Why it exists
By the 2010s, classical CAN was running out of room: more signals, software flashing and security overhead all needed more bytes per frame. Replacing CAN entirely would have been expensive. CAN FD keeps everything engineers liked about CAN and raises the payload and speed.
What it is
- Same arbitration and error handling as classical CAN.
- Up to 64 bytes of payload per frame (instead of 8).
- Two bit rates: arbitration at the normal rate (e.g. 500 kbit/s), then a faster data phase (commonly 2 Mbit/s, up to 5 Mbit/s or more with suitable transceivers).
- Can coexist on networks with classical CAN nodes only if those nodes tolerate FD frames — in practice networks are migrated as a whole.
With BRS set, everything from BRS up to the CRC delimiter is sent at the faster data-phase bit rate.
How it is used
Brake, steering, airbag, battery management and inverter ECUs typically communicate over CAN FD with their zone controller or domain controller. Bigger frames allow grouping many signals into one PDU and adding SecOC and E2E data.
Example sizes only. E2E profile, truncated freshness and MAC lengths are configured per PDU.
When to use it
For MCU-based ECUs with control traffic, diagnostics and security needs. For bandwidth beyond a few Mbit/s or IP-based services, use Ethernet; CAN XL bridges the two.
Best practices
- Choose data-phase rates per network based on topology length and stubs; validate with physical-layer measurements.
- Use the extra payload for SecOC and E2E rather than squeezing them.
- Keep a consistent PDU design across platforms to simplify gateways.
Common pitfalls
- Mixing classical-CAN-only nodes into an FD network.
- Pushing data-phase speed too high on long, branched topologies, causing ringing and error frames.