The diode is conducting in both directions, so it is broken. This can likely also be inferred from the smoke. The datasheet for this diode is at https://www.st.com/content/ccc/resource/technical/document/datasheet/19/78/9... The datasheet has a picture on the second page, of typical voltage/current curves for bidirectional and unidirectional devices. Pretty sure a diode is a unidirectional device. The curves include both negative and positive voltage and current. I'm guessing that chart is displaying positive forward voltages to the left of the Y axis, and negative or reverse voltages to the right. Not totally sure. It looks like there is a minimum voltage called V_F needed for the diode to conduct in the forward direction, and a maximum voltage called V_RM (maybe safety), and V_BR (breakdown), above which it will conduct in the reverse direction, and it roughly doesn't conduct in-between these voltages. With my diode broken, it seems this would imply that it was provided excess voltage in either direction, producing enough current to melt its innards. I'm using the factory-provided charger. It's probably been plugged in for weeks. There were various issues before this failure. Notably the power LED would blink red and the system would power off, while it was running. The charger right now is behaving in a funny way, where its green light is pulsing on and off. Measuring the voltage shows at the start of a pulse, it gives the correct 5V, and then this quickly drops to zero for the rest of the pulse. When connected to the shorting diode, it provides about 2.8V across it, continuously, while making a strange staticy noise at the wall. On the third page of that datasheet, the diode electrical characteristics are given. This is a SMBJ6.0A, so it looks like V_RM=6.0 V, V_BR=6.70-7.05 . I'm guessing that V_F is so small it doesn't matter. There's another item in that graph on page 2. I_PP, the peak pulse current, and its voltage V_CL, which they call the clamping voltage. Not sure what that means precisely, but on page 3 it has footnotes regarding their numbers:
3. To calculate V_CLmax versus I_PPappli: V_CLmax = V_CL - R_D x (I_PP - IPP_appli) where IPP_appli is the surge current in the application 4. Surge capability given for both directions for unidirectional and bidirectional devices
There are two values given for I_PP, V_CL, and R_D . One labeled 10/1000 us, and one labeled 8/20 us. I'm guessing that's the duration of the surge. The SMBJ6.0A has these numbers: 10/1000 us: V_CL = 10.3 V I_PP = 61 A R_D = 0.048 Ohm 8/20 us: V_CL = 14.8 V I_PP = 270 A R_D = 0.027 Ohm It sounds like this device will break if more than 10 V is applied for more than 1 ms, in either direction. I'm not sure of this, though. With a little bit more reading, maybe I could calculate how much current and heat is generated if the laptop is reverse-powered, and whether that would break it or not. -- Another thing going on is that I had the laptop plugged into the old RS-232 serial port of the mainboard I'm poking at, via a usb-serial adapter. Additionally, the laptop charger is ungrounded. An RS-232 serial port can have voltages between +-14V of ground. And I'm wondering if the ungroundedness of the laptop could have produced a different ground between them, and maybe a strong ground current when the cable was replugged. I don't really know. But after reading about this, and seeing it needed >10V difference to likely burn, I'm going to be thinking there's a reasonable chance it's either the grounding or some voltage from the RS-232 port escaping through the usb adapter. I'm also thinking that's most likely to happen when I hard repower the mainboard, which I am doing a lot. When I repower the mainboard, I often see garbage on my serial terminal, and sometimes the terminal is strangely forced to close. This is supporting information that there could be voltage spikes coming in through USB. ---- Whew ! :)