Physical Layer Simulation of CAN

[roTTer]

I would like to simulate the Physical Layer of the Controller Area Network(ISO 11898) to evaluate the performance of the system under varying conditions(Temperature, Twists/turn, Length, Termination type, Unfitting Node Terminations, Interconnections, Splices, Unbalanced Loading etc.)

I will be quite frank and confess that I do not have a deep-knowledge on the modeling of Transmission Line. That's the reason I am here!

My plan was to model the Shielded Twisted-Pair cable and get the appropriate Transceiver model(TJA1051 from NXP) from the supplier end, and simulate it for a simple 2-node system, and verify the results with an experimental setup, so as to validate the modeling of the Transmission Lines. At a later stage, I can take up complex systems for simulation.

I require help in identifying the necessary documents, papers and books that I can read up on to get this going.I have studied the Specification Sheets of CAN, it's data layer J1939 and other relevant documents.

Though I have few generic doubts that I am listing out below:

1. The necessity of chokes in the network(if yes, the formulas for it?), and also the reasoning behind it?

2. I read somewhere, that CAN is not affected by EMI? External EMI, maybe yes, because of he usage of STP, but what would be the effect of CrossTalk on it?

3. Ballpark figures for acceptable distortions, propagation delay and rise/fall times of the signal for reliable communication?

4. Do the Transceivers(CAN) support Reflected Wave Switching or are they only Incident Wave Switching Drivers?

5. Acceptable limits to the asymmetry in the signals that occur.

6. What do you mean by an Eye Diagram? I checked wiki and there's not much of data there. How do I define an Eye-Diagram for this simulation?

Please do comment if it's the right way to approach this problem and I do welcome any suggestions that you may have regarding this.

PS: This simulation will be for vehicular applications

 

[berkeman]

I would like to simulate the Physical Layer of the Controller Area Network(ISO 11898) to evaluate the performance of the system under varying conditions(Temperature, Twists/turn, Length, Termination type, Unfitting Node Terminations, Interconnections, Splices, Unbalanced Loading etc.)

I will be quite frank and confess that I do not have a deep-knowledge on the modeling of Transmission Line. That's the reason I am here!

My plan was to model the Shielded Twisted-Pair cable and get the appropriate Transceiver model(TJA1051 from NXP) from the supplier end, and simulate it for a simple 2-node system, and verify the results with an experimental setup, so as to validate the modeling of the Transmission Lines. At a later stage, I can take up complex systems for simulation.

I require help in identifying the necessary documents, papers and books that I can read up on to get this going.


I have studied the Specification Sheets of CAN, it's data layer J1939 and other relevant documents.

Though I have few generic doubts that I am listing out below:

1. The necessity of chokes in the network(if yes, the formulas for it?), and also the reasoning behind it?

2. I read somewhere, that CAN is not affected by EMI? External EMI, maybe yes, because of he usage of STP, but what would be the effect of CrossTalk on it?

3. Ballpark figures for acceptable distortions, propagation delay and rise/fall times of the signal for reliable communication?

4. Do the Transceivers(CAN) support Reflected Wave Switching or are they only Incident Wave Switching Drivers?

5. Acceptable limits to the asymmetry in the signals that occur.

6. What do you mean by an Eye Diagram? I checked wiki and there's not much of data there. How do I define an Eye-Diagram for this simulation?




Please do comment if it's the right way to approach this problem and I do welcome any suggestions that you may have regarding this.

PS: This simulation will be for vehicular applications

That's a pretty big project! How long do you have to get up to speed and get it done, including the physical testing part?

It will probably be easier for you to use Cat-5 cable to get your simulation environment set up -- there is probably a lot more literature available about Cat-5 modeling than CAN modeling. I did a google search on Simulation of Cat-5 Transmission Lines, and got some good hits. Here is a short one that is pretty good:

http://www.google.com/url?sa=t&rct=...hhkQEr8Te56J1a6IBEGhGZQ&bvm=bv.46471029,d.cGE

Which SPICE are you going to use for the simulations? What other simulation software do you have available? There are a number of tricks to simulating transmission lines that may not be in the literature -- they are often considered proprietary IP by the companies that use them to help them design network transceivers (like the company that I work for). But you can get pretty far with what has been published in the open literature.

Eye diagrams are a technique used to measure the amount of jitter in a communication channel or transmission line. You monitor the received differential signal at one of the RX devices with a differential probe and oscilloscope, and you trigger on the Transmit Enable (TXEN) signal at the sending node. You put the 'scope on infinite persistence, and let the display build up over many transitions of the RX signal. If there is no jitter, you should see the waveform overlaying itself pretty well, trace after trace. But if there is jitter in the waveform (from inter-symbol interference, ISI, for example), then the "eyes" of the overlapping traces fills in more. Worst case it closes completely, which means that the receiver will have a very hard time distinguishing between "1" and "0" symbols.

Do a search at Google Images using this: "eye diagram" transmission line.

 

[roTTer]

Hi,

I have close to around 5 months to finish this. I have no other commitments, so I can focus on this completely.
I am hoping that I'll be able to finish studying the papers/books in 30 days.

To answer your second question, I'll be using Saber for simulation purposes. Though getting MATLAB or SPICE is also an option, but seeing that I have a Saber License already, it would be feasible(cost-wise) to try do it in Saber itself.

As for identifying the literature, that is the hardest part for me right now. I do not have IEEE/SAE access, though I have built up a list of papers, but by only checking the abstracts. I am not too sure that if I buy them they would be of any help.

If you have in mind any Books on Best Practices or Techniques used in Simulation of Electrical/Communication Systems, or some papers that you are aware of that could be of any use. I would be appreciate it, if you could guide me on that.

Thanks for your suggestion on the Cat5 Wire and the explanation on the Eye-Diagram. I really need to master the art of 'Googl-ing'!

As of now, I am doing some ground work, where I am looking into the different wire types used in the automotive industry and which would be the best to simulate upon.

Modeling a real-time experimental setup for results I could refer, is also on the cards, and I'll be doing it simultaneously.

Any suggestions on what more I could do, or what I shouldn't do are welcome.

PS: I do not have a mentor or any specialist with me whom I could refer to. I am on my own! Any help would be greatly appreciated!

 

[Sanchit Tyagi]

Hi,

I have performed a comparison study on simulation and hardware circuit of 4 CAN Transceiver Nodes connected in star topology with a junction connector at the centre with 5 metre AWG19 cable length each. Developed a CAN Transceiver model in OrCAD PSpice and simulated the same circuit in PSpice, then compared the results of the same to study to effect of different Common Mode Chokes and variation of cable impedance by varying the Cable length. Study was done to analyse the effect of transients / ringing observed over the CAN bus.
Having replicated the 4 node CAN Transceiver circuit behavior in the simulation, now I want to go ahead with vehicle level simulation.
Few challenges I for see are:
1. Too high simulation time with nodes upwards of 32 (Nearly one hour)
2. Number of signals to be watched become too high with high number of nodes.
3. Due to the above two, each iteration in circuit makes the process longer.

Looking for some guidance, if anyone has done such an activity earlier.
I want to do such an activity to study the impact of changing any component on the vehicle without incurring the hardware implementation cost.