(k) is an appropriately delayed version of Din(k). Steepest Descent Algorithm: e2 a2 step size a1 optimum setting • Algorithm moves coefficients in direction of decreasing mean-square error. • Step size µ should be made sufficiently small to guarantee convergence. • Requires knowledge of properties of mean-square error; usually not available. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 28 Adaptation (2) Least mean-square (LMS) algorithm: FFE output signal: Dout (k) = a0 ×Din (k) + a1 ×Din (k -1) + ×××+ an ×Din (k - n) e 2 (k) = [Dout (k) - Dout (k)] ^ [ ] = -2 D^ d e 2 (k) dai ( out ) - Dout × 2 dDout dai = -2 × e(k) ×Din (k - i) Analog version of LMS: 1 ai (t) = e(t) ×Din (t - iT )dt t both signals are available on chip. ò EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 29 Adaptation (3) Types of adaptation: 1. Training Sequence A training sequence with known properties is sent through the channel + equalizer. The equalizer output is compared to the original sequence and an error signal is generated. 2. Blind Adapation Adaptation is continually performed while system is running. Only limited properties of the signal are known. An error signal must somehow be generated without having the original sequence. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 30 Adaptation (4) Generation of error signal: Din FFE ^ Dout Dout _ + e • Slicer restores logic levels and opens eye vertically. • Bit sequences at slicer input & input are identical. • Slicer has no effect on placement of zero crossing. • Slicer can be realized using CML buffers with sufficient gain and speed. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 31 Decision Feedback Equalization (DFE) Din (t) T T T FFE structure: a0 a1 a2 an Dou t (t) Noise applied to FFE input will be retained (perhaps filtered) at the output. Din (t) DFE structure: Dou t (t) + - - bm b2 T EECS 270C / Winter 2016 b1 T Prof. M. Green / UC Irvine T 32 Decision Feedback Equalization (2) Din (t) + Dou t (t) - - - bm b2 T b1 T T • Slicer is embedded in the structure; Dout is a digital signal. • Delay elements are digital -- commonly realized by DFFs. • Use of slicer suppresses input noise. • Cancels post-cursor distortion only. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 33 Decision Feedback Equalization (3) Din (t) Dou t (t) + - - post-cursor distortion 1-tap example: 2/3 Din (k) 1/3 1 bm b2 Dout (k) (desired) consistent with b1 1 T T T • Tap weights provide a “look-up table,” canceling post-cursor distortion based on last m bits of output sequence. • DFE can sometimes “latch up” with wrong tap weights during adaptation. Dout (k -1) 2/3 Din (k) - 1 ×Dout (k -1) 3 b1 = EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 1 3 34 FFE + DFE Din (t) a0 T T a1 T a2 an Combined FFE and DFE can be used to cancel both pre- and post-cursor distortion with low noise. Dou t (t) + - - bm b2 T EECS 270C / Winter 2016 Prof. M. Green / UC Irvine b1 T T 35 Front-End Circuits for DSP-Based Receivers from channel Vin PGA VA ADC Dout [1:n] ADC requires strict control over its input amplitude VA. VC AGC Automatic Gain Control Programmable Gain Amplifier (PGA): VA (t) = G(VC ) ×Vin (t) æV ö where G(VC ) = V1 × expç C ÷ è V2 ø EECS 270C / Winter 2016 “Linear in dB” gain characteristic gives settling time independent of input amplitude. Prof. M. Green / UC Irvine 36 PGA Design 1. Differential Pair: Iout- 2. Source Degeneration: Iout+ Vin+ Vin- Iout- Iout+ Vin+ Vin- 3. Op-Amp with Feedback: Rf RS + V_in RS + V_out 2RS ISS Rf + VC_ For biasing in weak inversion: ISS 2nVT æ V -V ö = ID0 × expç C T ÷ è nVT ø Iout = gmVin » ISS Iout = » gm ×Vin 1+ gmRS Vin RS Vout = - Rf ×Vin RS for gmRS >> 1 RS varied with constant dB per step. EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 37 PGA Example (1) C.-C. Hsu, J.-T. Wu, “A highly linear 125-MHz CMOS switched-resistor programmable-gain amplifier,” JSSC, Oct. 2003, pp. 1663-1670. Vout = Rf ×Vin RS Realization of RS: 2 dB steps EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 38 PGA Example (2) J. Cao, et al., “A 500mW digitally calibrated AFE in 65nm CMOS for 10Gb/s links over backplane and multimode fiber,” ISSCC 2009, pp. 370-371. Av = gmR = N × mnCox W IO × R L gain of single diff. pair where N = number of diff. pairs turned on EECS 270C / Winter 2016 Prof. M. Green / UC Irvine 39