FlexRay transceivers for signal integrity and EMC simulations

Automotive bus systems like FlexRay or CAN transmit safety critical data. To ensure correct functionality under all circumstances, extensive investigations about signal integrity and EMC have to be performed. To be able to use simulation in this validation process, suitable models for the components of the bus systems have to be developed. This paper shows how a combined transceiver model for signal integrity and EMC investigations can be created. The model shows good results in comparison to measurement data.


Introduction
Automotive bus systems are used to connect control units, intelligent sensors and actors in vehicles.From economical point of view extended cable networks are desirable.Systems are often operated close to their specification limits.To provide safety critical applications under these circumstances or electromagnetic disturbances, the functionality of a bus system must be ensured with sophisticated methods.
An approach to detect problematic behaviour and its causes is the computational investigation of the transmitted signal quality (signal integrity) of the physical layer of the bus system.Models for any component of a bus system have to be developed.Accuracy of models and methods has to be qualified by measurements with realistic systems.
To ensure a high immunity against electromagnetic disturbances, the transceivers should be qualified with special RF immunity tests.EMC simulations may help to investigate possibly critical cases.Therefore accurate EMC behavioural models are needed (Hilger et al., 2010).
This paper shows the capabilities of bus system signal integrity and EMC simulation and focuses on modelling of the transceiver device.SI transceiver models mainly deal Correspondence to: H. Günther (harald.guenther@tu-dortmund.de)with output characteristics while EMC models reflect the input behaviour of a device.However, intersections between those two fields exist and can be used to create a transceiver model to reflect both behaviours.As result the combined model is able to reflect the signal integrity behaviour of a real transceiver device in a wide range of input voltage and under electromagnetic disturbances and considers operation thresholds which limit normal operation ranges.

Signal integrity simulation model of FlexRay transceiver
A FlexRay transceiver consists of general functional parts, e.g. for sending, receiving, and decoding data.Figure 1 shows a general structure of interconnection between those parts, which is the basis for the transceiver signal integrity simulation model (Günther et al., 2010).
For signal integrity investigations the most important part of the bus transceiver model is the output driver.The model is based on a physical approach which is shown schematically in Fig. 2. The three possible bus levels belonging to the internal transceiver states idle, active low and active high are realized by interconnection of two MOSFET transistors.They work as controlled switches based on the MOSFET equations and connect the bus output pin with ground or supply voltage, depending on the state of the transceiver.The adaption of the switching behavior and of the bus levels is done with a differentiable control function and the series connection of two diodes.When idle, both transistors are in high impedance mode and the transceiver is ready for signal reception.An additional ohmic voltage divider is used to model the correct impedance behavior in this case and provide the bias voltage.To ensure correct behavior in different load conditions and to model the slope correctly, capacitive effects of the transceiver device are modelled using output capacitors.
Published by Copernicus Publications on behalf of the URSI Landesausschuss in der Bundesrepublik Deutschland e.V.A FlexRay transceiver consists of general functional parts, e.g. for sending, receiving, and decoding data.Figure 1 shows a general structure of interconnection between those parts, which is the basis for the transceiver signal integrity simulation model [1].For signal integrity investigations the most important part of the bus transceiver model is the output driver.The model is based on a physical approach which is shown schematically in Figure 2. The three possible bus levels belonging to the internal transceiver states idle, active low and active high are realized by interconnection of two MOSFET transistors.They work as controlled switches based on the MOSFET equations and connect the bus output pin with ground or supply voltage, depending on the state of the transceiver.The adaption of the switching behavior and of the bus levels is done with a differentiable control function and the series connection of two diodes.When idle, both transistors are in high impedance mode and the transceiver is ready for signal reception.An additional ohmic voltage divider is used to model the correct impedance behavior in this case and provide the bias voltage.To ensure correct behavior in different load conditions and to model the slope correctly, capacitive effects of the transceiver device are modelled using output capacitors.With several parameters like switching voltages of the MOSFETs and diodes and parameters of the control function it is possible to match details of signal integrity behavior of the model with the behavior of the real transceiver device.Datasheet information and measurement data is used for this.
An advantage of this behavioral modeling approach compared to behavioral modeling approaches simply using measured characteristic curves [2] is the possibility to integrate for example temperature and supply voltage dependencies of the transceiver device in the model quite easily.By using basic physical equations the influences of several quantities can be integrated directly.

