STM32MP157C-DK2 Long Turn On Time

[ASuar.1]

I'm working with the STM32MP157C-DK2 dev board and I've noticed that sometimes when powering up the board there is a 20-30 second delay where I don't even see the boot up messages. It's doesn't appear to be U-boot related either.

Has anyone else notice this? and if so were you able to fix it?

[Olivier GALLIEN · Accepted solution]

Hi @ASuar.1​ 

From your description I think you are facing a known restriction due to STPMIC V1.2 assembled on our DKx board.

Depending of your STPMIC sample you may encounter a delay between 0 and 20s before the STM32MP1 to be properly powered and boot.

This only appears on very cold boot. ( board shutdown for a while)

There's no workaround.

The issue is fixed in STPMIC V2.0 now available for new design.

Olivier

[ASuar.1]

That's good to know. Thank you very much I'll look out for that.

[Mariano Abad]

is V2.0 a silicon revision or a different IC?

[Olivier GALLIEN]

Hi @Mariano Abad​ ,

Sorry for misleading words, I meant STPMIC1 cut 2.0.

New silicon revision of same IP

Olivier

[MVass.1]

Hi @Community member​ ,

I was also experiencing this issue on my 157c-dk2 and 157a-ev1 boards every morning the first time I plug them in. (between 5-10 seconds delay)

I am glad that the new revision fixes this issue, but I have another related question regarding this issue:

I noticed that regular cold boot also carry a small delay, about ~700ms

We plan on using stm32mp157c with custom board, and one of the requirements is that we need to perform a certain action within 3 seconds from getting power. so boot speed is very important to us.

We did tests with slow motion video of turning on the power source, a milliseconds precision timer and console logs.

And we saw that it takes about 700ms from LED of power source turning on to first console prints of TF-A "NOTICE: CPU: STM32MP157CAC Rev.B"

You can probably contribute 100ms out of this delay to time it takes for the power supply's capacitors to charge.

Still it doesn't make sense that ROM code takes this much time.

Do you think this issue is also related to old STPMIC revision ? (The this delay will also be smaller on new HW)

Thanks,

Michael

[PatrickF]

Hi,

according to latest STPMIC1 datasheet, time when VIN is available and NRST is released is about 16ms (see Figure 56. "Auto turn-on condition sequence" and Figure 54. "STPMIC1 POWER_UP and POWER_DOWN sequence example").

Then one NRST is released, there is few internal init (about 10ms) and BootROM start to init SD-Card (this is done at 400KHz) and fetch TF-A (about 247kbytes @16MHz 1 bit, which is about 160ms).

We can say this should take about 200ms. You should see similar time when issuing a pulse on NRST (with button or watchdog reset).

Latest PMIC revision might behave as STPMIC1 datasheet details (thanks to correction of the bug). This device is present on all 800MHz boards, such as STM32MP157F-DK2, 157D-EV1, etc...

See also https://wiki.st.com/stm32mpu/wiki/How_to_optimize_the_boot_time.

Regards.

[SPele.1]

Hi,

Can I know from PMIC package date code what is the version of ST PMIC?

According to the package - assembly year is 0 and assembly week is 22. it means that We have version 1 PMICs?

Thanks!

Shlomi

[ChristopheB]

Hi Shlomi,

It is difficult to check the version of the PMIC package according to the marking code. If you provide the complete marking code, I may be able to provide you the PMIC version.

Nevertheless, the best way to get the real PMIC version is to check the linux consol during boot and find string "PMIC chip version" during boot: That read the silicon version of the PMIC via I2C at address 0x06 "VERSION_SR" (see STPMIC1 datasheet)