What is "safe" temp for HD's ?

swayzak · Post by **swayzak** » Mon Jul 16, 2007 3:10 pm

hi

I am concerned about the temp of a HD on my old P4 3GHz Northwood PC.

It runs at 47-56 degrees C.

There are 2 other HDs, one running at 40ish max, the other unknown (older drive with no internal temp sensor).

All are WD Caviars PATAs with 8MB cache.

It's a typically noisy, P4 Northwood system, with the cpu running at 55 degrees + & fans (Zalman flower cpu cooler, case fan + 5 1/2" drivebay bay HD cooler fan).

I've done everything I can but have reached the limit of how much I can cool this beast.

WD have advised any drive that runs at 55+ should be replaced !

Can anyone reassure me that I can get away with running these sorts of temps ?

thanks

swayzak

stromgald · Post by **stromgald** » Mon Jul 16, 2007 3:42 pm

This answer probably won't reassure you, but . . . The only time I've had a hard drive reach 50Â°C is when I had a 300GB Seagate 7200.9 (hot drive) hanging outside the case with almost no flow on a hot Houston day. With decent airflow, I don't think I've ever seen a hard drive get up past 50Â°C.

With that being said, are you sure you've done all the cooling you can? Coolers that fit in 5.25" bays use small fans which can be REALLY noisy.

After looking at some pictures of the Nexus iStyle case, I can honestly say that it's probably one of the worse cases for airflow. Go buy a cheap Antec 3000B for $40 or something, move everything over, trash the 5.25" bay cooler and get a front fan for your hard drives. I doubt your hard drives will ever reach 45Â°C with decent airflow.

BIONIC_EARS · Post by **BIONIC_EARS** » Mon Jul 16, 2007 3:59 pm

Personally, I'd have a very tiny amount of crap in my pants if my hard drives hit the high 40s.

But maybe that temp is normal for your older drive.

Many people here are comfortable with HD temps in the 50s mounted in ham and cheese sandwiches, but considering that data integrity is crucial above all else, I make sure my HD cooling is more than adequate to appease the reliability gods.

A bitch in heat is just asking for trouble.

If your drive has adequate spot cooling or additional mounted heatsinks, then just make sure the data is backed up if it's important, and count the days as your other, cooler drives eventually fail, while that one hot SOB keeps on ticking and makes me look like an ass.

zistu · Post by **zistu** » Mon Jul 16, 2007 4:22 pm

Most of the Western Digitals have a max operating temperature of 55 to 60 degrees. If you look up the specific type of drive you have (Windows Device Manager), you should be able to get a WD number. Enter the first series of characters (WDxxxxcc) on the following url:

http://support.wdc.com/products/index.asp

You should find your drives operating temperatures and all other info.

You're running them close to the max. I am not sure if WD incorporated some kind of safety margin into their values, but running them at temperatures that high isn't really recommendable.

Are you using all the drives constantly? If not, and you use one or more for archive or backup purposes, you could consider putting those in an external enclosure and connect them through USB or Firewire.

Another option could be to replace two smaller drives by one big drive. Hard disks aren't that expensive anymore these days.

By removing one or more drives you may improve your cooling so all drives will be properly cooled again. I've had the same thing once with 4 drives that were running really hot. I replaced 3 of them by a new drive and ended up with 2 drives in total running at acceptable temperatures.

~El~Jefe~ · Post by **~El~Jefe~** » Mon Jul 16, 2007 5:06 pm

hd's shouldnt be over 40 something C..... just shouldnt! also, they might be not registering properly. my test, and its a good freakin test, put your finger on the drive. hold it there. if it doesnt go OUCH MOFO, and you can hold it there comfortably, your sensors are completely wrong. if it hurts your finger... start making funeral arrangements for your data.

Das_Saunamies · Post by **Das_Saunamies** » Mon Jul 16, 2007 11:10 pm

I'm with zistu, check with your manufacturer for drive-specific values. Over 50'C is high, even my old Seagates do better(47'C highest but it has gone to 56'C at worst with no airflow).

dukla2000 · Post by **dukla2000** » Mon Jul 16, 2007 11:26 pm

Can I recommend this old sticky in the Storage Forum - Max safe temp for hard drives?