B. Behavior of SI model in typical FlexRay networks
The models for the bus system component were verified with measurement results for each component individually.The signal integrity behavior was validated at several load conditions.Good agreement between simulation and measurement data could be achieved.
Below the simulation results of two FlexRay bus system topologies shown in Figure 3 are presented as example and compared with measurement data.It is shown that with the help of simulation based investigation conclusions about real world behavior of bus systems can be drawn.Topology B also includes 5 bus nodes connected with a network containing different line lengths.Node 1 is sending data onto the bus. Figure 5 shows the signal traces of simulation and measurement at node 3. Again very good agreement between simulation and measurement data can be seen.With several parameters like switching voltages of the MOSFETs and diodes and parameters of the control function it is possible to match details of signal integrity behavior of the model with the behavior of the real transceiver device.Datasheet information and measurement data is used for this.
An advantage of this behavioral modeling approach compared to behavioral modeling approaches simply using measured characteristic curves (IBIS Open Forum, 2009) is the possibility to integrate for example temperature and supply voltage dependencies of the transceiver device in the model quite easily.By using basic physical equations the influences of several quantities can be integrated directly.
orrectly, capacitive effects of the transceiver device are ed using output capacitors.
Schematic Structure of Output Driver of FlexRay Transceiver Model h several parameters like switching voltages of the Ts and diodes and parameters of the control function it ible to match details of signal integrity behavior of the with the behavior of the real transceiver device.eet information and measurement data is used for this.advantage of this behavioral modeling approach ed to behavioral modeling approaches simply using ed characteristic curves [2] is the possibility to integrate mple temperature and supply voltage dependencies of sceiver device in the model quite easily.By using basic l equations the influences of several quantities can be ted directly.
havior of SI model in typical FlexRay networks models for the bus system component were verified easurement results for each component individually.gnal integrity behavior was validated at several load ons.Good agreement between simulation and ement data could be achieved.
ow the simulation results of two FlexRay bus system ies shown in Figure 3 are presented as example and ed with measurement data.It is shown that with the simulation based investigation conclusions about real ehavior of bus systems can be drawn.
ology A is a chain connection of 5 bus nodes connected us lines with a length of 2 meters each.Node 1 is .Figure 4 shows the comparison between simulation easurement data at node 4. The agreement between behavior and measurement values is very good.Topology B also includes 5 bus nodes connected with a network containing different line lengths.Node 1 is sending data onto the bus. Figure 5 shows the signal traces of simulation and measurement at node 3. Again very good agreement between simulation and measurement data can be seen.slope correctly, capacitive effects of the transceiver device are modelled using output capacitors.With several parameters like switching voltages of the MOSFETs and diodes and parameters of the control function it is possible to match details of signal integrity behavior of the model with the behavior of the real transceiver device.Datasheet information and measurement data is used for this.
An advantage of this behavioral modeling approach compared to behavioral modeling approaches simply using measured characteristic curves [2] is the possibility to integrate for example temperature and supply voltage dependencies of the transceiver device in the model quite easily.By using basic physical equations the influences of several quantities can be integrated directly.

B. Behavior of SI model in typical FlexRay networks
The models for the bus system component were verified with measurement results for each component individually.The signal integrity behavior was validated at several load conditions.Good agreement between simulation and measurement data could be achieved.
Below the simulation results of two FlexRay bus system topologies shown in Figure 3 are presented as example and compared with measurement data.It is shown that with the help of simulation based investigation conclusions about real world behavior of bus systems can be drawn.
Topology A is a chain connection of 5 bus nodes connected with bus lines with a length of 2 meters each.Node 1 is sending.Figure 4 shows the comparison between simulation and measurement data at node 4. The agreement between model behavior and measurement values is very good.Topology B also includes 5 bus nodes connected with a network containing different line lengths.Node 1 is sending data onto the bus. Figure 5 shows the signal traces of simulation and measurement at node 3. Again very good agreement between simulation and measurement data can be seen.

