FlexRay Protocol and FlexRay Protocol and ArchitecturesArchitectures

Chris Quigley

WarwickWarwick ControlControl TechnologiesTechnologies

Presentation OverviewPresentation Overview

AimAim: Give an overview of FlexRay protocol and network : Give an overview of FlexRay protocol and network architecturesarchitectures

1.1. FlexRay propertiesFlexRay properties

2.2. FlexRay FrameFlexRay Frame

3.3. FlexRay TimingFlexRay Timing

4.4. FlexRay ConsortiaFlexRay Consortia

BMW FlexRay Business CaseBMW FlexRay Business Case•• Recognised that NEW technology with HIGHER performance will Recognised that NEW technology with HIGHER performance will

result in higher nodal costsresult in higher nodal costs

•• Nodal costs can be reduced in future by:Nodal costs can be reduced in future by:--

1.1. Using microcontroller with integrated FlexRay controllerUsing microcontroller with integrated FlexRay controller

2.2. Using a low cost transceiverUsing a low cost transceiver

•• However on a system wide basis, cost savings include:However on a system wide basis, cost savings include:--

1.1. Replacing several CAN buses and their cablesReplacing several CAN buses and their cables

2.2. Reducing number of gatewaysReducing number of gateways

3.3. Reduction of design effort for system partitioningReduction of design effort for system partitioning

Network Architecture of FutureNetwork Architecture of Future-- Many proposed uses of FlexRayMany proposed uses of FlexRay

FlexRayFlexRay

High speed backboneHigh speed backbone

XX--byby--WireWire

Airbag deploymentAirbag deployment

LIN Sub BusLIN Sub Bus::

DoorsDoors

Seats etc.Seats etc.

CAN/TTCANCAN/TTCAN ––Applications:Applications:

Powertrain/bodyPowertrain/body

TTCAN deterministic TTCAN deterministic powertrainpowertrain

MOSTMOSTInfotainmentInfotainment

FlexRay Main Features (now 2.1)FlexRay Main Features (now 2.1)• Multi-Master

• Bit rate:-• Maximum Bit rate of 10Mbit/s

• Network Topologies:-• Star, Bus, Mixed

• Physical Layer:-• Twisted pair - electrical

• Safety:-• Dual Channel redundancy

• Bus Guardian for Babbling Idiot avoidance

FlexRay Frame FormatFlexRay Frame Format

FlexRay FrameFlexRay Frame• Header Segment:-

• Control Bits

• inc. Sync Bit

• Frame ID (11 bits) – Analogous to CAN ID

• Length (7 bits) – Length of payload in word (0 to 127 words = 0 to 254 bytes)

• Header CRC (11 bits) – Protects header

• Cycle (6 bits)

• Payload Segment:-• Message Data (0 to 254 bytes)

• 1st two bytes can be used for Message ID

• Can be used to Transpose J1939 PGN values

• Trailer Segment:-• CRC (24 bit)

TDMA inTDMA in FlexRayFlexRay• Cycle subdivide into a static and a dynamic segment.

Exclusive bus access enabled for short time in each case.Dynamic segment for transmission of variable length information.Bandwidth used when it is actually needed.

[http

://w

ww

.ixxa

t.de/

engl

ish/

prod

ukte

/flex

ray/

flexr

ay_i

ntr

Static segment

Channel 1Channel 2

Channel 1

Channel 2

Communication cycle

Dynamic segment

oduc

tion.

shtm

l]

Structure of Flexray networks

• Bus guardian protects the system against failing processors, e.g. so-called “babbling idiots”

Node Architecture Node Architecture -- Bus GuardianBus Guardian

• BD – Bus Driver• Electrical Physical layer

• BG – Bus Guardian• Protects message schedule

• Stops “Babbling Idiot” failure

FlexRay Network TopologiesFlexRay Network Topologies

• Dual Channel Cascaded Active Star:-

• Additional safety

• Mixed Topology Network:-

• Mixture of Star and Bus topologies

Numerous topologies,

include:-• Passive Star:-

• Low cost star

• Active Star:-

• Fault tolerant star

• Linear Passive Bus:-

• Similar to current CAN bus

• Dual Channel Bus:-

• Dual redundancy

• Cascaded Active Star:-

• Multiple couplers

Dual Channel BusDual Channel Bus

Active StarActive Star

Message routing results in lower loading on some branches

Cascaded Active Star NetworkCascaded Active Star Network

Mixed Channel NetworkMixed Channel Network

Mixed Topology NetworkMixed Topology Network

Network topology overview

Electrical & optical physical layer

allows for high data rates, increases error containment

passive medium, most experience, cost efficient

tolerates one faulty channel

Dual Channel

reduced wire-harness, experience, cost

Single Channel

Multiple StarBus

Network topology Network topology –– passive buspassive bus

• Maximum 8 stubs in a 10MBit/s network• Maximum 0.2m length of stub• Maximum 12m cable length between nodes• Acceptable EMC performance achievable• All in all feasible, but NO fault containment

Short onbus wires

Short onbus wires

Network topology Network topology –– active staractive star

• Maximum 16 branches• Maximum 24m length of branch (without stubs)• Maximum 12m length of branch (with stubs)• Improved EMC performance• Fault containment on branches

Network topology Network topology –– active staractive star

N

M

• Active stars can be cascaded, which means connected to each other with a point-to-point connection.

