MR Interface Analysis including Chord Signaling Options

Similar documents
Brian Holden Kandou Bus, S.A. IEEE GE Study Group September 2, 2013 York, United Kingdom

Summary of NRZ CDAUI proposals

Draft Baseline Proposal for CDAUI-8 Chipto-Module (C2M) Electrical Interface (NRZ)

100Gb/s Single-lane SERDES Discussion. Phil Sun, Credo Semiconductor IEEE New Ethernet Applications Ad Hoc May 24, 2017

Ali Ghiasi. Nov 8, 2011 IEEE GNGOPTX Study Group Atlanta

System Evolution with 100G Serial IO

Performance comparison study for Rx vs Tx based equalization for C2M links

Presentation to IEEE P802.3ap Backplane Ethernet Task Force July 2004 Working Session

Practical Receiver Equalization Tradeoffs Applicable to Next- Generation 28 Gb/s Links with db Loss Channels

LOW POWER DIGITAL EQUALIZATION FOR HIGH SPEED SERDES. Masum Hossain University of Alberta

64G Fibre Channel strawman update. 6 th Dec 2016, rv1 Jonathan King, Finisar

100G PSM4 & RS(528, 514, 7, 10) FEC. John Petrilla: Avago Technologies September 2012

PAM4 signals for 400 Gbps: acquisition for measurement and signal processing

Application Space of CAUI-4/ OIF-VSR and cppi-4

CAUI-4 Chip to Chip Simulations

The Case of the Closing Eyes: Is PAM the Answer? Is NRZ dead?

CAUI-4 Chip to Chip and Chip to Module Applications

Measurements Results of GBd VCSEL Over OM3 with and without Equalization

M809256PA OIF-CEI CEI-56G Pre-Compliance Receiver Test Application

Exceeding the Limits of Binary Data Transmission on Printed Circuit Boards by Multilevel Signaling

Ali Ghiasi. Jan 23, 2011 IEEE GNGOPTX Study Group Newport Beach

PAM-2 on a 1 Meter Backplane Channel

Comparison of NRZ, PR-2, and PR-4 signaling. Qasim Chaudry Adam Healey Greg Sheets

Comment #147, #169: Problems of high DFE coefficients

100G CWDM Link Model for DM DFB Lasers. John Petrilla: Avago Technologies May 2013

BER margin of COM 3dB

New Serial Link Simulation Process, 6 Gbps SAS Case Study

New Techniques for Designing and Analyzing Multi-GigaHertz Serial Links

Combating Closed Eyes Design & Measurement of Pre-Emphasis and Equalization for Lossy Channels

Combating Closed Eyes Design & Measurement of Pre-Emphasis and Equalization for Lossy Channels

Measurements and Simulation Results in Support of IEEE 802.3bj Objective

Further Investigation of Bit Multiplexing in 400GbE PMA

Approach For Supporting Legacy Channels Per IEEE 802.3bj Objective

Clause 74 FEC and MLD Interactions. Magesh Valliappan Broadcom Mark Gustlin - Cisco

50 Gb/s per lane MMF objectives. IEEE 50G & NGOATH Study Group January 2016, Atlanta, GA Jonathan King, Finisar

Validation of VSR Module to Host link

C65SPACE-HSSL Gbps multi-rate, multi-lane, SerDes macro IP. Description. Features

100GBASE-SR4 Extinction Ratio Requirement. John Petrilla: Avago Technologies September 2013

100G EDR and QSFP+ Cable Test Solutions

DataCom: Practical PAM4 Test Methods for Electrical CDAUI8/VSR-PAM4, Optical 400G-BASE LR8/FR8/DR4

32 G/64 Gbaud Multi Channel PAM4 BERT

SMF Ad Hoc report. Pete Anslow, Ciena, SMF Ad Hoc Chair. IEEE P802.3bm, Geneva, September 2012

52Gb/s Chip to Module Channels using zqsfp+ Mike Dudek QLogic Barrett Bartell Qlogic Tom Palkert Molex Scott Sommers Molex 10/23/2014

SECQ Test Method and Calibration Improvements

Line Signaling and FEC Performance Comparison for 25Gb/s 100GbE IEEE Gb/s Backplane and Cable Task Force Chicago, September 2011

