nRF5340 takes too long from Power ON to start executing the main application

[neo_here]

Hi, we have a custom nRF5340 based device. Has NISB and MCUBoot Enabled. It takes about 2.5 seconds to get into the application and BLE advertisement to start.

Wanted to know how do we go about narrowing down the cause and potential solutions.

We are on nRF Connect SDK 2.6.1

[Elfving]

Hi, 

The first two things from the top of my head of things to try, is to see if you get the same results with the same sw on a nRF5340DK, and if you get the same with a default sw on your board. Do you have a DK? And could you try a sample?

Regards,

Elfving

[neo_here]

Was able to test a sample on the DK and our board

1) DK, Pretty quick most of the time, did see a couple instances where it took about 500ms but that was about 1 in 10 times. Most times it was pretty quick




2) Same Sample on our board, seems to consistently take about 1 second or more to first advertisement. This is much lower than 2.5 seconds but I'm guessing the bootloaders also add some overhead in that scenario. 




Do you think the Power/Clock subsystems are taking much longer to stabilize on our board? Also how to lower the time taken by the bootloaders themselves? 

Also does the current draw pattern suggest anything for our board for why it's taking so much longer. 

[Elfving]

neo_here said:

2) Same Sample on our board, seems to consistently take about 1 second or more to first advertisement. This is much lower than 2.5 seconds but I'm guessing the bootloaders also add some overhead in that scenario. 

I see that I could've been clearer there. In case there were any misunderstanding - I was wanting to try a default sample on your custom PCB. Is this a default one you just tried (ie. one directly from the SDK)? Does it react the same way on your board and the DK, chances are that your PCB is okay. 

Have you gotten your HW reviewed by us on DevZone?

Regards,
Elfving

[neo_here]

Yes, it's a default sample. It does not react the same way on the DK vs our PCB. 

On the DK it takes less than 100ms vs about 1.3seconds on our PCB

I believe the HW team did get the review on devzone. 

[Elfving]

neo_here said:

Yes, it's a default sample. It does not react the same way on the DK vs our PCB. 

In that case it seems like there is something odd/wrong/different on the PCB. Could you find the devzone case in which this was reviewed?

If not, could you upload the schematics here so it can be reviewed here? I can put the case to private so that only you/whoever you wish to and nordic will be able to see what is shared in it.

Regards,

Elfving

[neo_here]

I was able to find the Devzone case used by the HW team. 

devzone.nordicsemi.com/.../312445


Also, what would you suggest to investigate SW Delays, especially NISB and MCUBoot ? GPIO Toggle at different stages ? 

[Elfving]

neo_here said:

GPIO Toggle at different stages

Yes, try this.

neo_here said:

I was able to find the Devzone case used by the HW team. 

And your HW was not changed after this point?

Regards,

Elfving