TCAN1043-Q1: nFault toggling

*TCAN1043A-Q1

Hi Marcus,

Thanks for clarifying the exact device. The nFault pins behave slightly differently across the A and non-A versions.

What data rate are you operating at?

How is the CAN bus terminated?

In order for nFault to report properly, there are certain conditions that need to be met (this is detailed fully at the end of section 8.3.7.1.4 of the datasheet).

Best,

Ethan

Hi Ethan,

Data rate is CANFD 500 / 500 kbps.

BR

Marcus

Thanks for that Marcus.

So technically the tCBF time dictates that 400 Kbps is the maximum data rate where this device can reliably report errors. However, 500 kbps is close enough and *should* still work fine. Operating above 400 kbps just means that the nFault will not be able to capture every fault perfectly at higher data rates. 

On your schematic, I see you have effectively ~120 Ohm termination. During your test, what was connected to the CANH and CANL lines? Is there another 120 Ohm termination present somewhere on the line?

The parameter spec R_CBF also defines that valid characteristic impedance that the CAN bus should have is 45 to 70 Ohms. In a properly terminated CAN bus, there should be two 120 Ohm resistors at opposite ends of the bus, which is an effective resistance of 60 Ohms. 

Also as a follow up question, in your original post you say that CANL is shorted to GND first, then you say CANH is shorted to GND. Can you clarify which is showing this behavior?

Hi Ethan

Thank you for claryfing the maximum data rate.

Yes, 120R termination is used at the other end as well.

Sorry, it should be always CANL. For CANH everything is fine.

BR

Marcus

Hi Marcus,

Are there any other nodes that would be affecting the total characteristic impedance? 

I would re-test with a slower data rate to see if this issue resolves itself. 

As some further background for nFault, a simple way to think of it (but internally much more sophisticated) is that the device is adding up the currents on each of the dominant transitions. If it reaches a certain current threshold at a time greater than tCBF, then the error flag is raised. The reason your nFault pin is toggling is because the current threshold is just barely triggering and then resetting. Out of all the CAN bus fault conditions (CANH shorted to VSUP, CANL shorted to VCC, etc), CANL shorted to GND accumulates the smallest amount of current across these 6 tests.

Additionally, TCAN1043A will still likely communicate despite CANL shorted to GND. Since CAN is differential, the CANH line will still transmit and a differential signal can still be read.

Best,

Ethan

Hi Ethan,

"Are there any other nodes that would be affecting the total characteristic impedance?" No, there are no other nodes.

We tested again at 400Kbps (problem is still seen) and 250Kbps (toggling is not seen).

BR

Marcus

Hi Marcus,

That is interesting. Could you send a new waveform with CANH, CANL, and nFAULT while at 400 kbps? Please zoom in enough on the time domain (less than 50us/div would be ideal) when nFault toggles high and when it toggles low. Two separate waveforms for each might be best.

Could you also share the kind of cabling you are using for the CAN lines? A datasheet here would be great so I can check for impedance specifications. 

-Ethan

Hi Ethan,

sorry for the delay.

CANH/L measured at connector side, not CAN receiver side.

Yellow: nFault

Blue: CAN H

red: CAN L (shorted to GND)

Thank you

BR

Marcus

Hi Marcus,

I am catching up on E2E threads as I was out of office. I aim to get a response to you tomorrow.

-Ethan

Hi Ethan,

we see it on multiple ECUs.

Just for understanding, it is the desired behavior of the TCAN1043A that the nFault pin changes between high and low, although the short circuit is still present?

Marcus