• No stub nodes on the connection between two active stars.

• A communication element that is sent from a node M to a node N may pass at most 2 active stars on its way.

Timing HierarchyTiming Hierarchy

Timing HierarchyTiming HierarchyFour levels on an internal node basisFour levels on an internal node basis

Communication Cycle Level Communication Cycle Level -- numbered from 0 to numbered from 0 to cCycleCountMaxcCycleCountMaxApart from startup, Communication Cycle is:-• Executed periodically• Consists of a constant number of Macroticks• Contains the following components:-

• Static Segment (Compulsory):- TDMA• Dynamic Segment (Optional):- Mini-slotting• Symbol Window (Optional) – Wake Up Symbol etc.• Network Idle Time (Optional)Arbitration Grid LevelArbitration Grid Level• Static Slot – Static Segment• Mini-Slot – Dynamic SegmentMacrotickMacrotick LevelLevelDesignated boundaries – action points – instants at which transmissions startMicrotickMicrotick LevelLevelDerived directly from the oscillator – a local node time – not used in synchronisation

Timing HierarchyTiming Hierarchy –– Bit TimesBit Times

Derived from clock ticks of the node’s oscillatorDerived from clock ticks of the node’s oscillator

Message Exchange OptionsMessage Exchange Options

Messages exchanged within framesMessages exchanged within frames

Frames sent within the recurring communication cycleFrames sent within the recurring communication cycle

Frame scheduling performed autonomously by the Frame scheduling performed autonomously by the protocol engineprotocol engine

Statically scheduled framesStatically scheduled frames• Repetitive message transfer with bounded communication latency • Example: distributed control loops

Dynamically scheduled framesDynamically scheduled frames• Spontaneous message transfer• Example : Diagnostic Information

How is Synchronisation Achieved?How is Synchronisation Achieved?Some definitionsSome definitions•• Global time: Common time within a cluster or networkGlobal time: Common time within a cluster or network

•• Local time: the time of the nodes clockLocal time: the time of the nodes clock

Synchronisation is the minimisation of the difference Synchronisation is the minimisation of the difference between Global and Local time between Global and Local time •• Achieved using Sync Frames in the Static Segment:Achieved using Sync Frames in the Static Segment:--

•• Sync Frame if Sync Bit is setSync Frame if Sync Bit is set

•• If set, receiving nodes should use for synchronisationIf set, receiving nodes should use for synchronisation

•• Sync frame configuration ruleSync frame configuration rule

•• At least 3 nodes shall be configured to be Sync nodes At least 3 nodes shall be configured to be Sync nodes --> Fault tolerance> Fault tolerance

•• Sync nodes with 2 channels, sends one per channelSync nodes with 2 channels, sends one per channel

•• NonNon--Sync nodes do NOT send Sync FramesSync nodes do NOT send Sync Frames

How is Synchronisation Achieved?How is Synchronisation Achieved?

Time MeasurementTime Measurement•• Every receiving node shall Store (by channel), the time differeEvery receiving node shall Store (by channel), the time differences (in nces (in microticksmicroticks) between the Local Time and the observed arrival time of all Sy) between the Local Time and the observed arrival time of all Sync nc framesframes

External Clock SynchronisationExternal Clock Synchronisation•• Synchronisation of different FlexRay networksSynchronisation of different FlexRay networks

Bus ArbitrationBus Arbitration

Arbitration based on unique Frame Identifiers in Static and Dynamicsegement

The Frame Identifier determines the transmission slot

Frame Identifier: 1 to SlotIdMaxCollision free arbitration via unique IDs and minislot counting• Frame sent when scheduled frame ID matches slot counter

Statically Scheduled FramesStatically Scheduled Frames

Frames of static length assigned uniquely to slots of static duration• Frame sent when assigned slot matches slot counter

BG protection of static slots

Dynamically Scheduled FramesDynamically Scheduled Frames

Dynamic bandwidth allocation• per node as well as per channelCollision free arbitration via unique IDs and minislot counting• Frame sent when scheduled frame ID matches slot counterNo BG protection of dynamic slots

