I’m trying to recover files from a hard drive with bad sectors, but it keeps disconnecting in the middle of the scan. The drive is still detected sometimes, and I’m worried continued attempts could make the data loss worse. I need help figuring out the safest way to recover data from a failing hard drive without causing more damage.
Seeing a cyclic redundancy check error, or having File Explorer hang when you open one folder, usually means one thing to me. Stop touching the drive.
If it was an external drive, I unplugged it. If it was my main system drive, I shut the PC off and left it alone. Reboots feel harmless, but they aren’t. Each spin-up gives Windows and the drive firmware more chances to hammer the same damaged spots again. I learned this the hard way once with an old 2 TB WD drive. It went from “some folders won’t open” to “drive vanishes mid-copy” in one afternoon.
What those errors often point to
You’re usually looking at one of two problems.
Soft bad sectors, sometimes called logical bad sectors. This is where the data or its error-check info no longer lines up. I’ve seen this after power cuts, forced shutdowns, and one time after somebody yanked a USB cable during a transfer. The storage area itself might still be okay.
Hard bad sectors are uglier. Those tend to mean physical damage. Worn platter surface, weak head, failing internals, stuff like that. No software fixes hardware. At best, software helps avoid damaged areas long enough to pull data off.
Where people make it worse
A lot of people go straight to CHKDSK. I wouldn’t do that first if your files matter.
CHKDSK is for making the filesystem usable again. It is not built around preserving every damaged file record for recovery work. I’ve seen cases where it cleaned up the mess in a way Windows liked, but recovery got worse after. Filesystem repair and data recovery are not the same job.
What I’d do first instead
Clone or image the drive. Full byte-for-byte copy, if the drive is still readable enough.
The point is simple. You want one pass over the failing disk, not ten. Pull what you still can onto a healthy drive, then do the rest of the work from the copy. Once the image exists, leave the original powered off.
For DIY stuff, I’d pick recovery software over repair tools. Disk Drill is one of the better-known options for this kind of situation. What stood out to me is the backup side of it. On unstable drives, software like this tries not to get stuck forever on unreadable sections. It skips rough spots, copies easy data first, then circles back with smaller reads. That matters. Fewer long stalls usually means less stress on a dying disk.
The rough workflow
- Connect a healthy target drive with enough free space.
- Make a byte-for-byte image of the problem drive.
- Scan the image, not the original disk.
- Save recovered files to a different drive.
That last part gets missed a lot. Don’t restore recovered files back onto the original drive. Don’t dump them onto the same drive holding your image either. Use a separate destination.
When I’d stop the DIY attempt
If the drive clicks, grinds, beeps, fails to spin, or stops being detected at all, I’d tap out there.
Those symptoms usually point to mechanical failure, not a simple filesystem issue. Software won’t fix a dead head stack or damaged internals. Pushing harder can make the result worse. If the files matter, send it to a lab.
Two names people bring up often are Gillware and Techchef. They handle damaged drives with proper hardware and cleanroom work. It isn’t cheap. You’re often looking at hundreds of dollars, sometimes a lot more. Still cheaper than losing wedding photos, tax records, client footage, or old project archives you can’t rebuild.
After recovery
Only after your files are recovered and checked would I even think about trying repairs on the original drive.
At that stage, you could do a full format or run chkdsk /r so the OS marks bad sectors and avoids them later. I still wouldn’t trust the drive for anything important again. Once a disk starts throwing CRC errors and freezing on reads, I treat it like a warning label, not a temporary glitch.
What I changed after getting burned
I stopped pretending one copy was enough.
The backup rule I stick to now is 3-2-1.
Three copies of your data.
Two different storage types.
One copy off-site or in the cloud.
It takes a bit more setup. Still beats spending a weekend trying to pull family photos off a drive making sick noises at 1:30 a.m.
2 Likes
Drive disconnecting mid-scan is a bad sign. I agree with @mikeappsreviewer on one part, stop doing full rescans on the original disk. Each freeze or reconnect event means the drive is struggling to read, power, or stay linked long enough to finish.
One thing I’d check first, before more recovery attempts, is the connection path. A lot of failing externals look “dead” when the real issue is weak USB power, a bad cable, or a flaky SATA-to-USB bridge board. If this is an external drive, try a new cable, a different USB port, and skip hubs. If it has its own power supply, swap that too. I’ve seen drives disconnect for hours, then stay up fine after replacing a $10 cable. Sounds dumb, but it happens alll the time.
I differ a bit from the blanket “never reboot” take. One controlled reboot after changing the cable or port is fine. Ten retries is where people cook the drive.
Next, check SMART stats with something simple like CrystalDiskInfo, only if the drive stays mounted long enough. Reallocated sectors, pending sectors, and CRC errors tell you a lot. CRC errors often point to cable or enclosure trouble, not only platter damage. If SMART is full of pending and uncorrectable sectors, stop testing and move to imaging.
For recovery, Disk Drill is still a solid pick because it lets you image first and work off the image. That matters more than the scan engine at this stage. Also worth reading actual user feedback here if you want a quick gut check on the software, real Disk Drill user reviews and ratings.
One more practical tip. If the drive is in an external enclosure, remove it and connect it straight to SATA inside a desktop, if you have access to one. USB bridge chips often drop out under read stress. Direct SATA is more stable. If it still disconnects on SATA, the drive itself is the problem, not the enclosure.
If the disk starts clicking, spins down, or vanishes from BIOS, stop DIY stuff. At taht point you’re feeding a failing drive more chances to die.
If it’s disconnecting mid-scan, I’d stop thinking of this as a normal “scan longer” problem and start treating it like a stability problem first. @mikeappsreviewer is right about not hammering the disk, and @cazadordeestrellas is right that the enclosure/cable can absolutely be the idiot in the room.
One thing I’d add: cool the drive down between attempts. Not freezer nonsense, just let it rest powered off for a while. I’ve had borderline drives stay online longer when they weren’t cooking themselves during repeated retries. Heat makes marginal reads worse, period.
Also, avoid file-by-file browsing. Explorer is weirdly great at freezing itself on one damaged folder and then dragging the whole drive down with it. If you can, use something that reads sequentially and keeps logs. Disk Drill is fine here mainly because you can create a disk image and then stop babysitting the original hardware.
And if you see what people call a hard drive bad sector and read error basics situation getting worse by the hour, that usually means the media is degrading, not just Windows being dumb.
Small disagree with the “check everything” approach: I would not spend a ton of time running extra diagnostics if the data matters. Every test is more reads. Swap cable, try direct SATA if possible, then image attempt. If the drive drops even there, lab time. Kinda sucks, but taht’s the line for me.
