The amount of
In 2006 and 2007, FlexRay was introduced successfully into seriesproduction on a first scale and the trend will be rapid growth of newvehicle projects containing FlexRay networks.
FlexRay is designed as a very flexible, high speed and faulttolerant bus-system, which supports real-time operation with embeddedevent triggered functionality.
FlexRay as a networking system in vehicles supports various types ofnetwork topologies at a data rate of 10Mbit/s. In automotive networks,such high bandwidth has not been achieved reliably before on standardunshielded copper cables.
Optical fibres or shielded copper cables, where bandwidths beyond10Mbit/s could be achieved, are unattractive due to high material andcable routing costs.
Flexibility in automotive networks describes the ability to usecertain topologies, combined with different cable length, correctlyterminated. FlexRay is designed to combine three network types in ahybrid network: passive line networks, passive star networks and activestar networks.
Passive networks are a challenge because of the correct applicationof network termination. FlexRay nodes connected to an active starshould always be low ohmic terminated.
However, 100 percent termination of every node in a passive networkdoes not easily support network expansion and makes it difficult tofulfill the DC bus load requirements.
Termination concepts forbid impedance steps along the line, whichwill happen by adding termination resistors in the transmission line(impedance leveling).
In passive networks, termination needs to be applied at the two nodesthat are the longest wiring distance apart. It is calculated as equalto the nominal cable impedance between the two points.
The best-recommended termination concept for FlexRay networks issplit termination, whereby onehalf of the cable impedance to ground isconnected between the two bus lines. Furthermore, a common mode choke(CMC) should be applied between bus driver and termination circuitry.
Bus termination, in some cases, contradicts the flexibility demandsof a network because flexibility implies on-demand insertion or removalof network nodes. The pre-calculated nominal impedance prior to nodeexpansion or contraction may have to change afterwards to preserveoverall network integrity.
It is recommended to use active star connections only for expandingor contracting a system, but even this requires initial design input toguarantee the desired flexibility of the active star. In many cases,costs of the base structure limits design and flexibility. On the otherhand, tar- geting for best flexibility will limit the stability of anetwork.
Designing a FlexRay node is a challenge. ECUs are developed forspecific applications and are provided by different suppliers. Re-useof ECUs is achieved by defining vehicle platforms with dedicated ECUsfor applications.
However, this means that ECUs in one network look different and havedifferent behaviors not only in terms of functionality, but alsobecause of secondary effects. These secondary effects need to beconsidered in the design of a FlexRay node to guarantee designfunctionality.
The main functional blocks on a typical ECU are MCU, memory, sensorinterface, bus system interface and peripherals for power supply,protection and supervisory circuits.
In general, MCUs are not placed side by side with the connectors.The bus system interfaces on the other hand should be placed in closevicinity with the connector pins. In the case of FlexRay, a nodeconsists of an MCU, the FlexRay communication controller (standalone orintegrated in the MCU) and the FlexRay bus driver (also called FlexRaytransceiver).
The electrical distances between the FlexRay communicationcontroller and the bus driver can be up to 15cm long on the PCB. Theinterface between the FlexRay communication controller and the FlexRaybus transceiver consists of two unidirectional wires (transmit dataline, receive data line) both running at 10Mbit/s.
The design of the PCB needs to ensure that disturbances on thisinterface do not harm the entire communication path. Output and inputimpedance of the devices need to match the wire impedance on the PCB,and the outputs needs to be capable of driving a certain speci- fiedworst-case load, whereby all device parameters need to be held withinthe FlexRay specified parameter range.
One of the difficulties in designing a FlexRay node is ensuring thatthe critical parameters for FlexRay are not violated. The currentFlexRay specification cannot address all the system parameters, and sothe target is to specify measurable parameters on device level.
Therefore, the designer gets additional responsibility whendesigning a FlexRay node, as the node might work without restrictionsduring ECU validation, but not work in the network constellation.
In other words, the design of the FlexRay node has a major effect onthe remaining system margin when operational. In the automotive world,stringent environmental conditions are given within which the functionhas to be guaranteed.
|Figure1: The components on the FlexRay communication path have directinfluence on system behavior.|
When specifying FlexRay, some key parameters that cannot be fullyaddressed by FlexRay components have a major effect on system behavior.These parameters are derived by considering the entire path of thesignal flow. On that path, the components shown in Figure 1 above have direct influenceon system behavior. Additionally, the layout of the PCB has a majorinfluence on the performance on of the FlexRay node.
In most cases, the OEM decides the network structure and even thebus termination. What are the FlexRay critical parameters to beconsidered in the design of the ECU?
