STM32F303CB - Low DC resistance between VDD and VSS (electrical short circuit) due to unknown reason

[mechatron]

Hi,

I have a setup on a PCB that resembles the following.

The 5V supply is pretty stable (0.5V tolerance) and the 3.3V LDO delivers a low-noise power to the microcontroller. The µC communicates with the outside world through CAN (transceiver present). There are already some precautionary measures present on this setup, as follows.

1. All the lines (power and CAN) are ESD protected (IEC61000−4−2 standard level-4.
2. Setup is completely coated (conformal coating on PCB). Hence, any physical contact from outside world will be difficult.

During routine operation, the communication with a sensor system (STM32F303CB on-board) was lost. On further investigation, the following was found.

1. Optical examination revealed no bad component. All the active and passive components on the PCB seemed intact. No burning out (or) abrasion on coated areas were found.
2. The resistance between 3.3V LDO output and Ground was <2ohms, which was narrowed down to the microcontroller itself.
   1. The resistance between VDD and VSS pins of the controller were between ~1.9 to ~6 ohms.

Table 1 - DC resistance measurement overview

Test Pin	Resistance against GND (ohms)
Pin 24	1.9
Pin 36	4.5
Pin 48	5.9

Note - During resistance measurement, the VDD pins were not in contact with the PCB or neighbouring pin (short is indeed "within" the controller).

Figure 1 - STM32F303CB pinout

I looked into possible sources and found that through any ESD events directly on uC pin (not through the input connector which is already ESD protected) or any over-voltage event outside the specs can cause the protection diodes within any GPIO to break and make a short circuit. But through several inspection (optical and electrical), I could find no evidence of such an event. 

Any ideas on why such a short might occur? If so, is there any other way in which I can find the cause for it?

Thanks!

[MasterT]

My guess is that power sequence was not respected. According to note 1 p. 58/149 DS9118 Rev 14:

VDDA must power on before or at the same time as VDD in the power up sequence.
VDDA must be greater than or equal to VDD.

Since VDD & VDDA divided over PI-Filter (What is that ?, and what P/N of LDO?) you have to ensure VDDA never falls before VDD. It could be as easy as installing one schottky diode or special power management IC.

[mechatron]

Hi.

The Pi-Filter consists of one ferrite bead and two capacitors (one on either side of the ferrite). So, the VDDA is always slightly higher than the VDDD. And both VDDA and VDDD are powered almost at the same time at the µC. Moreover, the setup failed during the operation mode, and hence, the power sequence plays little role here.

[Ozone]

Perhaps some EMI  glitch on a GPIO pin blew the internal protective circuit, and shorted it.

Since the board seems destroyed anyway, you could try to sever connections from GPIOs pins to the surrounding circuitry one by one, and see if the short circuit is gone. I would start with those connected to the "outside", like CAN.

[Andrew Neil]

---

@mechatron wrote:

a PCB that resembles the following.

---

Not terribly helpful - details matter!

Please post the schematic.

If it can't be posted in public, perhaps raise a support case direct with ST: https://ols.st.com/

Or contact your FAE or Distributor: https://www.st.com/content/st_com/en/contact-us.html 

[mechatron]

Hi. I will probably raise a support case directly with ST. Thanks.

[Andrew Neil]

Don't forget to report back here when you're done.

https://community.st.com/t5/community-guidelines/help-others-to-solve-their-issues/ta-p/575256

[mechatron]

For sure. Thanks for the reminder.