[Gontran]

I know the standard allows you to draw up to 500 mA or so from a USB port, but where does the current limiting happen?

If I design a device to run from USB power, should I limit input current or is it already limited by the host computer?

 

[Garbz]

the USB interface will run out of juice. You don't need to limit current on your device's side.

 

[DaKi][er]

While USB says 500mA there usually isn’t any current limiting at the computer end either, in one of maxim's journals they keep sending me I read they had successfully drawn up to 2A off a usb port without problems, and in a small test I did here off an old pentium2 era motherboard I managed close to 1A but with a falling output voltage that was equal to a 1.1ohm output impedance of the port but I never got around to testing on a more modern computer. MWP off the forums here built a usb dac and headphone amp in one that ran off the usb power and he said he had no problems with a bit over 600mA draw through the port off a laptop

 

[Gontran]

Good info, Thanks!

 

[mono]

- Because there isn't supposed to be over 500mA, you have no guarantee how far past that a board was "overengineered". Trying to draw more you are unlikely to blow a fuse but it's possible, or to overhead and lift a supply trace, possibly eventually onto it's failure.

- If your system is set to use 5VSB for USB (usually a jumper controls 5V vs 5VSB) you might run out of 5VSB current from the power supply if drawing too much.


- A USB device is "supposed" to report it's current usage. It may work without this report but that's the theory anyway.

- Inrush current on a USB device is limited by capacitance, USB spec upper limit is very low, like around 100uF. Normally this is not a problem but with some audiophiles adding thousands of uF worth of caps it becomes an issue. If this situation is present, yes you should limit current on the device, between the caps and the USB interface (supply line).

 

[star882]

It will simply trip a breaker if you go over the limit or short out the bus. How much current it takes to trip the breaker depends on whether it's a PTC or chip-based breaker and the exact design of the circuitry. Chip-based breakers don't give very much margin (expect to draw about 600mA to 800mA before it trips), while PTCs will typically let an amp or more through before going off, although PTCs have greater insertion resistance. There is also a (very inaccurate) ammeter that checks how much current each USB device is drawing.
BTW, I have actually made a USB power booster out of a spare regulated wall wart (5.4v) and a USB extension cable. It supplies up to 2.5A and I made it to supply my USB KVM switch. Before, the breaker in the hub would trip due to overload!

 

[mono]

Some boards don't even have resettable fuses or breakers, just a jumper wire where one could've gone.

 

[chaostic]

USB specifications are set in stone. Only difference between the specs and the implementation is that the implementation can improve on the specs at the makers wish, but has a minimum to achieve.

Now, most of the following is info I got from one of Maxim's engineer journal things they send out, but I pretty much assume they conform to usb specifications. There are two types of usb ports. High power and low power. Low power usb ports have upto 100ma of power at 4.25~5.05 volts with current increasing in 25ma blocks. Highpower ports can go up to 500ma at 4.75~5.25 volts with current increasing in 50ma blocks. All usb ports, to be certified usb compliant, must meet those standards. Now, the thing is, without, the word escapes me, but without syncronization of usb host (computer) and usb peripheral (the device) the port is limited to at most 100ma of power. (Thats because the usb controllers need atleast that much power to communicate). Without something on the device telling the computer what it is and how much power it needs, the computer's usb controller shouldn't give more then the 100ma of current.

So, unless you use chips with usb codes preprogrammed (Like TI's usb audio dac/adc), you should plan for the worst and assume 100ma. I don't know how much of this is true for usb hubs, powered that is (Unpowered would just pass thru to the computer's main usb controller)

 

[ericj]

Quote:

Originally Posted by mono Some boards don't even have resettable fuses or breakers, just a jumper wire where one could've gone.

Aint that the truth. I've seen more than one set of USB ports lose it's magic smoke because somebody plugged a wrong or miswired dongle into the header on the board.