-
Why Does Freezing a Hard Drive Give It Life
Posted by
bostonman
on 21 Sep, 2021 15:22
-
I have a dead drive and experimenting with seeing if I can get some life out of it (I don’t care if I do because I already have the files).
After many hours of letting it run, Windows never saw the drive, or it shows as a generic drive without files. Placed it in a zip lock bag in the freezer overnight, and Windows saw the drive along with about 10min of being able to transfer files before I lost it. Another 24hr freeze, and it worked again (although it seemed toast less time).
Why does this seem to help? Is it electronic related, cooling the motor winding, or causing the mechanics to shrink due to the cold?
-
-
From what I read the "not working" drive does spin but is not recognized?
A bad solder connection is the most likely cause. Temperature differences make different materials expand at different rates and this may cause a bad solder joint to connect temporarily.
If it's a QFP, (or anything not BGA) such bad solder joints can be "fixed" amateurishly by applying some flux and re-solering the chip, and more profesionally / reliably by first removing the IC and then soldering it properly.
The same principle may also apply to for example the silicon die itself and a bondwire. Such a fault is however not so easy to fix.
-
#2 Reply
Posted by
bostonman
on 21 Sep, 2021 18:08
-
Interesting.
So it’s actually electronics. I assumed it was mechanical differences that changed with expansion/contraction. Maybe for fun, I’ll look at the solder connections.
I don’t care about this drive, but I’m taking the opportunity to figure out last resort methods should I have an important drive fail in the future.
-
#3 Reply
Posted by
Kasper
on 21 Sep, 2021 18:26
-
Sometimes you can use cold spray to find the source of intermittant problems.
Air duster spray cans work too if you turn it upside down but I do not recommend that. They put something foul in it to discourage people from huffing it and it is hard to get rid of.
-
-
If you data is worth anything to you, then make backups on some reliable media, so your data is always on at least 2, but preferably 3 physical media (or more if you're paranoid).
Drives are just too cheap these days to even consider using an unreliable drive for important data, and messin' with an unreliable drive with your last backup would be a really silly gamble.
-
#5 Reply
Posted by
bostonman
on 21 Sep, 2021 18:36
-
All my data is backed up onto three drives, so I’m good…. I too am paranoid and tend to go overboard.
-
#6 Reply
Posted by
tooki
on 21 Sep, 2021 19:17
-
Interesting.
So it’s actually electronics. I assumed it was mechanical differences that changed with expansion/contraction. Maybe for fun, I’ll look at the solder connections.
I don’t care about this drive, but I’m taking the opportunity to figure out last resort methods should I have an important drive fail in the future.
No, you were exactly right. It is mechanical, not electronic. It’s got nothing to do with solder joints. The effect in question is (erroneously!) called “stiction”, but it’s actually lubricant failure. Freezing causes tolerances to change just enough to allow the platter to spin up.
-
#7 Reply
Posted by
tooki
on 21 Sep, 2021 19:18
-
From what I read the "not working" drive does spin but is not recognized?
A bad solder connection is the most likely cause. Temperature differences make different materials expand at different rates and this may cause a bad solder joint to connect temporarily.
If it's a QFP, (or anything not BGA) such bad solder joints can be "fixed" amateurishly by applying some flux and re-solering the chip, and more profesionally / reliably by first removing the IC and then soldering it properly.
The same principle may also apply to for example the silicon die itself and a bondwire. Such a fault is however not so easy to fix.
Sorry, totally wrong. It’s lubricant failure in the platter bearings. Nothing to do with electronics failure.
-
-
Sorry, totally wrong. It’s lubricant failure in the platter bearings. Nothing to do with electronics failure.
I had the impression the drive was spinning, but not recognized by the OS.
I do agree that my example of bad solder joints was also just a guess, there can be many other causes.
If there is anything I've learned from the few repair jobs I've done is that it's nearly impossible to analyze "weird" behavior to deduce a cause.
-
#9 Reply
Posted by
bostonman
on 21 Sep, 2021 19:45
-
Damn, maybe I should have included that the drive spins, it just makes odd sounds, and, after hours, never gets seen by Windows, and then I give up trying. Freezing it overnight seems to have little affect on the drive sounds, but, after about 10min, the drive gets recognized. At this point, I get about 10 to 15min of a working drive before it’s lost again by Windows before it goes back to the freezer.