Dynamically Scheduled FramesDynamically Scheduled Frames

Duration of Dynamic Slot depends upon whether or not frame tx or rxtakes place

Each mini slot contains an Action Point (macroticks) when transmission takes place

If transmission does not take place, then moves to next mini-slot

Communication ExampleCommunication ExampleCommunication Cycle Length

Static Segment Dynamic Segment

Cycle 0 m+1 m+2Static Slot 0 Dynamic Slot ID mStatic Slot 1

Static Slot 0Cycle 1 Static Slot 1 m m+1 Dynamic Slot ID m+2

Cycle 2 m+2m+1mStatic Slot 0 Static Slot 1

Back to Cycle 0 again

Bus Driver(at endnode)

Split termination

STP or UTP cable

C1

RT2

RT2

R1

BP

BM

%<10Matching of termination

|RT-Z0|/Z0Z0 differential mode impedance at 10MHz between 80 and 110Ohm

nF47CapacitorC1

Ohm<10ResistorR1

UnitTypeDescriptionName

Network topology Network topology –– cable terminationcable termination

Bus Driver(at endnode)

STP or UTP cable

Split termination

C1

RT2

RT2

R1

BP

BM

CBM

CDiff

CBP

Hint: Capacitances to be measured at test frequency fTest = 5MHz

pF< 40ECU's differential input capacitanceCDiff

pF< 50Capacitance of BM to GNDCBM

pF< 50Capacitance of BP to GNDCBP

UnitTypeDescriptionName

Network topology Network topology –– cable termination: cable termination: better ESD protectionbetter ESD protection

Network topology Network topology –– cable terminationcable termination

uH<< 1Stray inductanceLsigma

uHApplication specific selection necessary.Main inductanceLCMC

mOhm< 300Length resistanceRCMC

UnitTypeDescriptionName

• To improve the emission & immunity performance

• Forces the current to be of the same strength, but opposite direction.

• The parasitic stray inductance should be as low as possible (keeps bus oscillations low).

Bus Driver(at endnode)

Split terminationSTP or UTP cable

C1

RT2

RT2

R1

commonmode choke

BP

BM

CAN SignalsCAN SignalsBit Convention•Logical “0” = Dominant•Logical “1” = Recessive

Recessive Recessive

Vdiff

0 V

Dominant

CAN_High

VDiff2 V

CAN_Low

2.5 V

3.5 V

1.5 V

ISO 11898 CAN High Speed

CAN Physical Signaling CAN Physical Signaling

CAN Driver

Can_H

Can_L

CANChip

CANBus

t

V3.5 V

2.5 V

1.5 V

0 0 0 01 1 1

CAN Signalling1 = Recessive (Vdiff = 0V)2 = Dominant (Vdiff = 2V)

Can_H

Can_L

Electrical SignalingElectrical Signaling• Differential voltage uBus = uBP - uBM• Idle-LP is when no current is drawn, i.e. Power Off situation. BP and BM

are at GND.• Idle is when no current is drawn but BP and BM are biased to the same

voltage level• Data_1, BP is at a +ve level, BM is at a -ve level, Differential = +ve• Data_0, BM is at a +ve level, BP is at a –ve level, Differential = -ve

FlexRayFlexRay Differential VoltagesDifferential Voltages

Voltage LevelsVoltage Levels• The FlexRay PL has a buffer supplied by VBuf (typically

around 5 volts)• The idle level is half VBuf

– Typically around 2.5 volts

• Red shows BP• Green shows BM

Voltage Levels at StartupVoltage Levels at Startup• Shows rise from Idle_LP to Idle

Signal IntegritySignal Integrity

3. 3. Related Consortiums, Standards and Related Consortiums, Standards and Industry InitiativesIndustry Initiatives

FlexRay ConsortiumFlexRay Consortium

•• Started 2001Started 2001

•• Core MembersCore Members

•• BMWBMW

•• BoschBosch

•• DaimlerChryslerDaimlerChrysler

•• FreescaleFreescale (formally Motorola)(formally Motorola)

•• GMGM

•• PhilipsPhilips

•• VWVW

•• WCT WCT –– Development Member Development Member

ASAMASAMAAssociation for ssociation for SStandardisation of tandardisation of AAutomation and utomation and MMeasuring Systems easuring Systems

MembersMembers

•• Automotive OEMsAutomotive OEMs

•• 11stst Tier suppliersTier suppliers

•• Tool suppliersTool suppliers

www.www.asamasam.net.net

Key standards associated with FlexRayKey standards associated with FlexRay

•• XCP XCP –– Calibration Protocol, where “X” denotes any protocolCalibration Protocol, where “X” denotes any protocol

•• FIBEX FIBEX –– Equiv. Equiv. CANdbCANdb or LIN Description File (LDF)or LIN Description File (LDF)

ASAMASAMXCP XCP –– Calibration ProtocolCalibration Protocol

•• Used for data logging and parameter tuning during developmentUsed for data logging and parameter tuning during development

•• CAN, FlexRay and USB specifiedCAN, FlexRay and USB specified

FIBEXFIBEX

•• XML based file format for describing automotive control system XML based file format for describing automotive control system ::-- e.g. e.g.

