+ All Categories
Home > Documents > Short/Mid-Reach 10SPE PHY with Multipoint Claude Gauthier ...

Short/Mid-Reach 10SPE PHY with Multipoint Claude Gauthier ...

Date post: 15-Mar-2022
Category:
Upload: others
View: 5 times
Download: 0 times
Share this document with a friend
14
Short/Mid-Reach 10SPE PHY with Multipoint Claude Gauthier, OmniPHY
Transcript

Short/Mid-Reach10SPEPHYwithMultipoint

ClaudeGauthier,OmniPHY

A15/40mPHYCandidate• Ashort-reach(15/40m,4conn)forpoint-pointinterface– Goal- a50%reductioninpower,area,pinsrelativeto100BT1

July2017 2zimmerman_3NEA_01_061417.pdf

10Mb/sSPEPHYStrawman

• Simplifythesignalingtobinarylevels• DME(DifferentialManchesterEncoding)• TX- ClassABtransmitter@20MHz• RX- Analogequalizer,comparatorsforreceive,• RX- LargelyeliminateDSP,exceptforfloatingEQ

10Mb/sSPEPHYPackageDesign• Canthepackagefollowthesiliconareareduction?

• A100BT1discrete PHYmayhave36pins• Powersupplypinswillscalesignificantly• MIIis16pins,asignificantpercentageofthetotal

• Canreducethisto4pinswithoutchangingtheMACinterface:“xMII”• cordaro_thaler_10SPE_01a_0916.pdf

MLT3

• A14-pinpackageisreasonableforthisprotocol

-75%

10Mb/sSPEPHYDesign• Area<~25-30%of100BT1area

• This isapproximatelytheareaof100Base-TX

• PowerDissipation<~33%of100BT1• Powerreductionsupportsreductionofpackagepins• Importantgiventhenumberof10Mb/snodes inthenetwork

• Low-powerdriveslowpin-countpackages

Conclusion: Yes, it’spossible toreducesiliconarea/powerby>50%

WhatAboutMultipoint?• Increasessignal-integritychallengerelativetothebaseline,butitcanbemanaged

• Needtofixinsertionloss,delay,numberofnodes– Somemodelsexistforautomotive,needtoexpandthelistforotherusecases(i.e.cabinets,buildinginfrastructure)

• Biggerchallengeisaddressingtheimplicationsof“sharing”versuspoint-point

July2017 6

matheus_10SPE_01_0417.pdf

WhatAboutMultipoint• Forthissimple,traditionalPHYcanweaddcircuitrytoimplementCMAS/CD?– Showspromise,leveragesexistingPHYtopologieswithsomeadditionalmodifications

– Costsnetworkperformanceandplaceslimitationsonphysicalconstraints

• Alternative- TDMA-based– CanalsoleveragetraditionalPHYarchitectures,buttheoverallsolutioniscompromised:• Inefficient allocationofresources• Allocatetime-slots fornodesANDallocatetime-slots inbetween• Fixednumberofnodes, time-syncchallengesgivenvariablespacing

July2017 7

HowAboutFDMA?• Thefrequencydomainisacleanwayto“share.”– Everyonecantalkatthesametime,nodesgothroughaset-upandareallocatedafixed-channel

• Betternetworkperformance,opensthedoortolowerpower(analogimplementation),butit’sdifferent/new

July2017 8

FDMA-Conceptual• Leveragessimplelow-power/low-costmultichannelwirelesstopologies

July2017 9

– FrequencyAllocation• Firstconsider4channels• Usealltheavailablebandwidth (i.e.100MHz?)andallocateindividual10MHzchannels– ORuse36MHz(“sweetspot bandwidth”)andrelyonencodingtoget10Mb/sfrom4MHzchannels

ParallelHead-EndTransmitterConceptual

10

+xejω3t

Channel3modulation

xejω4t

Channel4modulation

xejω1t

Channel1modulation

xejω2t

Channel2modulation

Headendouttolineinterface

ParallelHeadEndReceiverConceptual

11

+xe-jω7t

Channel7baseband

xe-jω8t

Channel8baseband

xe-jω5t

Channel5baseband

xe-jω6t

Channel6baseband

Headendouttolineinterface

Single-NodeTransmitter/ReceiverConceptual

12

+

xe-jω1t

Channel1baseband

Outtolineinterface

xejω5t

Channel5modulation

MoreInformationNeeded• Auto-negotiationandchannelassignment• Link-Segment– Channelconditionsfornon-automotivemid-reachmulti-dropcases

– Fixonmax/minconditions– Connectorcharacteristics

• Powerestimationandcomplexityanalysisforbothtopologies

July2017 13

Thank-You!

14


Recommended