Migration of existing applications from FlexRay electrical physical layer specification revision 2.1 to revision 3.0
Back in 2006 the first vehicle including a FlexRay network hit the road. A relatively low-risk platform for learning, understanding, and experiencing was the starting point. The mentioned automotive application is known as adaptive drive (by BMW), and surely it presented an important milestone for the new technology, made necessary for competing with established network systems, e.g. controller area networks (CAN), in automotive environments. The deterministic communication method of FlexRay was soon valued by the system architects, opening up new vistas for the design of chassis, powertrain and safety-related applications.
To this day the number of car releases based on FlexRay electronic control units (ECUs) rises continuously. The system complexity regarding software, hardware and network topology has increased. This fact profoundly conflicts with the cost pressure weighting down newly introduced technologies. However, today the car makers (OEMs) transfer their know-how to medium-sized and small car platforms which clearly shows the degree of acceptance obtained for that bus system. Obviously this is one of the results of a close partnership between the OEMs, tier-1 suppliers, and semiconductor manufacturers. The latter make sure to provide cost-optimized FlexRay transceivers according to the best available technology and most recent requirements.
Before FlexRay networks became ready for automotive series production, years of close cooperation of networking experts had taken place. In 2001 the FlexRay consortium was founded with access to broad know-how of automotive systems, including specialists from different domains like software, hardware, electromagnetical conformity, testing, and semiconductor manufacturers. The goal to release an open standard with defined system behavior, interfaces, and parameters was reached successfully in 2005 when the base for the first generation of FlexRay transceivers, called electrical physical layer (EPL) specification revision 2.1, was officially published.
According to the FlexRay definition, a compliant physical layer communication channel consists of four elements: bus driver (BD), twisted pair (TP) bus line, bus termination, and an active star (AS). The BD (transceiver) is an interface between the microcontroller (including the FlexRay communication controller) and the FlexRay bus. The bus driver converts the digital data stream to differential analog bus voltage and vice versa. The twisted pair bus line, with defined surge impedance, connects two or more bus nodes. The bus line is terminated by a resistor. Usually an active star is placed in the center point of a FlexRay network. The active star can be used as a stand-alone device which redistributes the incoming bus signals to all other bus lines (branches). Optionally it can be used simultaneously as a signal router and a transceiver. In this case the locally transmitted data are sent to all FlexRay branches. All incoming data is mirrored at a single digital transceiver pin RxD. This functionality is needed e.g. in network gateway applications.
Because automotive network topologies are restricted by the car’s body geometry and its available mounting space, linear FlexRay network topologies unquestionable. In addition, crash-prone car regions, for instance the front part of the body, have been identified. In case of a branch failure, the remaining intact parts of the network can be separated from the defect ones.
Networks including an active star show improved EMC/EMI behavior compared to mere passive star networks and are preferable due to the fact of better parameter reproducibility and predictability. Automotive networks often require topology variants depending on the count of used electronic control units. The network partitioning always considers the minimum number of needed branches. On the other hand reduction of overall system costs can be achieved by utilization of the same hardware, connectors, wiring, etc. Therefore unused branches can be disabled by the active star.
Currently the FlexRay EPL specification revision 3.0.1 is available. Ten years after the formation of the FlexRay consortium, the second generation of FlexRay transceivers according to that specification release is in mass production. Therefore interoperability between 2.1 and 3.0 devices today, and for many years to come, is unavoidable. This applies to transceivers but also to the interaction between a communication controller and a bus driver which may be used in the same application.
The main goal during the definition process of the latest specification was to guarantee backward compatibility to EPL specification revision 2.1, a precise definition of active star parameters, and the best possible alignment to the physical layer parameter requirements of the Japan Automotive Software Platform and Architecture (JasPar) working group. This theoretical approach was refined by the test results of the actual silicon.
Conforming to the latest EPL specification revision is a must. FlexRay conformity not necessarily makes the customer decide for a certain product. The convenience is achieved in case the device’s specific functions are perfectly harmonized with the customer’s requirements.
