Hard drives are fragile and reliable all at once. It’s entirely possible to have a hard drive fail, even if your data is still in perfect condition on the magnetic platters inside. [Keith Sherry] was recently trying to recover data for a friend off a damaged hard drive, and demonstrated that modern twists on old tricks can still work.
The drive in question was an old 160GB disk that itself was being used as a backup. Of course, a backup you haven’t tested is no backup at all, and this one failed in the hour it was most needed.
The suspicion was that the controller board was the culprit, and that swapping the board out might bring things back to life. Back in the day, this was a common hacker trick. However, it often fails with modern drives, which store a great deal of drive-specific calibration data on the controller board. Without this specific data, another controller will be unable to access the data on the drive, and could even cause damage.
However, as [Keith] demonstrates, there is a way around this. A controller from a similar drive was sourced, albeit from a SATA version of the drive versus the original which used USB. A single chip is then removed from the original controller, containing the calibration data specific to that drive. Soldering this chip onto the new controller got everything up and running, and the files could be recovered.
If your data is invaluable, it’s likely worth paying a professional. As [Keith] demonstrates though, the old tricks can still come in handy as long as your techniques are up to date. DIYing your own data recovery can be done, it’s just risky is all.
Oh, and don’t forget — once you’ve recovered the files, throw the drive away. Don’t keep using it! Video after the break.
[Thanks to Amy for the tip!]
I DIY’d my own HDD recovery a few years ago. I wrote it up on my blog if anyone is interested:
https://miscdotgeek.com/adventures-in-hard-drives/
TL;DR: Freezer trick worked *just* long enough.
Freezer Trick did aswell for me, workmates didnt know what was hidden under the deep-frozen “premium” company food. got me 10, maybe 15 minutes to get the important data off. the ice buildup is usually like distilled water – not conductive. changing controller boards is a whole different story…
Another trick from an IBM serevice guy if the disc got stuck to give it a upright whack on startup (UPRIGHT!!!) to hopefully let it spin up a last time.
YMMV. All i know if you once read the smart “backup and replace” warning, its usually too late.
I keep business data on at least 3 different medias – tape, raid storage, usb stick.
a few years back i had a fried HDD and did the recovery with a donor board. however when inspecting the datasheet for the memory chip it stated that while the chip would happily handle the higher temperature from the soldering the data might not in fact survive high temps. i opted for an in situ chip reader and dumped the data onto the donor board instead of the soldering. worked like a charm.
I have soldered programmed NAND flash many times and the data was never corrupted. This is on a system with secure boot so even a single bit error will be seen.
I tried to repair multiple harddrives and even swapped the flashstorage to the donor board. but more problems because the main controller hide other specifics, some modern drives even embed some data inside.
check https://forum.hddguru.com/ first for if swapping controller boards is possible. many offer information for free or if asked nicely with enough details, experts might give hints.
These days people will run into a few problems:
1. the flash chip level unique sector mapping is drive specific (the PCB/chips often changes by factory and date)
2. the firmware is now locked on most oem drives, and requires special drive specific software to open
3. the extended SATA standard needed for low level copies is getting less relevant for SSD
Yes, sometimes paying $1700 to get the encrypted partition back from a mobile workstation (with crashed heads) that wasn’t backed up for 6 months is a painful reminder rsync+ssh has been around for decades.
Recently had a hdd failure, data recovery house couldn’t resurrect it. They replaced, heads 7 various other tricks. As it was unrecoverable I opened it up and found deep gouges in several platters. Didn’t think that was a thing in modern drives :(
About as recoverable as the IBM Deathstars.
The bad data recovery companies will change heads and because of doing that badly cause the headcrash you saw. Of course they will say it was like that.
I’ve owned my own data-recovery company. FYI: when I open up a drive and find a headcrash, I’ll report “unrecoverable” without investing time and money in a head swap. So I’m pretty sure they messed up. :-(
Of course if you took a peek yourself before sending it to the data recovery company, to be able to verify the above assumption…. then you ruined it yourself as you (probably) don’t have a cleanroom.
