Hi,
it is not possible to connect the CAN nodes directly. CAN transceivers have to be used.
CAN has several different physical layers that can be used, each specifying certain characteristics of the CAN network, such as electrical voltage levels, signaling schemes, wiring impedance, maximum baud rates, and more. Over the course of the last decade, two major physical layer designs have emerged as those used in most CAN applications. They both communicate using a differential voltage on a pair of wires and are commonly referred to as a high-speed CAN (ISO 11898-2) and low-speed CAN (ISO 11898-3).
High-speed CAN networks are implemented with two wires, allowing transfer rates up to 1 Mbps. High-speed CAN networks (linear bus) are terminated with 120 Ohm resistors on each end of the network.
Low-speed/fault-tolerant CAN networks also use two wires. They can communicate with devices at rates up to 125 Kbps and offer the ability for CAN data traffic to continue in the event of a wiring fault. In low speed/fault tolerant CAN networks (linear bus, star bus or multiple star buses connected by a linear bus), each device has its own termination. Termination at each node is a fraction of the overall termination resistance which should be about 100 ohms but not less than 100 ohms.
Due to different termination, high-speed CAN nodes (like TJA1050) and low-speed/fault-tolerant CAN nodes (like TJA1055) are not compatible.
See please this document where you can find which transceivers are for high-speed CAN and for Low-speed/fault-tolerant CAN:
http://www.nxp.com/documents/line_card/75017131.pdf
When you select a transceiver which fits your needs, see its datasheet for more details - you can find example circuit in a datasheet. For example:
http://www.nxp.com/documents/data_sheet/TJA1050.pdf
Regards,
Lukas
