Non normal operation of STM32F103x

[KUHAN JON]

Posted on February 07, 2018 at 09:31

Hi, everybody.

I have a trouble in using STM32F103RET6 MCU.

The problem is as follows.

My board has one STM32F103RET6 as a controller and uses SPI2, UART2, 4 channels of ADC and some GPIOs.

I downloaded the simple firmware using JLINK/SW mode.

The board worked well for some minitues and after sometime, looked as halted.

I doubt if MCU might be breakdown and so tested the resistance between VDD and GND using multitestet diode testing mode.

The resistance was measured as 50 or 200 or 19 in both direction.

But i found that after keeping the board as power off for several hours, that resistance was came back to the value of 412 in reverse direction and 1550 in forward direction.

I replaced with new chip but that phenomena was occured repeatedly.

I used LP5907MFX-2.8(2.8V/200mA LDO) as a power supply for MCU.

I cannot understand this phenomena.

Please help me to explain this phenomena.

Thanks.Kuhan.

[waclawek.jan]

Posted on February 07, 2018 at 09:43

The blame is always put on the most expensive part.... What about a reversed polarized capacitor?

JW

[S.Ma]

Posted on February 07, 2018 at 09:52

Any hot components on the board when things turn ugly?

[henry.dick]

Posted on February 07, 2018 at 12:06

'

I cannot understand this phenomena.'

it is definitely ***NOT*** that simple fireware.

Look somewhere else,

[AvaTar]

Posted on February 07, 2018 at 12:11

it is definitely ***NOT*** that simple fireware.

Look somewhere else,

My initial thoughts, too.

Does a debugger lose contact to the MCU during those 'events' ?

Or might it just be stuck in the hardfault handler ... ?

[KUHAN JON]

Posted on February 08, 2018 at 02:44

Hi, thanks for your replies.

Yes, AvaTar.

When STM32F103RET6 is halted, the debugger loses contact to the CPU.

And i add some information.

When that MCU was halted, i desoldered all peripheral componenets such as capacitors and LDO.

But the resistance situation was still measured as those wrong value.

I keeped it for 5 hours. After 5 hours, the resistance between VDD and GND was repaired to the normal values.

I again soldered the new peripheral componenets and tested.

That strange phenomena was repeated.

I am on verge of collapse!!

perhaps is the system clock unstable?

But anyway, i cannot analyse that phenomena.

I have other STM32 series board such as STM32F103VET6, STM32F103C8T6.

I tested those board, but those are all normal.

[KUHAN JON]

Posted on February 08, 2018 at 05:26

Hi, Clive One.

Thanks for your reply.

Yes, I have a schematic.

And i have not dirved any of the pins with 5V levels.

I will attach the schematic.

https://pan.baidu.com/s/1dHnjKNv

[Tesla DeLorean]

Posted on February 08, 2018 at 04:27

Do you have a schematic for this board?

Do you drive any of the pins with 5V levels?

[AvaTar]

Posted on February 08, 2018 at 07:56

When STM32F103RET6 is halted, the debugger loses contact to the CPU.

This would suggest there is a hardware-related issue during runtime.

Beside jan's recommendation about reverse soldered caps, I would check for  schematics errorss or placement errors as well.

Perhaps you switch a GPIO that inadvertently shortens the power supply via attached components.

Or the supply falters when you switch to a PLL-driven, higher clock rate.

Stepping through the code might help.

When that MCU was halted, i desoldered all peripheral componenets such as capacitors and LDO.

But the resistance situation was still measured as those wrong value.

Don't rely too much on such devices.

In the resistance setting, those multi-test meters apply a low voltage and measure the resulting current.

That fails miserably with nonlinear components, even more with components containing residual charges.

[KUHAN JON]

Posted on February 08, 2018 at 14:11

Thanks,Avatar.

I will do as you said.