cancel
Showing results for 
Search instead for 
Did you mean: 

NUCLEO-F767ZI External Power Start Up Problem

Yilmaz
Associate II

 Hi,

I have Nucleo-F767zi . When I give it an external power source through Vin. It doesn't always wake up. The board works if I reset it.

The graphic of my power supply;

 power supply.PNG

It arrives 9V at 400 mili seconds. The curve is almost linear.(No glitch on the wave.)

 

There is no issue if I use USB.

I switched up my external power sources. However, the outcomes remain the same.

Is there a recommendation?

 

Thanks

1 ACCEPTED SOLUTION

Accepted Solutions
Rubb
Associate II

@Yilmaz,

Think I found it after reading these:

https://stackoverflow.com/questions/76424572/nucleo-board-needs-reset-when-powered-externally
https://community.st.com/t5/stm32-mcus-boards-and-hardware/stm32f767zi-on-external-power-supply-is-stuck-until-reset-button/td-p/115194

And the user manual (UM1974) for nucleo boards where it states: 

Rubb_0-1725137544497.png

 

So it by default uses clock from the ST-LINK, X3 isn't assembled on the board and left empty.
It seems as it takes a while for the MCO to output the clock at power-on, with a reset the clock is powered on already and it therefore works.

This is also why it works if one connect the USB cable first and then power on E5V, then the clock pulse is already there and the board can boot up as normal.

If one doesn't want to be dependent on MCO one have to solder on the board to fit X3 and the capacitors and also set the correct solder bridges or use another external oscillator, or use the internal clock HSI.

 

EDIT:

If you don't care to much of startup time it is possible to increase the timeout for the HSE to wait for MCO.
In CubeMX this setting is found under RCC -> Parameter Settings -> HSE Startup Timout Value (ms).
I set it to 2500 ms and now my board boots at the first try with E5V and no USB connected.

View solution in original post

5 REPLIES 5
Andrew Neil
Evangelist III

So what does the ramp-up of the actual supply (3V, or whatever) to the STM32 look like?

 


@Yilmaz wrote:

The curve is almost linear. 


You think so? 

Looks exponential to me - tending towards around 13V.

Rubb
Associate II

@Andrew Neil , @Yilmaz 

I believe I have the same problem.
I power my board through E5V, have set JP3 to E5V and JP1 is off. 
No USB cable connected. 

If I start my power supply I can see that the continous current is around 150 mA, the card isn't fully on and don't work. 
If I then Reset the board on the button I see that the current jumps to a continous current of around 300 mA and my board starts to work. 
I tested this with only Ethernet connected and no other peripheral connected to something. 

Could it be that the ramp-up time of the power supply is to slow? If so, is it possible to delay the start-up of the Nucleo board or do one need to focus on fixing the power supply ramp-up?

EDIT: I tried to have the power supply on continously and then connect the STM32, it still don't boot the first time and get stuck on 150 mA. 
However, power off and then power on just a few seconds later it starts directly and draw 300 mA. To me this feels that there is capacitors charging in the beginning that prevent the board from getting enough current to boot. 
As it happens even when the power supply is started in advance I believe these are the capacitors on the Nucleo board and the power supply doesn't supply enough in-rush current.  Need to take the time to hook up the oscilloscope later to see if I can see something in the voltage levels/currents. 

Rubb
Associate II

@Yilmaz,

Think I found it after reading these:

https://stackoverflow.com/questions/76424572/nucleo-board-needs-reset-when-powered-externally
https://community.st.com/t5/stm32-mcus-boards-and-hardware/stm32f767zi-on-external-power-supply-is-stuck-until-reset-button/td-p/115194

And the user manual (UM1974) for nucleo boards where it states: 

Rubb_0-1725137544497.png

 

So it by default uses clock from the ST-LINK, X3 isn't assembled on the board and left empty.
It seems as it takes a while for the MCO to output the clock at power-on, with a reset the clock is powered on already and it therefore works.

This is also why it works if one connect the USB cable first and then power on E5V, then the clock pulse is already there and the board can boot up as normal.

If one doesn't want to be dependent on MCO one have to solder on the board to fit X3 and the capacitors and also set the correct solder bridges or use another external oscillator, or use the internal clock HSI.

 

EDIT:

If you don't care to much of startup time it is possible to increase the timeout for the HSE to wait for MCO.
In CubeMX this setting is found under RCC -> Parameter Settings -> HSE Startup Timout Value (ms).
I set it to 2500 ms and now my board boots at the first try with E5V and no USB connected.

Hi @Rubb;

Thanks for your interest;

We have putted X3 and capacitors. And Changed resistors.

Before Configuration Changes, Our Scope Outputs;

Before Configuration Vdd-NRST.PNG

(External 12V  Power)

Before Configuration Vdd-MClock External 12V Power.PNG

After Configuration Changes Our Outputs; (External 12V  Power)

After Configuration Vdd-MClock External 12V Power.PNG

 

 

 

Don't know if I fully follow, are you still facing problems?


If you have soldered on X3 and the capacitors.
I think you should only need to do some solder bridge changes:

Open SB149
Close SB8
Close SB9

Then you shouldn't have MCO at PH0 and it shouldn't be using it. After that you should also activate X3 in configuration. In CubeMX RCC -> HSE Crystal/Ceramic Resonator, to activate Osc Out (PH1).