Behavior of SI model in typical FlexRay networks
The models for the bus system component were verified with measurement results for each component individually.The signal integrity behavior was validated at several load conditions.Good agreement between simulation and measurement data could be achieved.
Below the simulation results of two FlexRay bus system topologies shown in Fig. 3 are presented as example and compared with measurement data.It is shown that with the help of simulation based investigation conclusions about real world behavior of bus systems can be drawn.
Topology A is a chain connection of 5 bus nodes connected with bus lines with a length of 2 m each.Node 1 is sending.Figure 4   Topology B also includes 5 bus nodes connected with a network containing different line lengths.Node 1 is sending data onto the bus. Figure 5 shows the signal traces of simulation and measurement at node 3. Again very good agreement between simulation and measurement data can be seen.

EMC simulation model of FlexRay transceivers
Because of the nonlinear behavior of the transceivers for EMC modeling, not only the determination of the stationary impedance is important, but also the dynamic behavior in case of bus communication failure (Hilger and Frei, 2008a,b).In other investigations the input impedance of IC's were mostly determined by the reflection measurements of the input pins and potential ESD protection elements were implemented as simple diode models (Boyer et al., 2007).The nonlinear effects near or over critical values are not considered in these examinations.In Hilger and Frei (2008b) investigations on modeling of CAN and FlexRay transceivers were made and very simple equivalent impedance models for EMC simulations were presented.However the behavior in case of transient disturbances or the partial destruction of bits by amplitude modulated signals in combination with ESD protection circuits and the transceiver internal signal processing is not considered.The receiver unit of a differential bus transceiver can be modeled as ideal high impedance comparator with low pass filter.High frequencies or fast transient pulses may not affect the correct detection of the signal and pulse width.In Table 1 the equivalent RC-circuits of investigated bus transceivers are shown.
Figure 6 shows a simple behavioral model of the bus receiver with equivalent impedance bus load, comparator and low pass filter.
The comparator detects and amplifies the differential bus voltage.The connected low pass will filter all high frequency In VHDL-AMS the transceiver behavior can be modeled as combination of discrete elements and variable input impedances (V/I-and f/Z-tables) which can be found by measuring the DC V/I characteristic and the impedance in failure point in frequency domain.

EMC Simulation Model of FlexRay Transceivers
Because of the nonlinear behavior of the transceivers for C modeling, not only the determination of the stationary edance is important, but also the dynamic behavior in case bus communication failure [4], [5].In other investigations input impedance of IC's were mostly determined by the lection measurements of the input pins and potential ESD tection elements were implemented as simple diode models .The nonlinear effects near or over critical values are not sidered in these examinations.In [5] investigations on deling of CAN and FlexRay transceivers were made and y simple equivalent impedance models for EMC ulations were presented.However the behavior in case of sient disturbances or the partial destruction of bits by plitude modulated signals in combination with ESD tection circuits and the transceiver internal signal processing not considered.The receiver unit of a differential bus sceiver can be modeled as ideal high impedance parator with low pass filter.High frequencies or fast sient pulses may not affect the correct detection of the nal and pulse width.In Table I  Figure 6 shows a simple behavioral model of the bus eiver with equivalent impedance bus load, comparator and pass filter.
The comparator detects and amplifies the differential bus tage.The connected low pass will filter all high frequency ts of the signal and the digital unit is modeled as ideal A/Dconverter.
In VHDL-AMS the transceiver behavior can be modeled as bination of discrete elements and variable input edances (V/I-and f/Z-tables) which can be found by asuring the DC V/I characteristic and the impedance in ure point in frequency domain.parts of the signal and the digital unit is modeled as ideal A/D-D/A converter.
In VHDL-AMS the transceiver behavior can be modeled as combination of discrete elements and variable input impedances (V/I-and f/Z-tables) which can be found by measuring the DC V/I characteristic and the impedance in failure point in frequency domain.
Figure 7 shows sample V/I-measurement results done with HPPI TLP (HPPI Transmission Line Pulser, 2010) of two typical FlexRay transceivers.At levels over 45-50 V internal ESD circuits may switch and the current increases critically.Below 45 V both transceivers have high impedances. In

