MCP2515 and MCP 2551 on a breadboard - troubleshouting

I am stuck with my project.
I have a working and tested sketch for an Nano working with one of these china CAN breakouts.

I than changed the bootloader of my Nano to output the internal clock (8mhz) to use it to clock a standalone mcp assembly.

I checked with scope- 8 Mhz looking fine.

I build up my circuit according to the schematic, but I dont get it to work. The TX of the mcp2515 is just high. If I send CAN data from outside to the MCP2551 I can see that the RX responds and outputs something but I have no acknowledge on the bus- just error frames if i send. I tested with and without termination, diffrent speeds- I think the CAN side is troubleshooted.

When I switch back to the china breakout is instantly works.

So back to the MCP2515, It has power on its pins, reset with 10k against 5V, CS looks clean in scope, SPI bus decodes on the scope. Somehow it is not awakening and i wonder do I miss something essential?

The breadboard picture: I started with a standalone Atmega328, but I removed that one and am running everything from a nano on another breadboard to use the serial output for troubleshooting and to faulttrace where the problem is.

I am using this Library:

I suspect 8MHz signals on breadboard circuits are not going to be reliable. Especially if you are using long wires to reach separate breadboards. The capacitances introduced by breadboards and flying wires are likely to corrupt the clock signal.

Maybe try having the nano and 2515 as close together as possible on the same breadboard, and connect them with short wires, laid flat to the breadboard.