500 m SMF Objective Baseline Proposal

PAM8 Baseline Proposal

Half-Rate Decision-Feedback Equalization Di-Bit Response Analysis and Evaluation EDA365

Simulations of Duobinary and NRZ Over Selected IEEE Channels (Including Jitter and Crosstalk)

100G SR4 Link Model Update & TDP. John Petrilla: Avago Technologies January 2013

AMI Modeling Methodology and Measurement Correlation of a 6.25Gb/s Link

More Insights of IEEE 802.3ck Baseline Reference Receivers

The Challenges of Measuring PAM4 Signals

Next Generation Ultra-High speed standards measurements of Optical and Electrical signals

Performance Modeling and Noise Reduction in VLSI Packaging

40GBASE-ER4 optical budget

100G MMF 20m & 100m Link Model Comparison. John Petrilla: Avago Technologies March 2013

Draft 100G SR4 TxVEC - TDP Update. John Petrilla: Avago Technologies February 2014

CDAUI-8 Chip-to-Module (C2M) System Analysis #3. Ben Smith and Stephane Dallaire, Inphi Corporation IEEE 802.3bs, Bonita Springs, September 2015

CAUI-4 Application Requirements

PAM8 Gearbox issues Andre Szczepanek. PAM8 gearbox issues 1

A Way to Evaluate post-fec BER based on IBIS-AMI Model

Further information on PAM4 error performance and power budget considerations

Development of an oscilloscope based TDP metric

On Figure of Merit in PAM4 Optical Transmitter Evaluation, Particularly TDECQ

50 Gb/s per lane MMF baseline proposals. P802.3cd, Whistler, BC 21 st May 2016 Jonathan King, Finisar Jonathan Ingham, FIT

Problems of high DFE coefficients

Practical De-embedding for Gigabit fixture. Ben Chia Senior Signal Integrity Consultant 5/17/2011

Powering Collaboration and Innovation in the Simulation Design Flow Agilent EEsof Design Forum 2010

Open electrical issues. Piers Dawe Mellanox

Duobinary Transmission over ATCA Backplanes

Technical Article MS-2714

Systematic Tx Eye Mask Definition. John Petrilla, Avago Technologies March 2009

N4917BACA Optical Receiver Stress Test Solution 100 Gb/s Ethernet

CU4HDD Backplane Channel Analysis

AMI Simulation with Error Correction to Enhance BER

100 Gb/s per Lane for Electrical Interfaces and PHYs CFI Consensus Building. CFI Target: IEEE November 2017 Plenary

100GBASE-DR2: A Baseline Proposal for the 100G 500m Two Lane Objective. Brian Welch (Luxtera)

XLAUI/CAUI Electrical Specifications

High Speed Serdes Devices and Applications

40GBd QSFP+ SR4 Transceiver

COM Study for db Channels of CAUI-4 Chip-to-Chip Link

Analyzing GBaud PAM4 Optical and Electrical Signals APPLICATION NOTE

New Technologies for 6 Gbps Serial Link Design & Simulation, a Case Study

Issues for fair comparison of PAM4 and DMT

40G SWDM4 MSA Technical Specifications Optical Specifications

DesignCon Pavel Zivny, Tektronix, Inc. (503)

Optical transmission feasibility for 400GbE extended reach PMD. Yoshiaki Sone NTT IEEE802.3 Industry Connections NG-ECDC Ad hoc, Whistler, May 2016

Comparison of options for 40 Gb/s PMD for 10 km duplex SMF and recommendations

PRE-QSFP-LR4L 100G QSFP 28 Dual Range Optical Transceiver, 10km. Product Features: General Product Description:

A 90 Gb/s 2:1 Multiplexer with 1 Tap FFE in SiGe Technology

Component BW requirement of 56Gbaud Modulations for 400GbE 2 & 10km PMD

Thoughts on 25G cable/host configurations. Mike Dudek QLogic. 11/18/14 Presented to 25GE architecture ad hoc 11/19/14.