D. Investigations of FlexRay Transceivers in Combination with ESD Protection Circuits
To perform EMC simulations in combination with ESD protection circuits for signal integrity investigations the transceiver models can be modularly extended with VHDL-AMS models of protection elements.
ESD protection elements are intended for the protection against rarely occurring pulses.A permanent operation with CW-coupling above the breakdown voltage is not possible.The thermal power, generated by a coupled CW-current, can not be bypassed and will leads to destruction.The immunity simulations with sinusoidal disturbances were exemplarily done with a TVS diode array from Protek type GBLCS05C.The TVS array has a breakdown voltage of 6 V@1 mA and a parasitic impedance of 1.4 pF || 100 kΩ.
According to the DPI method [9] the noise voltage will be coupled to both bus lines through the termination resistors and a capacitor.In Figure 9 the simulation and measurement setup is shown.A comparison between simulation and measurement is given in Figure 10.The voltages are limited by the protection element.At disturbance amplitude of about 8 V, the transceiver can not detect the differential voltage anymore.The disturbed differential bus signal with the resulting bit errors can be seen below.The correlation between simulation and measurement results is very good.
-1 -0,5 0 0,5 1 -10 To verify the low pass filter characteristics and the digital behavioral model in combination with ESD protection circuits measurements at high and low frequencies were done.In Figure 11 a comparison between the simulation with the behavioral model and a measured signal of the FlexRay transceiver can be seen.The transceiver is combined with the Protek TVS array.Interference frequency is 50 MHz and the injected amplitude is 20 V. The correlation of presented curves is good.

III. A COMBINED FLEXRAY TRANSCEIVER MODEL FOR SI AND EMC SIMULATIONS
The presented SI model of the FlexRay transceiver mainly focuses on the output behavior of the bus driver, while the presented EMC model reflects the input behavior of the circuit.Integration of the two models to a combined SI and EMC transceiver simulation model is possible.Figure 12 gives an overview of the common parts of the two models.

Investigations of FlexRay transceivers in combination with ESD protection circuits
To perform EMC simulations in combination with ESD protection circuits for signal integrity investigations the transceiver models can be modularly extended with VHDL-AMS models of protection elements.ESD protection elements are intended for the protection against rarely occurring pulses.A permanent operation with CW-coupling above the breakdown voltage is not possible.The thermal power, generated by a coupled CW-current, can not be bypassed and will leads to destruction.The immunity simulations with sinusoidal disturbances were exemplarily done with a TVS diode array from Protek type GBLCS05C.The TVS array has a breakdown voltage of 6 V@1 mA and a parasitic impedance of 1.4 pF || 100 k .
According to the DPI method (IEC 62132-4:2006(IEC 62132-4: , 2006) the noise voltage will be coupled to both bus lines through the termination resistors and a capacitor.In Fig. 9 the simulation and measurement setup is shown.
A comparison between simulation and measurement is given in Fig. 10.The voltages are limited by the protec- A comparison between simulation and measurement is given in Figure 10.The voltages are limited by the protection element.At disturbance amplitude of about 8 V, the transceiver can not detect the differential voltage anymore.The disturbed differential bus signal with the resulting bit errors can be seen below.The correlation between simulation and measurement results is very good.
-1 -0,5 0 0,5 1 -10  A comparison between simulation and measurement is given in Figure 10.The voltages are limited by the protection element.At disturbance amplitude of about 8 V, the transceiver can not detect the differential voltage anymore.The disturbed differential bus signal with the resulting bit errors can be seen below.The correlation between simulation and measurement results is very good.
-1 -0,5 0 0,5 1 -10  The transmitting part with the output drivers is present only in the SI model, thus it will be used for the combined model.Lowpass filter, digital behavior, and output impedances are present in both SI and EMC models, but show big similarities in general parts.The EMC model contains additional information about differential impedances and operation limits.

