Bitcoin Forum

My rigs are outside here.  Right now it's -24C outside and my rig's hard drive is running at 16C (my CPU amazingly is at 12C).

My question whether or not the hard drive will stop working below a certain temperature?  It's get into -40 to -50C here and it's possible that the HDD temperature will go below 0C.

Also at what temperature do brushless fans (on the video cards/CPU) generally stop working?

I just looked up a Deskstar datasheet, and it specifies 5C to 60C operating, -40 to 70C non-operating. WD Caviar is 0-60C operating. Looks like it would be best to not freeze your hard drive.

It sounds like if the fans stop working, that would actually be what you want to keep the system warm...

Holy crap where do you live?!

We had a trick for recovering data from old hard drives with wheel bearings (which aren't in use anymore, most spinners now use some sort of fluid bearing) where if the bearing had frozen, sometimes we could put them in the freezer overnight. Get them down -10 or 0C, and plug them in really quick and see if we could read the drive. 16C for a hard drive is fine, but i'm not sure about continuously running the drive below freezing.

I'm in North Alberta

Okay, thanks.  I guess I'll have to wrap some kind of insulating material around it to keep it warm.  Hopefully the motherboard won't cold bug at those temps either (right now the northbridge is 30C and the memory is 12C).

That's ri-dicu-cold. Why not take your rigs inside to help heat your home?

Interestingly, it's actually a pretty good idea to insulate your drives to keep them a bit warmer:

http://static.googleusercontent.com/external_content/untrusted_dlcp/research.google.com/en/us/archive/disk_failures.pdf

Drives between 15-30C fail at a higher rate than 30-40C almost across the board.

Interesting article. It might seem that way at first, but look at the temp vs time graph. Seems that if a lower temp drive is going to fail, it's going to be in the first 3 months. Between 6 months and 2 years, the failure rates are about the same, and beyond 2 years, it's actually a lower failure rate.

The issue will be getting your data back off...........

Consider that:
1. These devices are mechanical.... simple physics states that materials  expand and contract with temperature.
It means that as the temperature falls/rises. the tolerance of the mechanics may go out, you may record data on disk surfaces that have contracted/ with disk arms that are smaller, bearings that are tighter. When the mechanics return to the "correct" size it IS a problem.

2. Materials generally lag tracking the ambient temperature, which means that the drive will be colder than its surroundings when it get warmer, so a shit load of water may condense out on the electronics/inside the drives.

Mining rigs.. why have a HD at all? Boot off a USB and you won't have anything to worry about..

Where abouts in AB? Edmonton over here.. but I keep my rigs inside still, helps keep the gas bill low ;)

Well, across the board, if a drive is going to fail, it is most likely to do so within the first 3 months, it is simply the most likely to do so in colder climates. [3months, 2years) all drive failure rates drop significantly, while cold climates remain the most likely to fail.
To quote:

Which suggests that at the extremes (Cold and Hot) you do in fact correlate temperature to increasing failure rates, perhaps for mechanical reasons such as hardcore-fs lists. Therefore it does seem wise to attempt to keep drives operating at more reasonable temperatures, and insulating at extreme cold temps like -24C is very highly suggestible.

Still works. Just did it on a dropped laptop harddrive a week ago. Vacuum Baggy,some glue to seal air out, long power/sata cables running outta the freezer and the data was all brought back :P

See my sig. You were very lucky then. ;)

kk, I think I'll put it in a static protected bag and then wrap it in a bunch of bubble wrap, which is pretty insulating.  Then I'll see how that works and if it can bring the temps back into the 20s.

why not run your rig from a usb drive (or SSD if you need the capacity although I can't see why)?  No moving parts means it really shouldn't care hold cold it gets (within reason).

There was a gy that tested a Harddrive on LN2 .... Cant find the link ... :(

But i found something else: http://www.pcmag.com/article2/0,2817,2393390,00.asp

That's an SSD. Of course it's going to have no issues below freezing. The main issue with a spinning hard drive is the bearings.

