The noise tolerance, and sometimes even operation of the CAN bus depends on

* termination (well known, typically 120 Ohm at/near each end of the differential pair).

* voltage BIAS

----

Voltage BIAS

Unlike the RS485 transceivers, the CAN transceiver only

* pulls the CANL signal LOW (when actively transmitting)

* pulls the CANH signal HIGH (when actively transmitting)

TJA1050 has weak 25k bias toward 2.5V (VCC/2), but pullup/pulldown would seem desirable:

* CANL: 10k pullup to +5V

* CANH: 10k pulldown to GND

It might work reliably without these pullup/pulldown (it is easy to add externally, and the dash likely has it anyway)

* actually it worked well during ALL field tests, never experienced any problem with any setup in car or in the lab due to the lack of these bias (like in this [circuit].

* but it sounds safe and friendly to just include 10k (or at least a weaker 20k if very cautios).

* or at least some analog OscilloScope verification would be nice for a couple of installs, just in case. Eg measuring h/l voltage levels for both CANL and CANH, taking notes. Also measuring with known bias resistors, eg. 10k (or 4k7 as experiment)

----

TJA1050 => Saleae

A spare modul with TJA1050 can be used as a CANH/CANL differential reciever, the saleae can capture pin4 RXD data output (with good sample settings).