STM32F405 behavior when VCap pins aren't connected

[Agatti]

Good morning,

I'm using STM32F405 for my embedded project and this PCB version showed itself a real MCU killer... So far, it has already burned 4 MCU. 

Here's my situation. It's not my first time designing using an STM32 (I've already used STM32F103), but it was my first time using a PCB with this particular MCU, so it was my first time dealing with VCap pin. During project phase, I misread the datasheet and connected the VCap to the VDD 3.3V source power, so whenever the MCU was soldered to the PCB, I had 3.3V and GND short-circuit. To get around this situation, I bent up both VCap pins so they didn't touch the pad, leaving them floating. It worked, the MCU turned on, sink recognized it and it was working as expected.

But something weird started to happen. My PCB acts like a central communication board that interconnects other PCB peripherals (via UART) with the computer via USB. Two PCB peripherals controls motors, so I have to power then with 24V and 5V, besides the communication pins. They work like a charm when debugging both separated. I already ran some tests with both connected and powered at the same time, it has already worked as expected, but sometimes it just kills the MCU from the main board.

Some PCB power information:

• Just the main PCB, no peripheral connected: 55~70mA

• Main PCB + 1 peripheral: 150~160mA

• Main PCB + 2 peripheral: ~250mA

• only Main PCB connected after MCU burned: ~280mA

What is bothering me is this inconsistent behavior. It works for quite a while, then, out of nowhere, burns the MCU. I've tried pretty much everything while debugging to solve this problem, but the only conclusion I can come up with is because the VCap pin is floating.

I already made another version for the PCB with correct power supply (and a few other modifications), but I'm still waiting it to arrive. Meanwhile, I'm researching about this VCap to understand why it messed up so bad with my PCB.

Can someone help me explaining the VCap pin function and what should happen if it's not connected and/or floating? I've already read some forums posts and the datasheet, but all I know is that it's part of the internal voltage regulator. Is there any information beyond that?

[TDK]

If VCAP is floating, the chip generally won't work at all. It may work sometimes if the load on the internal logic is low, such as with very low clock rates and not a lot of internal switching happening. But if it's not connected, no guarantees are made on how the chip will behave.

Cut your losses and redesign the board. Feel free to post schematic here for review before you have it made. Consult the hardware design guide for helpful information on laying out a board.

https://www.st.com/resource/en/application_note/an4488-getting-started-with-stm32f4xxxx-mcu-hardware-development-stmicroelectronics.pdf

[SofLit]

---

What is bothering me is this inconsistent behavior. It works for quite a while, then, out of nowhere, burns the MCU.

---

I don't think burning the MCU has something related to Vcap pins. It's more likely related to your level of isolation between low level and high level voltages since you have a motor to control @24V.

But Vcap pins must be connected to the GND through capacitors with a low ESR:

Without these capacitors you face issues and instability of your MCU.  

[Agatti]

This is pretty much the power supply for the PCB. It has a filtering stage, a buck 12V/5A for another motor (has a completely different PCB project from the 24V one, consumes up to 5A) and the LDO regulators stage. The board itself is powered by 24V from a medical power supply device (127Vca / 24Vcc). 

Both 24V and 12V are sent directly to theirs respective PCBs. Should I use any method to grantee some level of protection against peak voltage fluctuations?

[SofLit]

You can for example refer to this Application note: https://www.st.com/resource/en/application_note/an4694-emc-design-guides-for-motor-control-applications-stmicroelectronics.pdf