Oh SSD should work...

found something different:

http://www.google.de/imgres?hl=de&safe=off&client=safari&sa=X&tbo=d&rls=en&biw=1426&bih=840&tbm=isch&tbnid=0XdsYthZHx17UM:&imgrefurl=http://hacknmod.com/hack/how-to-recover-dead-hard-drive-data/&docid=iMSE_MaCuwpwsM&imgurl=http://www.hacknmod.com/wp-content/uploads/2008/11/hard-drive.jpg&w=320&h=240&ei=gk61UKzCOc3bsgbG8ICICw&zoom=1&iact=rc&dur=383&sig=106743804277992879291&page=1&tbnh=137&tbnw=190&start=0&ndsp=36&ved=1t:429,r:6,s:0,i:100&tx=101&ty=82

http://www.google.de/imgres?hl=de&safe=off&client=safari&tbo=d&rls=en&biw=1426&bih=840&tbm=isch&tbnid=g_DdxDrspWrAsM:&imgrefurl=http://www.jimwilliamson.net/misc/2003-07-30-cold-hard-drive/index.htm&docid=QnHhlqw3gW6PvM&imgurl=http://www.jimwilliamson.net/misc/2003-07-30-cold-hard-drive/5.jpg&w=405&h=202&ei=xU61UMmDIY7Mswby64GgDA&zoom=1&iact=rc&dur=165&sig=106743804277992879291&page=1&tbnh=141&tbnw=282&start=0&ndsp=35&ved=1t:429,r:2,s:0,i:88&tx=60&ty=65


And i saw someone on xtremsystems (cooling with LN2 or dry ice) supercool their hard drive... What didn't went to good :)

HAHA - didn't even notice. Must of been an older hard drive in that laptop.

Static bag = conductive. Not a good idea.

Just bubble wrap then?

Edit: cardboard is r4, so I guess I'll make a little cardboard tomb for it.

welp, I fixed this without insulation by adjusting the position of my cards so that they blew on the hard drive.  Now it's -22C out and the HDD is at 27C and the cards are chugging away in the 50's.

I'm pretty sure some of the drives I've split open have had dessicant inside. Dunno if it'd be enough with that much delta T though. Depends on the humidity too I suppose.

Modern hard drives use fluid dynamic bearings.  As long as you keep it spinning, it should be fine.  While the drive is spinning, the bearing should be producing enough heat to keep from freezing even if the environmental temperature is well below the manufacturer's spec.

That said, you should use old drives that are out of warranty for this.  Although I think you will be fine, the manufacturer would not be amused by out-of-spec operation.

Oh, and drives don't contain desiccant because they aren't sealed.  What they normally have is a HEPA filter to trap particles pulled in through the pressure-relief port.  It looks like a little tea bag.

Back in the late 90s, I had what I now believe to have been the first in-car MP3 player in Minnesota.  It was really just a computer fed into my stereo by hacking the CD changer bus, but still...  On a few of the coldest days of the year, it wouldn't boot because the hard drive wouldn't spin up, but it was just a matter of waiting a few minutes for the CPU and PSU heat to thaw out the bearings.  That drive survived through several years despite below zero winters and over 100 degree summers.

Sorry.... Totally irrelevant about the bearings.... you have obviously not understood my original posting.
The amount of structural change in the materials over a given temperature range, dictates the reliability of the read/write circuits.

OLD hard drives, strictly recorded ones/zeros onto (maybe in a manchester or other recording system) the media, so they were more reliable over extended temperature ranges, due to the massive size of the encoded magnetic domains.

NEW hard drives DO NOT do this, and the data recovery is based on the probability of the signal recovered from under the noise floor being a one or zero NOT on a hard fact (very frightening but true).
By changing the temperature you change the probability, due to material expansion/contraction, which means data recorded at a temperature outside the drives operating range may ONLY be reliably recovered at that temperature.

Heh.  Mechanical contraction and expansion are not issues here.  Drives haven't used absolute positioning for like 25 years.  The heads center themselves over the servo tracks using analog voice coils and feedback, not stepper motors.  As for the magnetic systems, the entire read circuit is built around looking for the difference between parallel and anti-parallel domains, not at absolute levels.  As long as you stay away from the Curie point, the changes in magnetic properties are tiny compared to the other forces that a hard drive has to deal with, and well within the capabilities of our modern DSP techniques and redundant encoding schemes.

If the spindle is physically capable of spinning, the drive should be fine.

is now the living in happyness