The following migration analysis is based on a transition from AS device E910.56, conform to EPL specification 2.1 and predominantly used in gateway applications. This device will be replaced by E981.56 in line with the EPL specification revision 3.0.1. Both devices are the flagships of each FlexRay transceiver family generation manufactured by ELMOS Semiconductor AG.
The latest electrical physical layer specification includes 101 AS-related FlexRay parameters which were analyzed. 67 changes were identified compared to revision 2.1. A parameter is affected in case the parameter name, value, or measurement definition has changed. Besides that, a parameter replacement and a new parameter are also regarded as changes. The parameter analysis covers the influence on the application and system designed according to revision 2.1 including an active star in line with revision 3.0. The most critical FlexRay parameters are mentioned in the following.
The FlexRay communication controller uses internal synchronization mechanisms requiring accurate propagation delays, asymmetric delays (difference between the propagation delay of the falling and rising edge), and signal length changes which can occur at the beginning (dStarTSSLengthChange) and at the end of the FlexRay data frame (dStarFES1LengthChange), if applicable.
In the past, active star parameter definitions were often derived from the single transceiver. From the system point of view some timing parameter limits for the active star were unnecessarily narrowly defined and could be reduced. This happened in electrical physical layer specification revision 3.0. Since the AS is placed in the center of the network, the propagation delay parameters regarding the signal path from the branch to the RxD pin (dStarRx01, dStarRx10) and from the TxD pin to the branch (dStarTx01, dStarTx10) were extended. The new time budget analysis allowed also extending the asymmetrical delays for this signal path. In contrast, the signal propagation delay from branch to branch (dStarDelay01, dStarDelay10) as well as the activity and idle detection time (dStarActivityDetection, dStarIdleDetection) have been reduced.
The recent device provides a backwards compatibility to its predecessor. SPI handling has been simplified. Provided but unused features have been removed for the benefit of cost reduction. A case in point, the extended bus diagnosis was reduced to the minimum. The bus guardian functionality was added as well as the capability to effect a device reset (warm start). However, in case of an active backwards compatibility mode, the additional functionality is not available in order to avoid possible collisions with the existing application software implementations. The IC automatically recognizes whether to activate the backwards compatibility mode or to provide the new functional subset.
As a result of the alignment to the JasPar specification, the minimum differential bus level voltage (uStarTxactive) must fulfill 900mV (FlexRay 600mV) which is optional but probably will be implemented in most upcoming transceivers to cover the needs of the Japanese market. From the EMC point of view, this requirement might lead to higher radiated emission on chip and system level. Even if the transceiver passes the EMC tests required on chip level, higher bus voltages caused by the active star might negatively influence the entire FlexRay network. The electromagnetic disturbances might rise inside the integrated circuit but also along the bus lines due to surge impedance jumps and therefore could result in higher signal reflections which finally occur in the emitted frequency spectrum. Unfortunately this effect is multiplied by the number of used branches which are switched simultaneously. Current applications e.g. use up to eight branches. Extended EMC testing on chip and system level show that the best possible symmetry of the differential signal and well defined driver adjustment to the network are the most relevant parameters and that the OEMs’ EMC requirements can be fulfilled.
Conclusion
A malfunction of the active star device would immediately lead to a stranded vehicle which quite understandably is a “red rag” to the OEMs. Therefore any (minor) changes at this point are critically analyzed by the customer.
Systematic analysis of the FlexRay parameters according to EPL specification revision 3.0 combined with lab, system and in-vehicle tests demonstrate a migration from 2.1 to 3.0 within the existing hardware feature without serious issues. In case of the active star there is no need for hardware changes in existing applications due to a well-founded concept of backwards compatibility.
The drop-in replacement idea proves itself and allows a gradual transition until new hardware developments will be available.
In addition, EPL specification revision 3.0 could be verified as a solid base for the ISO standardization process which has been started.
About the authors: Radoslaw Watroba is a Field Application Engineer at ELMOS Semiconductor AG. Dr. Rolf Weber is Product Line Manager for Interface Products at ELMOS Semiconductor AG.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