Max Slowik · Post by **Max Slowik** » Mon Jul 16, 2007 11:53 pm

According to Google, there is no practical correlation between hard drive temperatures and failure. I read that whole damn whitepaper, sleep sand, drool, and all.

http://www.engadget.com/2007/02/18/mass ... ing-thing/

swayzak · Post by **swayzak** » Mon Jul 16, 2007 11:55 pm

Thanks people

stromgald wrote: After looking at some pictures of the Nexus iStyle case, I can honestly say that it's probably one of the worse cases for airflow. Go buy a cheap Antec 3000B for $40 or something, move everything over, trash the 5.25" bay cooler and get a front fan for your hard drives. I doubt your hard drives will ever reach 45Â°C with decent airflow.

I think you're absolutely right. I got this case because, at the time, it was recommended as a quiet case (for using as a DAW).

But the airflow is atrocious, although not helped by the large number of cables inside.

I will look at replacing case (and possibly HDs).

swayzak

jaganath · Post by **jaganath** » Tue Jul 17, 2007 1:11 am

Max Slowik wrote:According to Google, there is no practical correlation between hard drive temperatures and failure. I read that whole damn whitepaper, sleep sand, drool, and all.

http://www.engadget.com/2007/02/18/mass ... ing-thing/

that study was rather controversial, see viewtopic.php?t=38343&highlight= . From my own personal experience I would say there is a rather strong correlation between high operating temps and premature HDD failure.

nutball · Post by **nutball** » Tue Jul 17, 2007 1:18 am

jaganath wrote:that study was rather controversial, see viewtopic.php?t=38343&highlight= . From my own personal experience I would say there is a rather strong correlation between high operating temps and premature HDD failure.

It was controversial for controversial reasons

Max Slowik · Post by **Max Slowik** » Tue Jul 17, 2007 1:47 am

I think I would take Google's findings over my own. Even if I microwaved a bunch of disks, and set them on fire, and maybe even stabbed them, just a little, to see what killed the drives deader, I would trust Google more than me. I make mistakes, whereas Google makes, er, billions.

woodsman · Post by **woodsman** » Tue Jul 17, 2007 11:19 am

I have two "legendary" Seagate Barracuda IV 40GB silent drives (ST340016A) and one 60GB version (ST340021A). The Seagate SeaTools Diagnostic utility informs me that "Worst Temp" is 56C. In my primary box, an old box with only one chassis fan undervolted to 7 volts and not the best air flow around the drive, I never have seen my drive exceed 42C.

My test box is open and after a few hours the drive hovers in the mid thirties for temps.

There are many monitoring utilities to help users avoid hard disk tragedies.

my test, and its a good freakin test, put your finger on the drive.

This a reasonable test in many situations, not just computers. A related tip: always use the back of the hand or finger. This is the standard method taught in fire fighting and damage control classes. If the object you are testing is hot enough to burn, then the damage will be to the back of the hand or finger rather than the front. A serious burn to the front or inside of the hand could effectively render the hand useless for several days or weeks. A burn to the back does not, although still painful.

Granted, any hard drive that is hot enough to burn skin is already a dead drive or soon will be. But the back of the hand or finger tip is useful in many heat tests.

Das_Saunamies · Post by **Das_Saunamies** » Wed Jul 18, 2007 5:07 am

Whatever the results of the temperatures, the fact is that with proper airflow the HDDs have the capability to get cooled to certain temperatures. I think it would be a mistake not to provide this airflow, as it can be done silently and not much is needed.

Google may have done some real-world testing on a respectable scale, but I would still rely on the manufacturer for the guidelines of safe and recommended operation. And besides: Google has millions of HDDs and uses them in farms. I've got just three, and they're crammed into this metal box kicking around in my living room.

