CAN bus termination methods

I noticed that the schematic shows that the CBT offers an easy way to terminate the CAN bus when necessary, by shorting what looks like a solder pad to place a 120 Ohm resistor between CAN Hi and CAN Lo. While this method should suffice for most applications, everyone should be aware that there are other termination methods and some good reasons to use them that I will try to describe here.

The higher the bus speed the more important proper termination is to bus integrity. The entire bus topology needs to be considered before some of the more obscure termination methods are deployed. Remember that some buses you may want to tap into send safety critical information from one ECU to another that can affect the safe and reliable operation of the vehicle.

Termination is probably needed at the CBT when the CBT is the end node on a CAN bus, which will be normal when bridging (severing a CAN bus) and probably not very normal when just monitoring. The normal topology is that the bus is a daisy chain, meaning that it goes from ECU to ECU to ECU to ECU, with the ECUs that are serving as endpoints on the daisy chain requiring termination and all the ECUs in the middle generally requiring no termination. Of course like all rules, these rules are often broken to save cost or for practical convenience, especially for test tools that don't remain on the bus for long periods of time. By "daisy chain", it means that the 2 CAN wires are continuous from one end to the other, with no splices, tributaries, "T" connections or stubs.

http://daeinstrument.com/images/Daisy Chain.jpg

This is why many ECUs have the same CAN bus on two different sets of pins, to avoid having more than one wire per pin (weather proof connectors don't allow more than one wire per terminal) and to allow the ECU to make the daisy chain connection internally with as short of an internal "stub" as possible. However, sometimes a short stub is impossible to avoid, like when connecting the CBT to an existing bus for bus monitoring. Even if you connect the CBT to an existing bus to monitor it by tapping into the bus wires, a "T" connection is made where you splice in and the stub is the length between your splice point and the bus transceiver internal to the CBT. I would keep that distance less than 6 inches at 500 kbps and higher as a general rule of thumb.

If the bus speed is 500kbps or higher, you'll also want to pay good attention to termination requirements. At 1Mbps this becomes critical. Thankfully, most CAN buses are 500 kbps or slower. When a stub can't be avoided it can be acceptable to use what is known as weak termination. This is a value of resistance much higher than 120 Ohms, but I'll not get into why that is here unless someone asks. You can read about it at the following link though if you'd like.

http://www.nxp.com/documents/application_note/AN10211.pdf

What is important for all of us to know is that a more common termination method these days in automotive is called "split termination". This is when the 120 Ohm terminator is split into two 60 Ohm resistors placed in series and a capacitor goes from the resistor-to-resistor junction to CAN ground. This will minimize the noise that the CBT can inflict on the CAN bus, and can make the connection more reliable for the CBT when a lot of noise is present. This is an EMC concern.

To address other EMC concerns, it is also common to add additional capacitors as filters and a common mode choke. I have seen split termination and common mode chokes solve huge problems in certain cases, such as in a hybrid or electric vehicle where high voltage is present.

Here is a depiction of split termination, the common mode choke, filter capacitance, and ESD protection. I can envision making an add-on board that does this for the current CBT, and any future iteration of the CBT having some or all of those components built-in and the split termination similarly selectable as the single terminator is now. These parts need to be as close to the CAN transceiver as possible, hence why making them part of the main PCB design is preferred.

http://www.datasheetdir.com/Common-Mode-Chokes-In-Can-Networks-Source-Of-Unexpected-Transients+Application-Notes

Tommy,

Awesome writeup. Thank you for the information :)

administrators

Great write up, thanks Tommy!

Looks like your connection to CANBus Triple was lost, please wait while we try to reconnect.