Self-powering a bus-powered USB device?

February 20, 2008 2:58 PM

Anyone ever managed to power a USB flash drive independent of the bus-power?

I am examining it for a friend of a friend. Their usual IT support channels threw up hands and pronounced it dead. I am the last-ditch measure they passed on to because I see this sort of thing as part challenge and part playground.

When connected to known good USB ports, the LED does not light and the device does not appear on the USB bus at all.

It was pried out of its rubber casing when I received it; I have examined the circuit board and see none of the black craters I usually associate with power issues on circuit boards. I was thinking about hooking up a +5V DC power source to the power pin past the USB connector, probably with an alligator clip or small wire, to see if I can eliminate the connector itself as the source of the problem.

I would, of course, block the pin on the USB connector to prevent it from damaging that part of the USB. I figure I'd use a cheap USB male-female extension cord, cut the power wire, then toss or "repurpose" when I'm done with this.

Am I totally off base here? I'd love a better idea that I hadn't thought of, preferably not requiring too many specialized parts.

posted by britain to computers & internet (10 comments total)

I think you're wasting your time. But if you're really determined to do this, use a powered hub in between the computer and the flash drive.

But most likely this is a "MOSFET static-electricity death" scenario -- and those don't leave black craters on the circuit board. There's no visual indication at all that it's happened, except that the circuit no longer works.

A circuit board failure is a vanishingly small possibility, and I think it can be ignored. Bypassing the connector the way you're talking about will serve no good purpose.
posted by Class Goat at 3:03 PM on February 20, 2008

I don't see why you couldn't give it a shot, though if there isn't any obvious damage to the connector or the connection between it and the board, I wouldn't hold out a lot of hope.

Maybe you can just test the continuity of the circuit from the connector to the board before going to the trouble of modifying a cable and getting an outside 5V source.
posted by jjb at 3:05 PM on February 20, 2008

I'll add my thoughts that it will likely be a waste of time, and a perfectly good cable. That said it sounds like something I'd do if no one was around to talk me out of it.
posted by -t at 3:27 PM on February 20, 2008

Flash drives are the new floppies. Let this be a lesson to your friend of a friend. Always use two if you care about your data.
posted by flabdablet at 3:50 PM on February 20, 2008

Simpler than cutting up a cable is to use an ohm meter to check for continuity between the USB connector pins and the power and ground traces on the PCB. You can also plug the device into a hub and check for 5 volts between the power and ground traces on the PCB using a voltmeter.

Note that the lack of the LED does not necessarily mean that the board is not getting 5V power. The LED is not a power indicator. It is driven by the USB controller chip on the board and it may simply mean that the chip is defective and not properly executing code for some reason.
posted by JackFlash at 6:09 PM on February 20, 2008

A couple of years ago I had a client beg me to retrieve the data from his recently deceased USB flash drive. It would not power on at all when plugged in, which lead me to believe that it had an internal power issue.

I purchased an identical USB flash drive online, cut the memory chip out of the old one, and transplanted it into the new one. It was a long, difficult, fiddly job, as the memory chip had something like 48 very small legs. Getting all of those legs to stay in place at once took hours and hours.

Finally, we got the legs to sit right, and painted a couple that still weren't connecting right with conductive paint.

We got everything just right for the span of about three minutes until entropy kicked in and left it ruined for good. But in that three minutes, we got the data off and surprised the hell out of my client (he didn't have much faith that he'd get his precious data back).

To sum up: This is a shitload of work. If you're just doing it for fun, that's fine. If you're doing it on an hourly bench rate, that's fine, too. But go slow, and seriously downplay the possibility of getting the files back to the client. Then if you do get them, it's a party. If you don't (and you probably won't), it's no big deal.

Aside: I had my wife take a picture of me with that flash drive right after I had successfully copied the data from it. I'm wearing my biggest smile on record in that photo.
posted by SlyBevel at 11:55 PM on February 20, 2008

You should search for "USB Power Injector". The Nokia Internet Tablet folks make and use them to trick their little handheld device into acting as a USB host rather than a USB device.
posted by grumpy at 10:21 AM on February 21, 2008

FYI: the right way to deal with 48 very small legs is not to try to do it with hobbyist-grade soldering tools.

I have done this job with hobbyist tools, and it takes hours. I've done it with a professional reflow soldering station, and it takes about 15 seconds.
posted by flabdablet at 12:56 AM on February 22, 2008

I roped an robotics-hobbyist friend-of-a-friend into hooking it up and testing it on a PC and a Mac.

The device still didn't appear on the bus, but on the Mac , OS X noted that the USB device was overpowered and it was just going to disable that port to protect itself kthxbye. So I think we're about at the end of the rope here.

Thanks all for the suggestions. I marked them all because you all made salient points. [Does that ruin the point? Maybe, but I wanted to recognize them all.]
posted by britain at 6:50 AM on March 13, 2008

[oops "a" robotics hobbyist]
posted by britain at 6:50 AM on March 13, 2008