ATWindsor · Post by **ATWindsor** » Wed Jul 18, 2007 8:08 am

Das_Saunamies wrote:Whatever the results of the temperatures, the fact is that with proper airflow the HDDs have the capability to get cooled to certain temperatures. I think it would be a mistake not to provide this airflow, as it can be done silently and not much is needed.

Google may have done some real-world testing on a respectable scale, but I would still rely on the manufacturer for the guidelines of safe and recommended operation. And besides: Google has millions of HDDs and uses them in farms. I've got just three, and they're crammed into this metal box kicking around in my living room.

And the manufacturers all(?) say that anything below 55-60 degrees is fine.

AtW

jaganath · Post by **jaganath** » Wed Jul 18, 2007 9:30 am

And the manufacturers all(?) say that anything below 55-60 degrees is fine.

the fact that there is a temperature limit at all should tell you that excessive heat impairs hard drive performance, probably because the heat affects the magnetic media in the hard disk and causes it to lose its magnetization, and thus your data; thus one would expect more risk of data loss and corruption as the temperature increases. I used to run one of my hard disks at temps above 50C (poor HD enclosure) for long periods of time, it died within a matter of months. As Das said, there is very little financial or noise cost to cool HDDs properly, whereas there is a rather large cost if one doesn't, ie risk of catastrophic failure and loss of data.

loimlo · Post by **loimlo** » Wed Jul 18, 2007 9:31 am

Das_Saunamies wrote: Google may have done some real-world testing on a respectable scale, but I would still rely on the manufacturer for the guidelines of safe and recommended operation. And besides: Google has millions of HDDs and uses them in farms. I've got just three, and they're crammed into this metal box kicking around in my living room.

I guess it has something to do with usage pattern. We don't do so intensive readings/writings like google in a short period of time. Our drives tends to idle at most of time.
Just my two cents. No proof to back my thoughts.

Das_Saunamies · Post by **Das_Saunamies** » Wed Jul 18, 2007 9:34 am

At sea level 0-60'C limitation for ambient(drive case should not exceed 69'C), measurements taken at 25'C ambient... standard fare for Seagates for what is possibly decades. All my drives have had the same basic set of rules for temperatures from Seagate.

So yeah, 60'C, and if you believe Google they can be reliably operated as Google did at those temperatures for long periods of time.

My concern is this: I am not Google. My methods of storage, use and cooling differ, as do the drive models I use, thus their research may not apply to my drives. As long as I'm left guessing, and the HDD manufacturer doesn't lay down clear guidelines, I'm gonna keep cooling my HDDs. I'd rather not go towards one end of the limited spectrum if I don't have to, instead letting the reference designs work out what happens. (thanks loimlo for spotting this point before I could get this posted)

And increased heat does increase movement of microscopic particles, making them more prone to losing cohesion, which magnetism is based on, not to mention thermal expansion increasing wear. But is all that happening to a relevant degree... I'm not a scientist, just a user, but with jaganath on this one.

Chang · Post by **Chang** » Wed Jul 18, 2007 2:48 pm

jaganath wrote:the fact that there is a temperature limit at all should tell you that excessive heat impairs hard drive performance, probably because the heat affects the magnetic media in the hard disk and causes it to lose its magnetization, and thus your data; thus one would expect more risk of data loss and corruption as the temperature increases.

I'm fairly certain there's little to no loss of magnetization until you hit around a 33% of the actual curie point of the material. I suspect that for temperatures we're talking about, there'd be no loss in magnetization due to heat.

jaganath · Post by **jaganath** » Wed Jul 18, 2007 4:21 pm

OK, if it's not magnetization then what is it in hard drives that is vulnerable to heat? you're telling me what it's not, can you be more helpful and tell me what it is?

BIONIC_EARS · Post by **BIONIC_EARS** » Wed Jul 18, 2007 4:46 pm

The motor or bearings can stick or seize, the platters can get warped from excessive thermal expansion, nuclear fusion can spontaneously occur if it gets really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really, really hot...

cmthomson · Post by **cmthomson** » Wed Jul 18, 2007 5:55 pm

The main temperature-related failure mode of HDDs is loss of lubrication of the main bearing. Most recent HDDs use a fluid lubricant, typically with some air mixed in. If the drive is too hot for too long, this lubricant either evaporates or migrates, which eventually causes the disk to rotate too slowly to function, or to seize.

