Has anyone experienced regular board failures with a #RaspberryPi 4?

I have many of them around, more or less half of them run RPi OS and half Arch ARM, but all with quite similar configurations and running similar software.

Some of them have lasted years so far, others have failed only when the SD card failed (only 2 failures in 7 years btw, over a total of a dozen of RPis), but in most the cases the boards have been the same since I first purchased them in 2018.

But I've got one in my living room that I've been replacing at least one every 1-2 years because *the whole board* at some point fries up.

I've had the same SD card with the same RPi OS installation for it since 2018.

At some point, every 1-2 years, with no notice, sometimes even while no meaningful activities are running on it, the unfortunate RPi that runs in my living room just dies.

Red and green LEDs stay still without blinking. USB-connected devices don't even get powered up. The system doesn't boot and nothing is sent to the HDMI output. The boot-time VGA test with the color square isn't shown. The EEPROM diagnosis doesn't even start when booting it without SD card. But if I put the same SD card in another device, it all works again without problems.

Since it happens also when no CPU-intensive stuff is happening on the device, and the core temperature is overall normal (<70C), I don't think it may be due to overheating (unless it's an unlikely instance of hardware slow fry).

The first time it happened I thought it was just a bad board. The second time, just bad luck. The third time, extremely bad luck. But today a RPi4 running on the same SD card died again, after less than 2 years of usage, for the fourth time, and in the same way. Since no other devices ever gave me so much trouble, I start to see a pattern.

The only meaningful differences from other devices I run are:

- This one is connected to an always-on screen over HDMI.

- It has 8-9 devices connected over USB (but nothing very exotic: an Arduino, a Z-Wave gateway over USB, a USB mic, a soundcard, an additional Bluetooth dongle, a USB camera and a couple of USB storage sticks).

- It sports a #Pimoroni Breakout Garden for easy expansion over I2C/SPI, and I have a couple of breakout sensors connected to it (an LTR559 for proximity/luminosity, a BME280 for temperature/humidity, a PMW3901 for motion, and a MLX90640 thermal camera).

Since the only meaningful differences from my other devices are in the hardware expansions, I suspect that some combination of I2C/SPI extensions may, in some very unlucky conditions, send voltage spikes that may screw up the board. Or maybe it's the combination of numerous USB peripherals + several breakout modules that puts too much stress on the power supply? But in that case it doesn't explain why the failure only occurs after so long.

Since the board is completely fried when this happens, I don't have many ways of doing diagnostics - and 4 data points so far, separated by 1-2 years, are still insufficient to see a clear pattern.

Has anyone who has toyed with I2C/SPI breakout extensions (and specifically with any of those I've mentioned above) also experienced this type of board failures?