Yup, had this happen before. Seems to be quite common on certain 7200rpm 500GB drives, unfortunately often used in 1TB externals. One of the two failed and utterly destroyed the data, even the thumbnails were corrupted.
No possibility of recovery though supposedly microscopy can sometimes get a percentage of the data on the undamaged platters back. ($$$$!)
Interestingly it was possibly caused when they changed the material in the ramp, and it swelled up over time thus causing problems and essentially contaminating the HDA closest to the ramp.
Where do you think the SMART data is stored. You know the stuff about bad areas not to be used, Power cycles and such.
Let’s say a bad spot was reallocated and the drive contains data you want to recover. Swapping the electronics is a sure way to loose that data. BTW, I did not save which drive the reallocation occurred on. Good data old drive, bad data new drive and visa-versa is possible.
HDDGURU has utilities to do the actual LLF of the drive. I have refreshed the LLF of drives with sector format errors and made new again. I wish I could get to the smart reallocation data and zero it out and re-LLF again. The drive MFG’s so called format is just zeroing the sectors and re-reading the data. Letting the drive re-allocate. But what do you do if the re-allocation table is full ? If I could clear the table then Re-LLF. Test the drive by MFG’s format. If at that time the table fills out again. It’s time to do some clay pigeon shooting using the drive as a target. 00 buck shot does wonders to ensure a non-recoverable drive.
“Oh, and don’t forget — once you’ve recovered the files, throw the drive away. Don’t keep using it!”
If it’s large enough to still be worth something, clearly mark it as a scratch drive and use it for noncritical stuff. Murphy says that it will die again in minutes if you try to use it for important data, but use it for seeding torrents and it will keep working for years.
When it the comment section to post this. Trowing away stuff that could be of use is a total nonsense in our ecologically-challenged world. I was surprised to read that on Hackaday.
No first you strip off the platter(s) for art projects, and save the magnets, and then you throw it out.
They are saying that this works sometimes to reciver data, but you shouldn’t rely on it.
+1
You know, that’s actually a really good point. Threw away some failing HDDs last month, next time I’ll make a point to use them for noncritical storage with a script to notify me on new errors
Some Western Digital drives tend to have their supply protection diodes fail. In my case it caused the traces in the SATA power connector to explode and the connector to melt. Removing the diode and replacing the connector (I used a Molex connector, no 3.3V needed) is all it takes to make it work again.
I just finished my own controller board swap a couple of weeks ago. I fried the board of one of my HDDs when I accidentally used the wrong modular PSU cable (I really wish they would standardize those). I bought an identical used board from eBay that was supposedly in working condition. I used a heat gun and tweezers to remove and swap the bios chips. I reinstalled the HDD and…..nothing. It turns out the new board had 2 bad TVS chips. I couple of quick snips to remove them brought the drive to life. It no longer has built in surge protection but I was able to recover the data.
I saved a extremely pretty and somewhat distressed student back in the day. By using another customers drives board with identical model number. It worked and I saved her uni work just before hand in.
I am not sure I would risk another clients data now but hey when your 19 and she is hot.
Now I am old I just say “Whers your backup?”
“If your data is invaluable, it’s likely worth paying a professional. ” If it’s that valuable, and you didn’t have it backed up multiple places you deserve to be in pain over it’s loss.
I recently got myself a “brand new” SEAGATE ST41200N (1.2G SCSI WREN VII circa early 90s) for my PDP-11 by buying a new-old-stock ST41200ND (differential SCSI) and swapping in the singled-ended controller from a dead drive. Seems like the systems that used differential SCSI (mostly IBM I think) are pretty rare, leaving behind the occasional cache of never used spare parts.
This might be a viable solution when you have a stack of nearly identical drives from a NAS or a backup set.
Or, worse, you might find that they’ve all got the same bad solder that’s slowly destroying the drive.
https://twitter.com/mdwyerfoo/status/1404899366335180802
For a brief time, WD was shipping 2TB Green drives that used a lead-free solder mix that has a sulfur creep corrosion problem. This causes the lines to the heads to short together. I have no idea if there’s a way to correct this other than replacing the whole board.