Propagation delays are derived by components in the signal path,starting from the sending communication controller, to the receivingcommunication controller where the FlexRay message is decoded. Thereare two kinds of propagation delays, the symmetrical delay (retentiontime of the signal through the signal path) and the asymmetrical delay(shortening and lengthening of the bit-time).
|Figure2: Bit length change due to asymmetrical propagation delays is shown.Asymmetrical propagation delays directly restrict the use of networktopologies.|
As shown in Figure 2 above ,the asymmetrical propagation delay is the more critical parameterbecause it directly restricts the use of network topologies. Inaddition to static effects, dynamic effects (i.e. electromagneticinfluences, glitches etc) influence the length of measured bits in thenetwork and must be considered during the design phase.
The SI parameters on the bus line are specified in timing anddifferential voltage levels between the two bus pins (BP and diagram ondifferent test planes is a common method to detect breaches from theexpected signal. Not every breach means that the network will notfunction, but it gives an indication that the network might beunstable.
Another approach for measuring signal integrity is to definesimplified receiver models that deliver a go or no-go result as towhether the signal can be correctly converted into a digital signal.
This can be measured at any point of the network in the analogsignal path. Both methods do not give feedback whether the decoding ofthe FlexRay message will work.
But in connection with the propagation delay considerations andsignal integrity, the function of a network can be foreseen up to acertain point.
Armed with this basic knowledge, the challenge in designing aFlexRay network is the configuration of all the nodes in the network.Unlike event-triggered networks, time-triggered technologies requiremore parameters to be configured in advance to get the communicationrunning.
Describing the protocol functionality of FlexRay will stretch thescope of this article, which is limited to describing some availablesoftware tools for configuring and sett up a FlexRay network. Toolvendors for FlexRay include EB Elektrobit, dSPACE, IXXAT, TTTechAutomotive, TZ Mikroelektronik and Vector Informatik.
Most of these tools provide an easy-to-use graphical interface toconfigure the FlexRay network and generate Fibex configuration files,which can then be interpreted by the IDE for the MCUMCU.
The configuration files include the FlexRay parameters necessary forevery node for the FlexRay communication to work. These parametersinclude FlexRay channels, static slot length, mini slot length, numberof static slots, number of mini slots and sample clock period.
|Figure3: To demonstrate a FlexRay network a gaming steering wheel wasupgraded to a FlexRay node and a converter box was developed.|
To demonstrate a FlexRay network, a steer-by-wire application waschosen (Figure 3, above ). Agaming steering wheel was upgraded to a FlexRay node and a converterbox was developed.
The converter box which is connected directly to a standard PCconverts signals from FlexRay into USB 2.0 and vice versa,. Also, twoFujitsu Starter Kits SK-91F467- FlexRay are connected to the FlexRaybus to expand the passive line network to four nodes.
The hardware for the steering wheel node and the FlexRay USBconverter box were developed from scratch. The setup described was ajoint development between austriamicrosystems AG and the FH JoanneumDepartment for Electronics and Technology Management. The principalfunctions blocks of the steering wheel node are shown in Figure 4 below.
|Figure4: The principal function blocks of the steering wheel node are shown.|
The steering wheel is mounted on a shaft with a permanent magnetfixed on the end. The magnetic rotary encoder AS5046 delivers thesteering position via SPI to the MCU. The MCU periodically reads theposition together with the button states of the steering wheel.
This information is forwarded to the FlexRay communicationcontroller. Finally, the AS8221 FlexRay transceiver fromaustriamicrosystems provides the differential bus voltages for theFlexRay channels.
The FlexRay messages are converted to USB messages so that the PCcan communicate with the steering wheel (Figure 5 below ). The MCU in thiscase acts as a host to provide the raw data received on the FlexRaycommunication controller interface to the USB link and vice versa.
|Figure5: At the converter box shown, FlexRay messages are converted to USBmessages so that the PC can communicate with the steering wheel.|
The Fujitsu Starter Kits are used to observe the FlexRay traf- ficand to record messages. This FlexRay setup allows flexibility whenexperimenting with terminations concepts.
On the steering wheel node and the converter box, a low ohmic splittermination together with a common mode choke is applied. On thestarter kits, high ohmic split termination with common mode choke isapplied.
This configuration showed the highest stability in networking. Thedemonstration setup is designed so that system critical FlexRayparameters like propagation delays and configuration parameters – canbe measured and evaluated while changing the configuration. Even thesubsequent expansion of the network topology into an active star may beconsidered when validating network communication.
Harald Gall is Product Manager BusSystems at austriamicrosystems AG and Peter Hintenaus is a lecturer andprofessor in the Department for Electronics and Technology Managementat FH Joanneum University of Applied Sciences.