STM32H747IGT6 - Extreme Heat

[Caller89]

Hi All

Hope everyone is well,

Im working on a large and expensive project that requires dual core processing, therefore prior to committing to the cost of my project I created a custom dev board  to ensure I can achieve what is required for my project.

My dev board I am soldering components using stencil and reflow plate. The Chip connects to Cube PRG and runs code I upload

When powering my board via USB (5v VBUS) the STM32 is getting extremely hot (Burn your finger hot), I have checked the following. 

1. VCAP and VCAPDSI tied to GND via 2.2UF CAPS

2. Don't appear to have any bridges across pins / pads , iv checked under microscope.

3. Board is drawing current of 1.2amps when using USB.

4. I am getting continuity between 3.3v and VCAP pins, which suggest I have a short, I can only think maybe a bridge on the rear of the pins that I cannot see.

5. When powered on Im reading 1V on the VCAP pins, suggesting the VCAP are being pulled done by over current.

6.  D+/D- pins when powered , no 5V leakage.

7. When I removed the inductor between the board and 3V3 buck converter and powered the board using bench supply the VCAPs measured 1.364V, board was getting hot however not as hot. Board then drawing current of 500mA, however idle.

I believe my issue is related to soldering/bridging, which is creating a short,  therefore Im going purchase more boards fully assembled by JLCPCB, however before I do so I would appreciate the community review my design below in case Im missing something obvious, thanks in advance.

[AScha.3]

Hi,

seems mostly ok, but whats wrong:

you connected Vdd12dsi to 3v3 , killing the chip. or create big current...if still alive.

+

you connected Vlxsmps to GND, but it should be open.

-- So try: disconnect (cut pin ) Vdd12dsi and Vlxsmps . See if chip still alive...or replace with new chip.

[percy4458]

Hey Caller89! Yeah, that kind of heat isn’t normal for the STM32H747 — even under load it shouldn’t burn your finger. Since you’re seeing around 1 V on the VCAP pins and drawing 1 A+ from USB, it really does sound like a short or bad solder joint under the MCU. Continuity between 3.3 V and VCAP is a big red flag — those pins shouldn’t be tied together directly.

Before sending it to JLCPCB, maybe double-check your stencil alignment and pad clearances, and inspect the underside of the chip with X-ray or hot-air rework to resolder it. The H747 is super sensitive to even tiny bridges between power domains. If your schematic and caps are correct (2.2 µF to GND on each VCAP), then yeah — this points to a hardware short rather than a design flaw.