Hi all, I've been working with the STM32H743ZI and lan8742AI, but when I connect RJ45, LAN8742 will overheat, even though only connect power, temperature also rises slowly, have anyone encountered a problem like this? thanks.

[CTing.1]

..

[Piranha]

No, that PHY works nicely and stays cool on ST's Nucleo and Discovery boards, my custom boards and countless other boards in the whole world. Most likely you have some hardware problem.

[CTing.1]

Sorry, my description is not clear, 

the PCB is a layout by myself,

here are my schematics below,

red X is not used,

now I test result is, RJ45 plug/unplug LED shows normal�?DHCP get successfully, and I use MQTT send/ receive normal,

but the temperature will rise(~50 degrees) after plugging RJ45,

I have no idea what problem is.

​

and I have another problem, NRST is necessary to set IO in stm32 MX or not?

if I set IO output, I reset(section of USER CODE ) when restarting is correct?

​

​

[Piranha]

> the temperature will rise(~50 degrees)

C or F degrees? Absolute or rise above the ambient temperature? Anyway even 50C with an ambient temperature 25C seems to be too much.

Double check the RJ45 socket pinout. Double check the PHY pinout, wiring, PCB routing, heat transfer design on the PCB. Check the solder quality, resolder the same chip, replace the chip.

> NRST is necessary to set IO in stm32 MX or not?

Either you just connect it to the NRST of MCU, or connect it to some GPIO and do a reset with a software. Unless you have some special requirement of PHY chip NOT being reset, when the MCU is reset, connecting PHY NRST to MCU NRST is easier to manage and does not require an additional GPIO pin.

[CTing.1]

Sorry I have not replied for long,

about ethernet layout, I have not modified anything, I really have not detected any other bugs,

besides ethernet, I spent a few weeks re-layout the other my all PCB, finally received it last week,

I test a week, the temperature �? NRST software reset is well,

maybe the problem is the PCB production process or the other circuit, everything looks good now.