•• NetworksNetworks

•• NodesNodes

•• MessagesMessages

•• Signals (e.g. Engine Speed)Signals (e.g. Engine Speed)

AUTomotive Open System ARchitecture,AUTOSAR AUTOSAR

•• WCT subcontractor advising on LIN software architectureWCT subcontractor advising on LIN software architecture

•• Motivation: Basic I/O functionality is not a selling point for Motivation: Basic I/O functionality is not a selling point for automotive automotive OEMOEM

•• therefore look to standardise software componentstherefore look to standardise software components

•• reduce costs of production of vehicle electronicsreduce costs of production of vehicle electronics

•• Advantages:Advantages:--

•• Ease of portability between different microcontroller architectEase of portability between different microcontroller architecturesures

•• Open up the ability to use different software suppliersOpen up the ability to use different software suppliers

AUTOSAR - FeaturesGoals• Scalability to different vehicle and platform variants• Implementation & standardization of basic system functions as an OEM wide “Standard Core“ solution• Transferability of functions throughout network• Integration of functional modules from multiple suppliers• Maintainability throughout the whole “Product Life Cycle”• Increased use of “Commercial off the shelf hardware”• Software updates and upgrades over vehicle lifetime

Software Architecture• AUTOSAR Software• Runtime Environment• Basic Software

JASPAR (Japan Automotive JASPAR (Japan Automotive Software Platform Architecture)Software Platform Architecture)

•• OnOn--going Japanese consortiumgoing Japanese consortium

•• Similar aims to AUTOSARSimilar aims to AUTOSAR

•• Parts of JASPAR involved in AUTOSARParts of JASPAR involved in AUTOSAR

•• www.www.jasparjaspar..jpjp

SAPECSSAPECS((SSecured ecured AArchitecture & rchitecture & PProtocols for rotocols for EEnhanced nhanced CCar ar

SSafety)afety)•• Capture Requirements of :Capture Requirements of :--

•• information around vehicleinformation around vehicle

•• telematictelematic information between vehicle & infrastructureinformation between vehicle & infrastructure

•• FlexRay DemoFlexRay Demo•• Develop and integrate FlexRay IP for demoDevelop and integrate FlexRay IP for demo

•• Demo of power train controlDemo of power train control

•• Analysis / Qualification tool for displaying dataAnalysis / Qualification tool for displaying data

•• Qualification standards for systemsQualification standards for systems•• Review of currentReview of current

•• Suggestion of new procedures and tools for qualificationSuggestion of new procedures and tools for qualification

SAPECSSAPECS((SSecured ecured AArchitecture & rchitecture & PProtocols for rotocols for EEnhanced nhanced CCar ar

SSafety)afety)

•• Project Partners Project Partners

•• Atmel: Project LeadAtmel: Project Lead

•• ValeoValeo: Automotive applications expertise: Automotive applications expertise

•• CS: Vehicle and CS: Vehicle and TelematicTelematic InformationInformation

•• AMIS: Semiconductor manufacturerAMIS: Semiconductor manufacturer

•• WCT: InWCT: In--vehicle networking tools and expertisevehicle networking tools and expertise

•• End date: June 2007End date: June 2007

EASISEASISElectronic Architecture and System

Engineering for Integrated Safety Systems

•• European consortium looking to propose future vehicle European consortium looking to propose future vehicle network topologiesnetwork topologies

•• Partners include:Partners include:--

•• PSAPSA

•• DaimlerChryslerDaimlerChrysler

•• Volvo Volvo

•• TRWTRW

•• etc.etc.

EASIS EASIS –– Functional ArchitectureFunctional Architecture

Recommended Network ProtocolsRecommended Network Protocols

EASIS EASIS –– System TopologySystem Topology

EASIS EASIS –– High End Recommended High End Recommended ArchitectureArchitecture

EASIS: Low End Recommended EASIS: Low End Recommended ArchitectureArchitecture

FlexRayFlexRay Future Future –– FlexRayFlexRay IIIIFlexRay 2.1• Dual channel• Multi-master• Large RAM requirements and no. gates – therefore expensive• Implemented in BMW X5 chassis control (single channel) as high speed

network NOT x-by-wire• Audi to implement FlexRay in future

FlexRay II• Aim to make FlexRay cheaper and ultimately replace CAN? • Single-master• Single channel• Lower silicon requirements