4 ms·
USB ports are supposed to be current limited. Given that those components cost money to buy and money to place… I think we know where low end equipment vendors
by jws 16y ago
USB ports are supposed to be current limited. Given that those components cost money to buy and money to place… I think we know where low end equipment vendors are going to end up.
- marshray 16y agoI'd expect current limiting to be a feature of even the cheapest USB interface ICs. It's really easy to test and potentially a short-circuit/fire hazard. Not the kind of thing any testing lab would be likely to overlook.
- jws 16y agoYou'd think. But the cheapest interface ICs only deal with the data lines. You have to go to extra effort to control the 5V line. It isn't trivial. You want to flow 100mA without dropping your 5V source much, because the USB device is still expecting it to be 5V.
- marshray 16y agoYou'd think. But the cheapest interface ICs only deal with the data lines. You have to go to extra effort to control the 5V line. That's insane. Clearly my problem is that I am insufficiently cynical. I'll just have to try harder.
- nowarninglabel 16y agoWasn't there a problem recently with a USB device, that was popular on Hacker News, getting overheated and becoming a fire hazard. The sleep/time one. I believe they recalled them. Granted, I don't know which side of the cable was at issue.
- chadgeidel 16y agoWakemate - it wasn't the cable, it was the "wall wart" charger. When I got mine (and discarded it) I noticed it was curiously similar to the iPhone 3G charger that was also recalled. Of course, my current iPhone charger looks identical as well.
- AlexandrB 16y agoFunny story. I assumed this as well - but while working on a school project I fried a very nice motherboard when our project got a short from a wire that had come loose. Lesson learned - use a separate power supply and keep the USB only for signalling.
- Natsu 16y agoAlso in this vein: try explaining to your bosses why spending money for an AC adapter for a USB I/O device in a business-critical piece of hardware that does not tolerate lost signals well is a justifiable expense when it is perfectly capable of getting power from a USB cable into a computer that is already powering entirely too many things. I actually bought it myself and smuggled in the upgrade just because the headaches it would cause me, personally, weren't worth it.
- dfox 16y agoThere is probably no USB Host-side interface that really implements USB current-limiting completely. Common solution (used on Intel's reference designs) involves placing solid-state fuse into +5V line (preferably for each port) and sensing if it trips (by routing +5V from actual connector to USB controller IC). Rating of such fuse is typically significantly larger than 500mA (which is maximal allowed current). Truly correct implementation of this would have array of such fuses in different ratings and would switch them according to required current declared by connected device (I don't know of any USB controller or hub IC that actually can do this, not to mention that such thing would be order of magnitude more expensive, which explains why it's not mandatory and why nobody does that). On the other hand there are also pretty common solutions that simply do not do current limiting in any way (or have non-resettable fuse soldered onto board).
- regularfry 16y agoHow very... trusting of you. In a similar vein, I would have expected a motherboard USB circuit to handle an overvoltage from a dodgy peripheral in a contained fashion. I can point out the blown 7A(!) fuse on the IBM T41 motherboard if necessary.