ATWindsor · Post by **ATWindsor** » Fri Jul 20, 2007 9:46 am

jaganath wrote:
And the manufacturers all(?) say that anything below 55-60 degrees is fine.
the fact that there is a temperature limit at all should tell you that excessive heat impairs hard drive performance, probably because the heat affects the magnetic media in the hard disk and causes it to lose its magnetization, and thus your data; thus one would expect more risk of data loss and corruption as the temperature increases. I used to run one of my hard disks at temps above 50C (poor HD enclosure) for long periods of time, it died within a matter of months. As Das said, there is very little financial or noise cost to cool HDDs properly, whereas there is a rather large cost if one doesn't, ie risk of catastrophic failure and loss of data.

I have never protested that excessive heat is a problem, but temperatures in the 50-range is probably not excessive, seeing at is within spec, and the results from google.

I have runned a HD in an enclosure with temps in the 50-range for years with no problem, one sample is way to little to use for telling if its safe or not.

AtW

subsonik · Post by **subsonik** » Fri Jul 20, 2007 1:03 pm

My opinion is that rather than elevated temperature itself, changes in temperature lead to earlier drive failure.

This might explain the controversy between Google's paper and the experience of home users. Google runs their HDD's 24/7, so there is absolutely no change in temperature. Their servers are located in nicely aircooled datacenters, with a constant ambient temperature.

A typical home user powers on and off his PC every day, so the hotter the HDD runs, the bigger the temperature changes are every day. If your HDD runs hot, leave it on 24/7

I personally have a few drives which have been running for over 30K hours continuously. That's almost 4 years of 24/7 operation! Any sane user would have thrashed the HDD's long time ago.

Das_Saunamies · Post by **Das_Saunamies** » Tue Jul 24, 2007 3:25 am

Well said. The Google environment is certainly more static than your average home with varying levels of temperature, humidity and even vibrations -- not to mention differences in drive use.

Aris · Post by **Aris** » Tue Jul 24, 2007 5:44 am

It all depends on what the manufacturers rated max operating temp is. Its usually 55c. Some drives are 60c. Long as your below your specific drive manufacturers rated max operating temp then your fine, but lower is always better. Only increase noise to lower temps if your over the max temp though.

Traith · Post by **Traith** » Sat Jul 28, 2007 2:51 am

The main thing is that all the heat generated by the drive is removed. i.e. the rate of heat removal = rate of heat generated.

I used to work at one of the major OEMs. We would put certain drives in certain cases assuming an ambient temperature of 30-40C for consumers, higher for business but lower for data centers. The idea being that the only variable would be the ambient temperature [you'd assume that there would be one person who would use his PC at 40C and that he would be transferring the mother of all files from one drive to another].

NOT. The main thing for us is noise. Few consumers notice the hot air coming out of their boxes but they DO notice the fan or drive noise. We have put FOUR IBM GXP 75s in a St. Francis casing. Note, there is no fan blowing over these and the drives are 'soft-coupled' to a casing that's mostly plastic anyway. I would see drive temps at 50C... idle ^ ^ and the STF casing typically only has the PSU as the sole exhaust/intake fan. The fact that it works points out the excellence of this tool-less casing [factory paid USD 19.89 for EACH casing] [One side effect of the plastics used, the hottest part of the casing to the touch was the rear top [i.e. the PSU] which you'd expect but even 3 minutes after shutdown, touching an IBM drive would hurt [unless you had opened the casing]

By far, the largest number of warranty claims I've approved was for hard drives.... followed by power supplies [we used Foxconn but... CHEAP Foxconns].