10Gb/s SFP+ ER 1550nm Cooled EML with TEC, PIN Receiver 40km transmission distance

400GbE AMs and PAM4 test pattern characteristics

ELECTRICAL PERFORMANCE REPORT

Development of an oscilloscope based TDP metric

10GBASE-LRM Interoperability & Technical Feasibility Report

Thoughts about adaptive transmitter FFE for 802.3ck Chip-to-Module. Adee Ran, Intel Phil Sun, Credo Adam Healey, Broadcom

10GBASE-R Test Patterns

Transcription:

MR Interface Analysis including Chord Signaling Options David R Stauffer Margaret Wang Johnston Andy Stewart Amin Shokrollahi Kandou Bus SA May 12, 2014 Kandou Bus, S.A 1

Contribution Number: OIF2014.113 Working Group: Physical & Link Layer WG (Electrical Track) Title: MR Interface Analysis including Chord Signaling Options Source: David R Stauffer david@kandou.com Date: May 12, 2014 Abstract: This analysis compares signal integrity and power analysis results for various Chord Signaling codes in CEI-56G Medium Reach (MR) channel applications. Codes are compared to an NRZ baseline. Notice: This contribution has been created to assist the Optical Internetworking Forum (OIF). This document is offered to the OIF solely as a basis for discussion and is not a binding proposal on the companies listed as resources above. Each company in the source list, and the OIF, reserves the rights to at any time to add, amend, or withdraw statements contained herein. This Working Text represents work in progress by the OIF, and must not be construed as an official OIF Technical Report. Nothing in this document is in any way binding on the OIF or any of its members. The document is offered as a basis for discussion and communication, both within and without the OIF. For additional information contact: The Optical Internetworking Forum, 48377 Fremont Blvd., Suite 117, Fremont, CA 94538 510-492-4040 phone info@oiforum.com Kandou Bus, S.A 2

Disclosure discloses that we own intellectual property related to Chord Signaling and other material described in this contribution. We are committed to RAND licensing of all of our technologies. We are committed to adhering to the bylaws of all standards organizations to which we contribute and maintain membership. We are committed to be good corporate citizens. Kandou Bus, S.A 3

CEI Medium Reach Application Application is defined as 500 mm reach one connector CEI-28G-MR Insertion Loss: approx. 20 db @14 GHz. 4

Channel Requirements Project Start states requirement that max loss is in the range of: 15 to 25 db at 14 GHz 20 to 50 db at 28 GHz This allows the project to pick a wide range of possible insertion loss equations; some are worse than current CEI- 28G-MR and CEI-28G-LR channels. This contribution: Projects three possible equations for channel insertion loss; Compares signaling options (NRZ, PAM-4, ENRZ) for each of these channel options; and Provides power analysis comparing NRZ and ENRZ signaling. Kandou Bus, S.A 5

Basic CEI Insertion Loss Equation: CEI Channel Equation IIIIIIIIII = CEI-28G-MR: ccc + ccc ff ff bbbbbbbb ff bb ccc + ccc ff ff bbbbbbbb ff bb + ccc ff ff bbbbbbbb ff bb ff mmmmmm ff < ff bb 2 ff bb 2 ff ff bb f bmax = 28.1, c1 = 1.083, c2 = 2.436, c3 = 0.698, c4 = -17.83, c5 = 2.694 CEI-28G-LR: f bmax = 25.8, c1 = 1.083, c2 = 3.35, c3 = 0.96, c4 = -9.25, c5 = 2.694 Kandou Bus, S.A 6

Channel Options Channel options generated using: f bmax = 56.2 c1 = 1.083 (same as CEI-28G-LR/MR) c2 = adjusted to produce desired loss c3 = c1 / 3.4895 (to match CEI-28G-LR/MR slope) c4 = calculated to avoid discontinuity at f b /2 c5 = 2.694 (same as CEI-28G-LR/MR) Channel options: Option 3: Extends CEI-28G-MR channel (c2 = 2.436) Option 2: Extends CEI-28G-LR channel (c2 = 3.35) Option 1: Set for 50 db loss @28 GHz (c2 = 3.663) Kandou Bus, S.A 7

