STM32C542RCT6 CANFD is not working with internal clock

[harishDST]

Hi,

I am working on CAN FD communication using the STM32C542RCT6. CAN FD is working properly when I use an external clock source. However, when I switch to the internal clock, the controller starts reporting errors such as Error Passive, followed by Bus-Off.

My suspicion is that the internal clock may have a slight frequency deviation, which could be affecting CAN FD communication. Since this is a new controller, I would like to understand whether the internal clock can reliably support CAN FD communication or whether an external clock is required.

My current CAN FD configuration is:

• Arbitration bit rate: 500 kbps

• Data bit rate: 2 Mbps

• Bit Rate Switching (BRS): Enabled

Could you please clarify whether the internal clock can support this configuration and what could cause the controller to enter Error Passive and Bus-Off states when using it?

[Andrew Neil]

when I switch to the internal clock, the controller starts reporting errors

That is to be expected:

 

My suspicion is that the internal clock may have a slight frequency deviation

So look at the datasheet - the accuracy will be specified there.

Then you can check if it meets CAN’s accuracy requirements ...

[LCE]

The internal clock / HSI is only usable for basic operation and some “slow” or clocked interfaces.

For CAN, audio, ethernet and some other peripherals: never ever use HSI.

[Ozone]

First, I need to second that. HSI doesn’t work for CAN, amongst others.
 

> However, when I switch to the internal clock, the controller starts reporting errors such as Error Passive, followed by Bus-Off.

Which means the CAN peripheral receives all messages, but discards them as erraneous or corrupted.
Consequently it disconnects itself from the bus.

[mƎALLEm]

Hello ​@harishDST and welcome to the ST community,

From this interesting application note: AN6363 “Introduction to clock requirements and calibration for STM32 MCUs” / Table 3, the recommended accuracy for FDCAN needs to be  =< 0.65%:

So what clock are you using: HSI or PSI? So according to that application note, you must use either a crystal in direct or via PSI i.e:

Knowing that the HSI drift according to the datasheet is +/-1% (TJ = -20 to 130 °C):

You can also refer to the section 2.2 FDCAN from the same application note for more clock accuracy calculation details. (especially the section 2.2.2 Oscillator tolerance).

Hope that helps.

[LCE]

​@mƎALLEm  I don’t know the C5, as far as I understand the PSI can only be as good as its source (why called “reference”), so it should be made clearer that for a stable PSI a stable HSE is needed - or am I wrong ?

[mƎALLEm]

From the table 3, it was said “in direct or via PSI” :

That means need to use a crystal: either directly or over PSI. Crystal = stable and accurate clock source.