It does seem that after each freeze, I get less time because I just tried it and didn’t get 2min. The first time I got about 20min.
If it’s bad lubricant, does that mean I can use a drop of oil? I know it can be messy, but I’m being serious and/or a bit exaggerated just to understand drive failures.
-
#10 Reply
Posted by
Kasper
on 21 Sep, 2021 20:06
-
Your freezer is probably not always the same temperature. I've seen one oscillate between -2°C and -10°C. So that may cause your changing results but I think it's more likely that it's just wearing out
I think in my few experiences with this, when cold spray affected a solder joint, the effect only lasted a handful of times. Same with board flex, sometimes you can press on a board to make it start/stop working but you seem to need more and more force each time. Again that's something I've not tested very thuroughly.
-
#11 Reply
Posted by
bdunham7
on 21 Sep, 2021 20:13
-
I've recovered the data from a few drives using an ice pack and a fan to keep them cool. It seemed to be an electrical issue with each of them, but not worth any effort other than to clone the drive if possible.
-
-
Louiss Rossmann has made a video in which he takes a faulty HDD to a specialized company for repair. Very informative to watch.
-
#13 Reply
Posted by
Manul
on 22 Sep, 2021 00:46
-
Sometimes heads stick to the platters if they touch. The surfaces are polished to nanometer scale, so they stick extremely strong. I've changed heads a few times on hdd and used slight heating to make them free. Different expansion of materials help to break the stiction. I think that cooling might also work, so this is one possible scenario for the fridge method.
Another scenario is that it might reduce read errors in some cases for the reasons which I don't know, but I've seen it happen. Probably it has something to do with the change of alignment in case of damaged/miscalibrated heads.
But generally it is not a good idea, because condesation or even ice might form and cause even more damage to the heads and platters. So it is just for fun and experimenting, not for cases of important data recovery.
-
#14 Reply
Posted by
BradC
on 22 Sep, 2021 01:16
-
I've done this a few times. Sometimes I've put the drive in a zip-loc with cables taped in and run it in the freezer with a fan pointed at the bag to keep it cold.
In *none* of my cases was it "stiction". None. The drive always spun up but wouldn't read unless it was cold. I had a bit of a "back and forwards" with a storage engineer at Maxtor (before the Seagate buyout) and he commented it was mechanical tolerances. The platter and head bearings are steel and the majority of the drive is aluminium, so the differential in thermal expansion moved the tolerances around enough that it sometimes worked.
Hightlight "sometimes". Stiction can generally be overcome with a quick flicking twist motion on power up to use inertia to help overcome the stick. I personally haven't seen a drive stick since the late 90's. Back then you could pop the lid, give it a spin and read all the data off. Now the lid fasteners help hold the heads in place and it's torque critical, so that's a no-go.
Like all failures, freezing works on some and not others, but the anecdotes are "freezing worked for me so it'll work for you" without any regard to your actual failure mode, or telling you that they had 10 failed drives and freezing worked on one of them.
Like everything else, it's a tool in the box. These days I rely on backups rather than recovery.
-
#15 Reply
Posted by
Zenwizard
on 22 Sep, 2021 03:23
-
In the REALLY old days this was a spindle motor torque issue. The grease in the main baring would get hard and the spindle motor did not have enough starting torque. the freezer would shrink the baring "JUST" enough to allow the spindle to start. Once the grease warmed up the motor could keep the spindle at speed so it worked. This issue generally resolved when the transition was made to fluid baring from the louder ball baring in spindles.
-
#16 Reply
Posted by
bostonman
on 22 Sep, 2021 04:00
-
A few weeks ago I opened the cover of a useless drive and then screwed it back on for what else.... fun. The result was the drive still worked. I was shocked because I thought once the cover is opened, dust gets on the platters and it's useless.
I believe running the drive without a cover didn't work until I replaced the cover.
After dismantling a few drives over the years, I've learned they only have electronics, a motor, platters, and heads (ignoring the coil and magnet). My assumption is the platters can't go bad unless the heads scratch them. The heads don't seem like they can break, although maybe the weight of the head pulls on the arm and/or arm can get weak over time and (as it was pointed out) stick to the platter.