Channel Options 8

Channel Models and Simulation Conditions SDD21 for NRZ / PAM-4 shown in top figure. SDD21 for ENRZ including all subchannels in bottom figure. Channel Models were constructed by concatenating: Package Model (5.4 mm) Daughter card PCB (5 inches) o FR4 (er=3.7, tand=0.019) Backplane connector (FCI) PCB trace (8, 13, or 15 inches) o FR4 (er=3.7, tand=0.019) Package Model (5.4 mm) Simulation Conditions: Tx Launch: 1000 mvppd 3-tap FFE, CTLE, 20-tap DFE No random jitter. Pass/Fail: 25 mvppd, 0.25 UI eye BER 1E-15 w/o FEC or 1E-6 w/fec. 9

Channel #1-49.8 db @28.1 GHz Channel #2-45.1 db @28.1 GHz Channel #3-33.5 db @28.1 GHz KEYE Results - NRZ Simulation conditions: NRZ @56 GBd Tx Launch: 1000 mvppd FFE (3-tap) CTLE (0 to 12 db) DFE (20-tap) BER = 1E-6 (assumes FEC) Results: Pass/fail criteria: o o > 25 mvppd eye height > 0.25 UI eye width Case 3 eye is open. Case 1 & 2 eyes are closed. 10

Channel #1-27.1 db @14.05 GHz Channel #2-24.6 db @14.05 GHz Channel #3-18.2 db @14.05 GHz KEYE Results PAM4 Simulation conditions: PAM-4 @28 GBd Tx Launch: 1000 mvppd FFE (3-tap) CTLE (0 to 12 db) DFE (20-tap) BER = 1E-6 (assumes FEC) Results: Pass/fail criteria: o o > 25 mvppd eye height > 0.25 UI eye width Case 3 eyes are open. Case 1 & 2 eyes are closed. 11

Channel #1-34.6 db @18.73 GHz Channel #2-31.4 db @18.73 GHz Channel #3-18.73 db @18.73 GHz KEYE Results ENRZ Simulation conditions: ENRZ @37 GBd Tx Launch: 1000 mvppd FFE (3-tap) CTLE (0 to 12 db) DFE (20-tap) BER = 1E-6 (assumes FEC) Results: Pass/fail criteria: o o > 25 mvppd eye height > 0.25 UI eye width Case 1 & 2 & 3 eyes are open. Sub-channel #2 eyes are shown at left (worst case). Sub-channel #1, #3 eyes are slightly better. 12

KEYE Results ENRZ w/o FEC SC#0 SC#1 Channel #3, 1e-15 SC#2 Simulation conditions: ENRZ @37 GBd Tx Launch: 1000 mvppd FFE (3-tap) CTLE (0 to 12 db) DFE (20-tap) BER = 1E-15 (no FEC) Results: Pass/fail criteria: o > 25 mvppd eye height o > 0.25 UI eye width Case 3 eyes are open. ENRZ will work w/o FEC on this channel. 13

KEYE Results - Detail Channel #1: Pkg (5.4mm) + 5in PCB + FCI conn. + 15in PCB + Pkg (5.4mm) Channel #2: Pkg (5.4mm) + 5in PCB + FCI conn. + 13in PCB + Pkg (5.4mm) Channel #3: Pkg (5.4mm) + 5in PCB + FCI conn. + 8in PCB + Pkg (5.4mm) 14

KEYE Results - Observations Results are consistent with prior published data: NRZ has open eyes for channel losses up to 30dB. NRZ operation can be extended up to 36 db using FEC. PAM-4 simulations tend to show similar results to NRZ for equivalent channels and signal processing. o Advantage of lower channel loss at lower baud rate is offset by lower launch amplitude. ENRZ performance tends to be better because: ENRZ takes advantage of correlation across 4 wires (not just 2 wires). Baud rate is lower than NRZ; less channel loss. ENRZ eye height is twice that of PAM-4. Kandou Bus, S.A 15

