STM32H735IGK6 cooking when power is applied

[CMJ-PEL]

Hi All,

We have recently spun a PCB with a BGA packaged STM32 processor. Unfortunately I am having real problems with the board bring-up. The 3V3 rail is getting clamped to around 1.9V when I have a 100mA current limit on my PSU. A thermal camera has confirmed that the MCU is quickly heating up.

Our design is almost an exact copy of the MB1520 EVM design (with the exception of the SMPS pins). Would someone be able to take a look at the schematic and tell me if I've made a mistake in the design. I'm not ruling out a short on the BGA pins, but as all 5 boards have the same symptoms this feels unlikely (plus that would prevent Vdd from reaching 1.8V).

Our power schematic:

Reference power design:

Our other pins (in case the issue is here):

Any help on this would be greatly appreciated, fault finding BGA micros is not proving to be a fun experience.

Many thanks,

Colin

[AScha.3]

Hi,

really this :

You have VCAP at 3v3, the core supply, (1.2V typ.) , this will kill the cpu instantly.

If its still alive...disconnect vcap from 3v3 + give it the  2x 2u2 or 4u7 , whatever is recommended for your cpu, (2x 2u2 here), to have the decoupling caps for the core supply.

+ vddsmps at gnd - maybe a problem , i dont know, look in ds /rm whats recommended, if using LDO only.

Only 3v3 at vddldo .  Then check...

[STOne-32]

Dear @CMJ-PEL ,

main issue is VCAP pins , must have Vcore value around 1.2V  

 

having 3,3V will destroy the device for sure !

Hope it helps 

STOne-32

[STOne-32]

Our replies crossed :) , did not see . thank you very much Dear @AScha.3 

[CMJ-PEL]

Thanks for your assistance with this one.
This is going to be a real pain in the proverbial to rework.

While a PCB respin will now going to be required, it would be useful to try and use our current boards to find out what other issues may be present before going through the expense of a V2 build.

We can probably remove the BGA and get new ones fitted.
I am almost certain that we can disconnect the Vcap pads from 3V3 while doing this.
Connecting them together and fitting LDO caps on the other hand might be a nightmare.

Looking at the layout I might be able to add capacitance to one of the three Vcap pins.
The datasheet states that all 3 Vcap pins need connecting together.
While not ideal, what are the chances of them being internally tied together?

Many thanks,
Colin