Repairing Dead USB Flash Drives

Over the last few years, [Tobias] has repaired a number of USB Flash drives. This strikes us as a little odd, given small capacity Flash drives are effectively free in the form of conference handouts and swag, but we’re guessing [Tobias] has had a few too many friends lose their thesis to a broken Flash drive.

In all his repairs, [Tobias] found one thing in common The crystal responsible for communicating with the USB controller is always broken. In a way, this makes a lot of sense; everything else on a Flash drive is silicon encased in an epoxy package, where the crystal is a somewhat fragile piece of quartz. Breaking even a small part of this crystal will drastically change the frequency it resonates at making the USB controller throw a fit.

[Tobias]’ solution for all his Flash drive repairs is to desolder and change out the crystal, bringing the drive back to life. Some of the USB Flash drives even have multiple pads for different crystal packages, making it easy to kludge together a solution should you need to repair a Flash drive five minutes ago.

28 thoughts on “Repairing Dead USB Flash Drives

  1. I recall the first time I diagnosed a walkie-talkie with a broken crystal, since then i’v seena few of them, all the way down to a multimeter. I believe that they are probably being overdriven.

  2. I have also had controllers with broken usb inputs (from ESD I guess). Fortunately the flash chip can be easily unsoldered and read out.

    The blocks are not in physical order due to wear levelling, but often if the chip has not been used too much it is still ok for photorec to make sense of.

  3. I’ve also made a temporary repair to a flash drive where the USB connector had a hairline crack in the solder. The solder on those connectors takes all the mechanical strain of repeated plugging-in and unplugging. A bit of careful re-soldering made it work again, at least long enough to recover the files.

  4. All the ‘dead’ USB memory sticks I’ve been given to repair all ‘died’ because the owner accidentally snapped the USB plug off. If the pads were ripped off the board I have to scrape off solder resist to solder onto to the fine data line wires that go into the controlling chip, which actually makes me look more awesome at soldering to people who’ve never picked up such an iron :)

    Thanks for the tip of the dead crystal in case I ever encounter such a non-working USB stick.

  5. Wow – I’ve had a few mice (that fell on the hard floor a few times) do the same thing “constantly dis- and then reappearing in the operating system within seconds.” I was thinking wear related fatigue in a wire or two, but a broken XTAL makes way much more sense.

  6. This is a great post – I just dug out an old thumbdrive which I was given by a friend who was hoping I could do something with it, swapped the 12Mhz crystal with one from a cheapo freebie USB drive, and bingo, it worked! I was totally resigned to having to read the NAND chip raw, so this saved me a serious headache.

  7. btw, nobody with a slightest knowledge of data safety would keep a single copy of anything important on a usb key. They’re intended for transporting data you already have a backup of, not for keeping it.
    as for the repair, that’s a really interesting find that deserves bookmarking. I’m sure the death rate of those xtals would drop considerably if the usb key manufacturers mounted them using proper vibration damping materials. A little sealastic like soft glue on a xtal can do wonders.

    1. I have heard a few stories of people who had their harddrive crash and realized that the only now existing copy of their work is on that old flashdrive they broke the connector off of or bleach washed, or ran over with the car or something.

      I actually did a flash chip swap on one where the connector was so badly broken that repairing it was impossible, we ended up contacting the company who’s advertisement flash drive it was, getting a couple of brand new ones and moving the flash chip over.

      I believe the owner of the flash drive had it connected to a laptop and then dropped the whole thing, breaking the USB connector on the flash drive as well as the laptop harddrive…

  8. This is cool as hell. In the past I’ve fixed a lot with connector problems, but some of the “unfixables” might very well be dead crystals. I have a stash of 12MHz for USB stuff but I never thought about them as a potential failure mode.

    I’ll have to go back and check some of those old clunkers. Thank you!

  9. I’ve fixed a few RC planes, which (after crashes of cause) have smashed their Xtals. My Hubsan X4 has done it a few times, luckily the 16Mhz chrystal was readily available in my drawers, but in a way too big house. solution ? – Use the one in the transmitter on the helicopter, and put in the way to big one in the transmitter :)

  10. So I was being an idiot (as usual) and somehow managed to nearly snap the connector off my USB flash drive: now it’s really quite loose and the connection to my computer doesn’t properly work unless I push it in. Is there any way I could fix this? Or should I just go buy a new drive? I have some important files on there..

  11. I have a USB flash drive with similar symptoms, but with SM3257 memory controller which has integrated crystal. Any ideas on repairing it? I was unable to find datasheets for the controller to see if there is an option for external crystal.

  12. Re. bad crystals.. onboard ones are a nightmare because the code in the e2prom is often vendor and size specific so even an identical drive made a month earlier or later might not work at all or read back with gibberish.
    I sent off one for recovery with a bad controller and it wasn’t possible to get any useful data back even though the chips were intact.
    For these temperature changes (cough dry ice /cough) *might* *possibly* work if the crack is fairly minor and the temperature induced frequency change brings it back within the USB controller’s pass band.
    Working on one of these at the moment, the giveaway is it will show “usb device not recognized” and the LED will blink in an irregular pattern if present.

  13. Later found that the pattern was actually an error code showing a faulty connector (!) or more specifically one of the USB2 D lines was bad. If this happens the drive won’t work because all lines are used in USB3 (3 channels) which is a bit of a weak point in the standard. Swapped connector from dead 256 to 64 and transplated casing, it still works!!! Alas the other one is totally dead and its controller was running hot (95C) even when just powered. Monolith ones sometimes have a hidden diagnostics connector on one end which can be accessed as an LED controller, potentially allowing cloning of the sector tables etc.

Leave a Reply to Mark GomersbachCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.