How Does Coding Save Power? Chord codes have the potential to save power: Less clocking power: CDR can be shared across wires. This increases number of transitions available, reducing run length constraints on the CDR, and amortizing circuit size/power across more than one bit stream. Lower equalization power: o Lower baud rate = less attenuation = less equalization required. o Better ISI tolerance can be achieved through proper code design. Lower Driver Power: Code design can reduce termination line power. This contribution analyzes ENRZ power consumption for a reference architecture relative to an NRZ implementation using an equivalent architecture. Kandou Bus, S.A 16

Power Analysis Methodology Purpose: Benchmark power of an ENRZ reference design to an equivalent NRZ reference design. Methodology applied: Kandou Wasp chip used as reference design for Serdes circuit and logic blocks (TSMC 28 nm, 28 GBd). Spice simulations used to determine power for circuit blocks of the reference design. Logic block power estimated based on synthesis results. Determine rules for scaling each block to other frequencies. Scale baseline power of each block to TSMC 16 nm process. Equivalent circuit architecture assumptions are used for all codes at all baud rates. (This avoids biasing results with architecture choices.) Benchmark: NRZ is used as a benchmark for the analysis methodology. This can be compared to power of existing examples of CEI-25G-LR Serdes products. Kandou Bus, S.A 17

NRZ Power Estimate (CEI-25G-LR Baseline) Baseline configuration: 112 Gb/s interface NRZ signaling 4 lanes @28 GBd CEI-25G-LR compliant TSMC 28 nm process DFE: 12-tap Typical power: 11.71 pj/bit Calculation is comparable to examples of CEI-25G-LR Serdes known to the author. CDR, 95.60 Rx Analog, 360.80 Power Breakdown (mw) DFE, 174.00 Tx Analog, 318.80 Clock Dist., 144.60 Digital Logic, 168.80 Power Total (mw) 1310.00 Data Throughput (Gb/s) 112.00 Energy per Bit (pj/bit) 11.71 PLL, 27.40 Termination, 20.00 18

NRZ Power Estimate (CEI-56G-MR) Baseline configuration: 112 Gb/s interface NRZ - 2 lanes @56 Gb/s TSMC 16 nm process DFE: 20 tap FEC required but not included in power budget. Typical power: 9.88 pj/bit Analysis shows 16% power reduction for next generation. Power advantage of next process node is offset by higher baud rate. Power is not scaling by 30% as has been the case for prior generations of NRZ Serdes. Industry requires larger power reductions than past generations. Rx Analog, 286.72 Power Breakdown (mw) CDR, 76.48 DFE, 157.97 Tx Analog, 268.80 Clock Dist., 130.10 Digital Logic, 141.54 Power Total (mw) 1105.15 Data Throughput (Gb/s) 112.00 Energy per Bit (pj/bit) 9.88 PLL, 33.54 Termination, 10.00 19

ENRZ Power Estimate (CEI-56G-MR) Baseline configuration: 112 Gb/s interface ENRZ - 1 lane @37 Gb/s TSMC 16 nm process DFE: 20 tap FEC not included requirement depends on baseline. Typical power: 6.79 pj/bit Analysis shows 42% power reduction for next generation. Baud rate scales more in line with process transistor speeds. Power reduction is exceeds 30% scaling typical of prior generations. CDR, 75.80 Rx Analog, 152.56 Power Breakdown (mw) DFE, 156.56 Clock Dist., 77.50 Tx Analog, 130.80 Digital Logic, 134.58 Power Total (mw) 759.14 Data Throughput (Gb/s) 112.00 Energy per Bit (pj/bit) 6.79 PLL, 26.34 Termination, 5.00 20

Summary Three channels were analyzed with various insertion losses consistent with the ranges in the project start. Only the best of these channels had open eyes for NRZ or PAM-4 signaling simulations. o FEC was required in both cases. ENRZ eyes were open for all channels. o FEC was only required for the two higher loss channels. NRZ and ENRZ power was analyzed: NRZ power is not scaling at the same rate as prior generations. ENRZ offers significant power savings over an NRZ implementation. Target of 6.79 pj/bit is projected for ENRZ. Analysis demonstrates advantage of ENRZ over NRZ: CEI-56G-MR channel can be defined for a higher insertion loss than can be supported for other signaling methods. Substantial power savings are possible. Kandou Bus, S.A 21

KANDOU reinventing the BUS Kandou Bus, S.A 22