The folks that designed servers for the datacenters also would target a specific temperature for certain drives/cases combo but they didn't have to worry about noise or cost. Among the stuff they told me, they aimed for a steady drive temperature. I guess this is similiar to what Google [or anyone] tries to accomplish in their datacenter.

----

My personal take on this, make sure your drives are at a relatively constant and low temperature. And to watch the temperature rise when load increases and make sure the cooling is adequate to minimise the rise, that way you know your cooling capacity exceeds the heat generated. Why worry about this? Put it this way, do YOU have your data backed up?

My own setup has 4 - 6 WD 320GBs in the forward bay of a Lian-Li V2000 with 4 Samsung 500GBs behind them. I put the 320s closest to the intake fan [I run the fan at normal speeds because when I dropped the RPM... I started to hear the seek sounds from the WDs] since they run hotter.

WDs = 33C [max of 40CC during 'full load'] This varies from drive to drive depending on where it is, [assuming Drive 1 is at the top] drive 1 is at 35C, drive 6 at 33C and drives 3 and 4 are at 32C. [all from SMART so take this with a grain of salt] I like the V2000's setup, just using the stock mountings already give low noise and the separation between drives means easy airflow even over the PCBs.

*Full Load being transferring 300GB of files from one WD to the neighbour immediately above or below it.

------

Off topic or maybe not, someone mentioned IBM GXPs, man I hate those drives. In my regional area, the reliability of those drives were so bad we estimated that half of all IBM GXPs would fail within 9 months. [Eventually we figured out that all the drives from the Phillipine factory should be changed while the Hungarian ones were "OK"] The reason being that the algorithm that calculated the effects of heat expansion on the platter were slightly off. I had dozens of cases where a user would switch off his PC, then find that it would not boot a day later [until I told them to leave the system at the BIOS screen for 4 hours.... warming the drive back up, LOL]

Another one for me would be those Quantum drives in the early 2000s, i.e. from the time Quantum pushed out those 4200 rpm [LCT] drives till shortly after the Atlantis [AS] drives came out. If it had a drive controller IC from Philips, the drive would fail within 12 months if not sooner ... UNLESS the drive was somehow kept at a low temperature [the design techs guesstimated it to be 20-25C depending on the casing]. The problem here was that the IC would burn out it it got hot enough. So PCs in libraries, homes and what not lost their drives in job-lots while those use as console workstations in datacenters were fine... Heh heh.

This Quantum disaster was very widespread, I think it even got public in a minor way. Now if Google had bought all the drives from Quantum, you'd see a very definite correlation between high drive temperature versus low reliability.

subsonik · Post by **subsonik** » Sun Jul 29, 2007 10:28 am

subsonik wrote:My opinion is that rather than elevated temperature itself, changes in temperature lead to earlier drive failure.

To quote myself... I just had a Samsung HD501LJ fail. It was in a USB enclosure, and got fairly hot in a short time because I was writing a backup to it. You can't make a statistic on just one drive, but it supports my theory.

Traith · Post by **Traith** » Mon Jul 30, 2007 12:36 am

Which enclosure do you use? I'm using an overpriced Cooler Master 'X-Craft eSata' enclosure and my HD501LJ definitely gets hot with it. The only ventilation with this enclosure is the mesh grill on the bottom of the enclosure so I elevate it a little to drop the temperature. I was getting high 40s C, dropped to low 40s C, still a lot higher than when mounted in the PC itself.

Are you using eSata or USB? I noticed that with USB, the drive ran longer for a typical job and seemed to get hotter too, naturally.

cmthomson · Post by **cmthomson** » Mon Jul 30, 2007 6:28 pm

Of course different drive families have different typical failure rates and modes.

There was a period about 3 years ago when nearly all laptop drives would die at any temperature if they were spun 24/7. This was due to lubricant failure.

But in general, if you let a disk get over 50C, and especially if you let it go through large temperature swings, you're looking for trouble.

As many have already pointed out, the Google data is for disks in a constant temperature & operation environment. But equally important IMO is that these are server disks in server platforms and environments. Home PC users have a different class of drives, and wildly different usage/environment.