TCAN1043A-Q1: Transition from SLEEP to NORMAL

Part Number: TCAN1043A-Q1
Other Parts Discussed in Thread: TCAN1043

Tool/software:

 The following is note 5 in Figure 8.4 of Chapter 8.4 of datasheet. 

It's mentioned that 'To move to Silent mode, the nSTB pin must be high and a high-to-low transition must occur on EN."

So, if a high-to-low transition on EN does not occur (EN is always high), will TCAN1043 move to SILENT or NORMAL? 

THANKS.

 

  • Paul,

    That is correct, but only if Sleep mode is entered through the SWE timer elapsing. If Sleep mode is entered in the traditional way (through the EN and nSTB pins), then moving from Sleep to Silent would work just as the state diagram specifies.

    Regards,

    Eric Hackett 

  • Dear Eric,

    Thanks for your reply.  I'm confused about the following case:

    1st step:SLEEP mode is entered through the SWE timer elapsing;

    2nd step: EN is set high;

    3rd step: nSTB is set high after 2nd step. The time interval is about 1.5us.

    I think, in this case, TCAN1043A will remain  SLEEP mode because there is no transition as note5 mentioned below occuring.

    But in fact, TCAN1043A moved to SILENT or NORMAL in test .

    So, is it correct that TCAN1043A will remain  SLEEP mode in that case? AND, could you please think of any reasons for the performance?

    THANKS,

    Paul

  • Paul,

    If EN is set high and then nSTB, I would expect the device to go to Silent mode. Is it possible to share waveforms of the EN and nSTB and INH pins while this is happening?

    Regards,

    Eric Hackett 

  • Hi Eric,

    Waveforms are following:

    FIGURE 1

    FIGURE 2

    Before what happened in FIGURE 1, SLEEP mode is entered through the SWE timer elapsing. Waveforms in red block of FIGURE 1 was zoomed in FIGURE 2. We can see that EN was set high, and then nSTB was set high after about 1.5us, and then INH was transitioned to HIGH. 

    I have an idea that VCC transition in FIGURE 1, which means restart of TCAN1043A, is related to the performance. Is this right?

    Looking forward to reply.

    Regards,

    Paul

     

  • Hi Paul,

    The tMODE1 requirement is only a maximum value. There is no minimum set for this time, so under some operating conditions, it is possible that 1.5us is sufficient to register as the correct input for the sequence here. However, this may not be true across all operating conditions - at some voltage and temperature conditions the 1.5s timing may not be enough to guarantee the correct movement into normal mode. Therefore I must recommend that the system is updated in this case to allow for the full 20us to pass between the edges on EN and nSTB. 

    Note that the sequence that transitioned nSTB and EN to their initial states after the SWE timeout will also impact the requirements here as there is no time limit for this sequence. In this test are nSTB and EN always low from power on until the point that we see in the waveform? 

    Regards, 
    Eric Schott

  • Hi Eric,

    Yes, in this test, nSTB and EN were always low from power on. And I will pay attention to the tMODE1 interval in design. 

    So, could you please think of any reasons? If the VCC transition in FIGURE 1 I mentioned below is relevant?

     

    Thanks,

    Paul

  • Hi Paul,

    I can see in figure 1 that there is some movement on all signals t-40ms before Vcc rises. Do you know if this is noise from another signal or do the EN and nSTB signals toggle at this point as well.

    Please keep in mind that the sequence shown in the datasheet is notated to guarantee a transition from SWE-Sleep mode to normal/silent mode based on which parts of the digital core are accessible during this fault case. This is not the only sequence that can be used to get the device out of an SWE-Sleep case, so the system here showing an exit from sleep mode is not odd behavior to me. My concern is that this case may not be reliable under all operating conditions, therefore I would recommend a revision to the system so that the datasheet sequence is used so that the behavior remains reliable across all possible conditions. 

    Regards, 
    Eric Schott