IV. SIMULATION OF BCI TEST WITH FOUR NODE FLEXRAY NETWORK
With a VHDL-AMS model of a BCI clamp from Fehler! Verweisquelle konnte nicht gefunden werden.it is possible to make virtual CAN or FlexRay network tests.In combination with a cable model [10] and the transceiver models an overall simulation model can be created.To investigate an extended network 4 FlexRay nodes are interconnected with cable models of twisted pair cables.In Figure 13  In Fig. 11 a comparison between the simulation with the behavioral model and a measured signal of the FlexRay transceiver can be seen.The transceiver is combined with the Protek TVS array.Interference frequency is 50 MHz and the injected amplitude is 20 V. The correlation of presented curves is good.

A combined FlexRay transceiver model for SI and EMC simulations
The presented SI model of the FlexRay transceiver mainly focuses on the output behavior of the bus driver, while the presented EMC model reflects the input behavior of the circuit.Integration of the two models to a combined SI and EMC transceiver simulation model is possible.Figure 12 gives an overview of the common parts of the two models.The transmitting part with the output drivers is present only in the SI model, thus it will be used for the combined model.Lowpass filter, digital behavior, and output impedances are present in both SI and EMC models, but show big similarities in general parts.The EMC model contains additional information about differential impedances and operation limits.
with two 1.3 kΩ resistors, nodes 2 and 3 with two 47 Ω resistors and each with 4.7 nF to GND.Common mode chokes or other protection elements are not used in this setup.The BCI clamp is positioned on the cable of node 2 in a distance of 0.2 m. Figure 14 shows a comparison between the simulated and measured voltages at different FlexRay nodes.The measurements were done in the frequency domain with a network analyzer and in the time domain with a signal generator and oscilloscope.Here higher input power and active voltage probes were used.The BCI clamp was supplied with constant power of 10 dBm over all frequencies.The parasitic effects of connectors, the PCB and discrete elements, like resistors and capacitors, were considered in the model.The model shows very good correlation up to 200 MHz between simulated and measured results.with 4.7 nF to GND.Common mode chokes or other protection elements are not used in this setup.The BCI clamp is positioned on the cable of node 2 in a distance of 0.2 m.
Figure 14 shows a comparison between the simulated and measured voltages at different FlexRay nodes.The measurements were done in the frequency domain with a network analyzer and in the time domain with a signal generator and oscilloscope.Here higher input power and active voltage probes were used.The BCI clamp was supplied with constant power of 10 dBm over all frequencies.The parasitic effects of connectors, the PCB and discrete elements, like resistors and capacitors, were considered in the model.The model shows very good correlation up to 200 MHz between simulated and measured results.

Conclusion
In order to investigate possible disturbances in bus system operation with simulation, behavioral models of bus transceivers for SI and EMC investigations were developed.SI model focus mainly on the output behavior of the transceivers, while EMC models reflect the input behavior.A detailed comparison of two developed models for these applications shows, that integration into a combined transceiver model for SI and EMC simulations is possible.Transmitter parts are taken from the SI model, since the EMC model gives detailed information about output impedances and operation limits.Digital receiving parts are present in both models.The combined model shows good matching with measurement results in an application example with disturbance signal injection through a BCI clamp.

Figure 1 :
Figure 1: Schematic Structure of Output Driver of FlexRay Transceiver Model

Fig. 1 .
Fig. 1.Internal structure of output driver of FlexRay transceiver model.

Figure 2 :
Figure 2: Schematic Structure of Output Driver of FlexRay Transceiver Model

Figure
Figure 3: Structure of FlexRay Topologies

Figure 4 :
Figure 4: Simulation and Measurement Results for Topology A

Figure 2 :
Figure 2: Schematic Structure of Output Driver of FlexRay Transceiver Model

Fig. 4 .
Figure 5: Simulation and Measurement Results for Topology B

Figure 5 :
Figure 5: Simulation and Measurement Results for Topology B Fig. 5. Simulation and measurement results for topology B.

