Re:Google engineer finds USB Type-C cable that’s so bad it f...
by cesarb on Wednesday February 03, 2016 @02:19PM
(#11028900)
This was not shorting +Vpp to GND; this was putting -Vpp in the +Vpp pin. All pins are referenced to GND, which is 0V by definition, so if you put +5V in it, it's the same as subtracting 5V from the voltage in all the other pins. And since they put the 0V in +Vpp, it became -5V. Most devices aren't prepared for negative voltages.
Re:Google engineer finds USB Type-C cable that’s so bad it f...
by ThePhysicist on Thursday February 04, 2016 @04:42AM
(#11032902)
Point taken, but I still think that the device should be protected against supplying it with a "malicious" input voltage, as there are cheap and robust components to do this. The NCP373 for example (http://www.onsemi.com/pub_link/Collateral/NCP373-D.PDF) protects against voltages between -30 to 30 V and is designed exactly with the "faulty USB cable" use case in mind, and the low-volume cost is just 50 cents (large-volume cost is probably much lower). So I think there really is no excuse for letting a faulty 5 $ USB cable destroy your 1000 - 2000 $ device, even if it supplies an input voltage of the wrong polarity.
Btw I already destroyed some parallel, serial and USB ports when connecting them to devices I built that malfunctioned or were ill designed (during prototyping), so as a measure of safety I now often use optocouplers to galvanically isolate my circuit from the port, which can help to prevent most kinds of damage.