The motor (in my experience) always seems to spin. This leaves the electronics which I imagine could develop cold solder joints (I've seen enough cold solder joints in my days of repairing), but this would mean the bad drives I've experienced all had bad solder joints in places that allowed the drive to spin, but not be seen by Windows.
With all this being said, if the motor doesn't come up to speed quick enough, I assume the electronics aborts the operation. But if the motor spins long enough and heats allowing inertia to get the motor spinning faster and faster, I'd assume a USB reconnect would allow the drive to be seen.
I know a hard drive has very tight tolerances, and millions of '0s' and '1's being read/written at very fast speeds, but seems basic surgery could be done to get a drive back to a working state.
I do have a 2.5" laptop drive that isn't mine, and MAY contain pictures I'd like to transfer (it's a long story, but bottom line is: it's a drive that may or may not have stuff on it). Unfortunately I hear the head swing back and forth and then come to a rest. Although research told me this wouldn't work, I tried buying a replacement PCB, however, as research stated, it didn't work. I believe due to a chip containing information on how it wrote the data on all the platters. This chip would need to be removed and installed on the new PCB.
After various tactics to get the drive working (including the freezer trick), I removed the cover to discover the drive didn't spin at all. After looking around, I learned the metal cover connected two screw holes electrically. Once I touched a wire to two screw holes (with the cover on) the drive spun, however, this only gave me a visual on the heads moving back and forth until they gave up.
The drive is still in my possession and I'd like to get the data off this. I'm keeping the drive in hopes of one day technology advances thus allowing me to get the data. Unfortunately not knowing whether the drive has anything valuable doesn't give me a reason to send it to a pro.
During research for this drive, I also learned someone removed platters from a bad drive and installed them in a good drive. He did this by placing tape around the edges of the platters to keep them aligned. From what I read about this Western Digital drive, as I typed above, the drive needs the electronics to know where data is or how it's aligned throughout the platters.
-
#17 Reply
Posted by
Psi
on 22 Sep, 2021 04:04
-
It's most likely the thermal rate of expansion.
You are changing the physical alignment of things.
Could be balls under a BGA start touching again,
Could be atoms in a logic gates inside a faulty chip moving around enough to start working again for a short time.
Could be the head alignment to the platters moving ever so slighting and allowing a read without CRC error.
Could free up a tight motor bearing.
Could be faulty electrical parts that have gone out of tolerance and shrinking them regains enough margin to work.
Could be a lot of things.
Freezing a HDD is a very mild form of atomic percussive maintenance
-
-
Whatever it is, it isn't universal. I have tried freezing on a number of drives, with zero success rate. In fact I have had little luck with any dead drive. Got one to spin up by the flick method mentioned before, but that only worked for a couple of turn on cycles.
Good brain fun to try to figure out what is going on, but the real answer as stated earlier is to use multiple backups so that the exercise is purely for fun and success isn't important. Tolerances have been too tight for consistent home lab success since shortly after ZIP drives.
-
#19 Reply
Posted by
amyk
on 22 Sep, 2021 06:00
-
A few weeks ago I opened the cover of a useless drive and then screwed it back on for what else.... fun. The result was the drive still worked. I was shocked because I thought once the cover is opened, dust gets on the platters and it's useless.
That will probably give you least a few bad, or weak, sectors. If there's enough dust it will definitely damage the heads.
More interesting info from someone who worked in the HDD industry:
https://www.eevblog.com/forum/beginners/what-to-do-with-a-hdd-motor/msg129452/#msg129452You may also want to search the Internet for the term "thermal asperity" and do some reading. Lots of (dense, specialised) information about particles on the platter and their interactions with the heads in HDDs.
-
#20 Reply
Posted by
Ranayna
on 22 Sep, 2021 09:18
-
In my opinion, Harddisks are one of the most incredible components in a modern computer. That these work as well as they do and not die immediately seems like a miracle.
The tolerances are incredibly tight, some models spin at ridiculous 15,000 RPM, and for some time they were reliably used in portable electronics like laptops or eben iPods.
The posts linked by amyk make this even more clear how amazing the engineering behind the humble and now often loathed harddrive really is.
