I'm still having trouble sending CAN-FD traffic from a Teensy MM while still being able to receive it perfectly fine. To aid in determining what is going on, I've taken two captures with a Saleae Logic Pro. The bad one comes from the Teensy MM and the good one comes from an ESP32 driving an MCP2517FD chip.
That's the bad one. Here is the code I'm using to send:
Code:
CANFD_message_t fd_out;
fd_out.id = 0x789;
fd_out.brs = 1; //do the baud rate switch for data section
fd_out.edl = 1; //do extended data length too
fd_out.len = 16;
for (int l = 0; l < 16; l++) fd_out.buf[l] = l * 3;
canHandlerBus2.sendFrameFD(fd_out);
The important part to note here is that I'm sending a 16 byte message. Yes, I did use setRegion(64) and can confirm that there are only 14 mailboxes. But, you can plainly see that the hardware is sending a DLC of 4 which is why I get protocol errors. It's actually sending 16 bytes of payload but since the DLC section says 4 anything trying to receive it gets very confused.
Here's the good one from an ESP32. The Teensy receives this just fine. You can also see that DLC is 0x10 (16) which is the correct # of bytes because that unit is also sending 16 byte messages.
So, it seems my problem is that the Teensy side sends an invalid DLC but the correct actual # of bytes thereafter. The ESP32 sends the proper bits.
Both ends are set to 500k nominal and 4M data speed. Both captures are right at the CAN-TX and CAN-RX pins of the FD transceiver on the Teensy containing board.
I have tried different clock speeds for the CAN hardware on the Teensy. Both 24 and 60MHz yield this same result. The Teensy can send standard CAN frames perfectly fine through this same interface. Both ends have FD transceivers. I've tried a lot of things and nothing really seems to resolve the fact that one side can send FD frames to the other all day but the other direction never works for FD traffic. This certainly isn't a CRC problem like I first thought. Oh, also, look at the code above. The bytes sent should be 0, 3, 6, 9 but they're coming out as 0, 6, 12, 18 which is twice the proper value each time. It appears that things are getting desynced by one bit. Could this be some sort of TDC error or propegation delay problem?
Edit: I broke out the code to a much simpler sketch (the program was otherwise extremely long and complex) and then started to toy with the speed. It seems it works fine at 500k nom, 2M data so it looks like I might have some board layout issues or something that is subtly messing up the timing at 4M. Perhaps this could still be termination related or something to do with propagation delay or something of that sort (maybe even TDC). So, I guess it seems I may have figured out my trouble.