Figure 6 .Fig. 6 .
Figure 6.Behavioral model of FlexRay receiver input Figure 7 shows sample V/I-measurement results done with HPPI TLP [7] of two typical FlexRay transceivers.At levels over 45 -50 V internal ESD circuits may switch and the Figure 7. TLP measurement, V/I characteristic of tested FlexRay transceivers In Figure 8 the frequency dependent critical failure voltages of the FlexRay transceivers are presented.Because of constructive variations and internal signal processing the transceivers have different immunity levels.

Figure 8 .Fig. 7 .
Figure 8. Critical failure voltage of tested FlexRay transceivers The voltage change between 90 and 100 MHz are caused by resonances in the measurement setup.
Fig. 8 the frequency dependent critical failure voltages of the FlexRay transceivers are presented.Because of constructive variations and internal signal processing the transceivers have different immunity levels.The voltage change between 90 and 100 MHz are caused by resonances in the measurement setup.www.adv-radio-sci.net/9/111/2011/Adv.Radio Sci., 9, 111-116, 2011

Figure 7 .
Figure 7. TLP measurement, V/I characteristic of tested FlexRay transceiversIn Figure8the frequency dependent critical failure voltages of the FlexRay transceivers are presented.Because of constructive variations and internal signal processing the transceivers have different immunity levels.

Figure 8 .
Figure 8. Critical failure voltage of tested FlexRay transceivers The voltage change between 90 and 100 MHz are caused by resonances in the measurement setup.

Figure 9 .
Figure 9. Test setup incoupling with DPI method

Figure 10 .
Figure 10.Simulated and measured differential bus voltage, interference frequency 1 MHz with FlexRay Type A and Protek TVS Array

Fig. 9 .
Figure 12.Common parts of SI and EMC transceiver modelThe transmitting part with the output drivers is present only in the SI model, thus it will be used for the combined model.Lowpass filter, digital behavior, and output impedances are present in both SI and EMC models, but show big similarities

Figure 9 .
Figure 9. Test setup incoupling with DPI method

Figure 10 .Fig. 10 .
Figure 10.Simulated and measured differential bus voltage, interference frequency 1 MHz with FlexRay Type A and Protek TVS Array

Figure 9 .
Figure 9. Test setup incoupling with DPI method

Figure 10 .
Figure 10.Simulated and measured differential bus voltage, interference frequency 1 MHz with FlexRay Type A and Protek TVS Array

Fig. 12 .
Fig. 12. Common parts of SI and EMC transceiver model.

Figure 13 .
Figure 13.BCI test setup with passive star and four nodes

Fig. 14 .
Fig. 14.Comparison between simulated and measured amplitudes at the four FlexRay nodes.
shows the comparison between simulation and measurement data at node 4. The agreement between model behavior and measurement values is very good.

Transceiver CAN Type A FlexRay Type A FlexRay Type B uivalent Impedance 12
the equivalent RC-circuits of estigated bus transceivers are shown.

Table 1 .
Equivalent impedances of investigated bus drivers.
EMC model reflects the input behavior of the circuit.Integration of the two models to a combined SI and EMC transceiver simulation model is possible.Figure12gives an overview of the common parts of the two models. presented

of BCI test with four node FlexRay network
(Zhang et al., 2008)passive star and four nodes.With a VHDL-AMS model of a BCI clamp fromHilger et al. (2010)it is possible to virtual CAN or FlexRay network tests.In combination with a cable model(Zhang et al., 2008)and the transceiver models an overall simulation model can be created.To investigate an extended network 4 FlexRay nodes are interconnected with cable models of twisted pair cables.In Fig.13the BCI test setup is shown.According to FlexRay Protocol Specification(2005)all nodes are split terminated.Nodes 1 and 4 with two 1.3 k resistors, nodes 2 and 3 with two 47 resistors and each 4 Simulation www.adv-radio-sci.net/9/111/2011/Adv.Radio Sci., 9, 111-116, 2011