Fried one of my test RPis by accidentally putting 5v from an Arduino onto a 3.3v GPIO pin, had a beer, and started over.

• Fried a small TFT display considered for using in a new product just yesterday, by accidentally swapping +3.3V and GND wires while setting up the test circuit. Damage: About 5 €. Lesson learned: Always measure current, and use a lab supply instead of relying on a development board to supply the device under test. Lesson re-learned: Double check supply pins (see below).
• Fried a custom prototype controller board on a test rig last year, by accidentally touching a wire connected to a 3.3V signal to the +24V supply line. Most semiconductors on the board were not amused and released the magic smoke. Damage: About 100 €. Lesson learned: Banana plug jumpers should better be isolated. And no, I did not build the test rig. That was the job of the hardware guys. But I wrapped the f**ing jumpers carrying supply voltages in electrical tape after that.
• Fried my first PC (a 90% compatible XT clone) years ago by connecting the non-standard PSU connector shifted by one pin. +12V from the PSU connected to the +5V line on the board. Magic smoke escaped. Damage: About 500 € in current money. Lesson learned: Double-check the power connectors before switching on. ATX PSU connectors can't be plugged in in a wrong way except by excessive force, so that's clearly a forgotten lesson, as shown above.
• A decade before that, I talked my dad into measuring the output pulse on the primary winding of an old door-bell transformer while feeding in a DC pulse from a 9V battery on the secondary winding. Using the good oscilloscope in the TV workshop, of course. He should have known better. The pulse killed one of the two input channels of the scope. No magic smoke, but sufficient damage to send in the scope for repair. I don't know how he explained the damage to his boss. And I don't know how much damage that little experiment has done to the scope. But I know for sure the pulse exceeded the max. input voltage of the scope by about one order of magnitude. So I guess it killed a good part of the input amplifier, full of manually selected, custom semiconductors. I think that the damage was about 1000 € in current money. Lessons learned: 10:1 probes aren't sufficient for measuring voltages in the kV range. And feeding DC pulses into unknown transformers is a really bad idea.

Alexander