What could cause a boot delay of 1-2s on 4/300 boards using STM32F7?

[SWint.2]

Hi,

We are using STM32F7 in our GUI board and we are experiencing an issue with 4 boards out of 300 where the processor is taking a longer time than usual to boot up. We have measured the time from when the board receives power to when the processor has powered up using an oscilloscope.

On a working board, the time is ~0.5s. On the failing 4 boards, it is 1-2s.

I have checked the errata documentation, reference manual and datasheet but I have not found anything which could explain this problem yet.

Does anyone have any ideas what could cause this boot delay?

Thanks in advance.

Sian

[TDK]

It's almost certainly the case that the processor begins executing code (almost) immediately after power-up and something in your program is causing the slowdown. Toggling pins in various stages of the boot-up process helps immensely with tracking down the cause.

One thing that comes to mind is if it's waiting for a clock to be stable, or waiting for another chip to begin communicating correctly.

[Tesla DeLorean]

+1 HSE Not starting (see also LSE, PLL)

Might also be weary of VCAP pins/caps

>>Toggling pins in various stages of the boot-up process helps immensely with tracking down the cause.

Processor is viable at Reset Handler, code some quick assembler code to enable GPIO clocks in RCC, and config/output some pins.

[Uwe Bonnes]

Look for bad soldered pins, resultiing in floating pins.

[waclawek.jan]

> to when the processor has powered up

What is "processor has powered up"?

JW

[Tesla DeLorean]

Assumption here is beyond, SystemInit(), main() entry, then SystemClock_Config() and HAL_Init() to start doing anything, so those would be candidates for closer inspection.