And with these tighly engineered devices, there is a boatload of stuff that can go wrong.
I personally heard that freezing can revive an old drive, explained by the "stiction" of old lubricant, but i have never done that myself.
I can't imagine that freezing helps with electronic issues, since the drive will really quickly go to room temperature and above when in use. Maybe remove the PCB and clean the springloaded contacts?
-
#21 Reply
Posted by
Psi
on 23 Sep, 2021 05:08
-
Whatever it is, it isn't universal. I have tried freezing on a number of drives, with zero success rate.
Agreed, there's lots of things that freezing a HDD isn't going to do shit for.
- Read/write head damaged
- Any IC that has failed in a destructive way (smoke/explosion on die).
- Shorted TVS diode on input.
- FW corruption
- Run out of reallocating sector pool
etc..
The best thing you can do if you have a failed HDD is buy a good quality SSD to replace it (not super high spec, those tend to run hot).
-
-
My understanding is that this is more urban myth than anything else. In desperation, I tried it once. Nah... Its more likely to do more damage than help recover the data. Getting a replacement HDD control PCB is the best starting point. Which I did also and it worked first time.
www.ebay.co.uk/sch/i.html?_from=R40&_nkw=HDD+pcb&_sacat=0&_sop=15I etc. If that fails then its off to the highly expensive recovery servcies.
-
#23 Reply
Posted by
bostonman
on 29 Sep, 2021 02:46
-
I've froze and used the hard drive almost daily for about a week-and-a-half.
My conclusion (at least with this drive) is that no additional damage was done due to freezing. I have noticed files have vanished, which could be from freezing, however, I just hear lots of odd noises at various times (similar noises I heard prior to experimenting) and imagine I'm losing files at fast speeds.
I even tried doing a restore of old deleted files, however, the drive is so slow that it eventually disconnects itself after extensive use (which has been a common issue), or the recovery software just sits at 3% complete while the drive makes odd sounds. The other day I let it run for over an hour and it stayed at 3%, but then it eventually just disconnects from the USB.
I found watching several Louiss Rossmann hard drive videos to be very educational (it was great advice to those who suggested it). I have a question regarding an approach I saw to recover a hard drive.
In the video the plates were removed using tape to keep them aligned and stuck into a working drive (I saw this done several years ago). After it was done, the drive was working well, but they had software that analyzed many aspects of the drive. It seemed simply swapping plates caused some errors that needed to be fixed with (expensive) software they were using.
From research on these topics before and now, I learned that Western Digital drives keep plate alignment information in memory on the PCB. Having learned this, simply swapping a PCB wouldn't work, which would mean swapping plates wouldn't work.
If I find a sacrifice drive, can I swap plates or is the above correct and it can't be done (and/or is it exclusively with Western Digital)?
Also, can simply installing replacement heads (using a head comb - something I learned in the Louiss Rossmann videos) work or does that too require software to set certain parameters once replacement heads are installed?
-
#24 Reply
Posted by
bostonman
on 14 Oct, 2021 01:54
-
I've continued freezing and running the hard drive.
After many times of trying to communicate with it prior to freezing, it seems (in this case) freezing makes a big change. Unfortunately (although I don't care about the files) the files seem to be vanishing, or show and are not accessible. Some directories were pictures and had the picture count as part of the name. Weeks ago the full amount of pictures were showing, now I have less than half in some directories.
Anyway, my testing to gain knowledge seems complete and my conclusion: obviously the safest way to preserve data is to back it up on a few different drives (or I guess the iCloud). A last resort to obtain data is to try freezing, however, it seems the drive dislikes getting warm or gets overloaded with activity and will disconnect. If freezing helps, quickly get the essential files off the drive.
I asked a question in a previous message, but it never got answered:
From research on these topics before and now, I learned that Western Digital drives keep plate alignment information in memory on the PCB. Having learned this, simply swapping a PCB wouldn't work, which would mean swapping plates wouldn't work. If I find a sacrifice drive, can I swap plates or is the above correct and it can't be done (and/or is it exclusively with Western Digital)?
A second (and believe last) question is: if I use DD Rescue (Linux) to make an image of a failed drive that I just took out of the freezer, does it also copy deleted files that could be un-deleted?