Products > Computers

Damaged SSD by shorting USB - can I ensure the Computer is ok to use still?


On my trusty old Lenovo T430, I once, the side with the Audio jack close to a wall, fumbled a audio TRS jack into the USB somehow where I was suspecting the audio jack must be, and that was the end of that USB3 port (R.I.P., I miss it dearly).
The other one still worked, and with an external hungry HDD connected, with a volt/amp measuring thingy in between, it looks like nice steady 5.1V or so while drawing 0.7A, to it seemed only one USB port was dead and I taped over it and forgot about the incident.

But months later I started noticing I was getting, first just OS errors, when trying to copy large files ( = high likelihood of containing one bad cell), and later it would turn into BlueScreenOfDeath in the Windows installation.

I researched a bit and, I didn't know, a "power surge" can apparently set the seed to make a previously fine SSD deteriorate rapidly.
Also I did not know before trying out Samsung Magician on this Evo something drive, that that laptop internally connects this via USB, no direct SATA/PCIe.
Which would make sense to create a power surge affecting the SSD then if I short something on USB.

Now I wonder:

Can I perform tests that tell me whether it is safe to still use this laptop? (for hardware connected in the future, likely fire hazard, and so forth)
And what would those be?

E.g., if there is nothing suspicious on the USB voltage, is it probably ok? (maybe rather look with scope than just averaged V/A display USB stick) What's a likely failure mode when I short USB (not sure which contacts...) and one port dies, anyway? Protection circuitry blasting something?

As an aside, maybe someone has tips for this sub-topic as well:
Unfortunately, when such a reading error occurs, even if I have it on a SATA of a desktop PC, with CloneZilla booted and trying "-rescue" / by-sector modes, the SATA controller seems to say bye-bye, and every following read attempt will be an error, even if that's clearly not the state on the SSD - which I know because I used a USB-SATA adapter before, copying some files with TotalCommander in Windows, and whenever an error occured, "skip" would not work and also yield only errors after the first - except if I un/re-plugged the USB-SATA adapter, waited 3 secs, and then clicked on "skip" - and presto, the next file that failed before, was copied without a problem. Only that this procedure is very laboriuos, hence why I tried CloneZilla with "-rescue".
I was advised to try DDRescue, but I'm not sure what more this thing can do unless it knows how to get the striking SATA controller out of error state, to be able to continue.

I would buy a small replacement hard drive. SSD doesn't fault the way spinning disks did.  A traditional hard drive would probably have survived it easily :)

I wouldn't trust any test you perform, in short.

An aside, but germane. This is exactly why you should never store any durable data that's not mirrored in some form on a local hard drive.  Especially a laptop, throwing away an OS should be a nothing process.

Isn't it 840 Evo by chance? I don't think SSD deterioration has anything to do with USB power surges. Especially considering that none of digital circuits inside SSD are powered from 5V directly, all through step down converters or LDO.

Yeah yeah, theory and practise, of "don't store data on" ... ;)

--- Quote ---I wouldn't trust any test you perform, in short.
--- End quote ---

That's what I thought, grrr ;)

It is an Evo 870, and I found this claim about power surges in several places.

Anyway: I made a USB boot stick from here:

This, alongside gparted (when starting into xfce desktop) to check what /dev/sd* drives actually arte that you are messing with,
the image also contains the savior program: ddrescue.

I wasted some weekends trying to copy stuff with totalcommander on windows (it's better than windows explorer at least!) and fopund myself un/replugging the hard drive frequently in case of USB-SATA adapter, because after one read error, I thought so anyway, the SATA controller must have striked, as all further reads were errors, until power cycle.
Now on internal mainboard SATA it did the same, and whether I tried this in Windows or Linux, didn't matter.
Clonezilla in "rescue" mode did not help that (although otherwise a nice tool).
So I thought no magic program could read the data without laborious, manual intervention - if it's at the level of the controller not playing along anymore.
But wrong! Ddrescue copied 99,99% of the data in 2..3 hours, apparently bypassing OS filesystem stuff, and all the retries then (that could damage a mech drive so you won't even get the easy data - smart order! ;) ), which took another 10 hours or so.
Ending up with 20KB worth of bad data of 1 TB, that's fair. Spread, of course, so there will be a bunch of files with one to a few wrong bytes or so.
But the bulk of the data is saved.

Also notbale I found that on that laptop, CrystalDiskInfo claims the drive was still 99% good (I guess determined from how much was written),
but on the SATA connector of another PC, it said 60% good, argh! That's quite a bit different.


[0] Message Index

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod