HDDs Cook too

[RKS]

A lot of cases these days have those quick install facilities for Hard Drives. They tend to be of a soft plastic material.

It didn't occur to me until today that the reason why my hard drive array was getting so hot, despite two cooling fans specifically for the HDDs, was due to those soft plastic install devices.

In the old days, HDDs were screwed to the case. This acted as a heat sink. The hard drive doesn't need that much of a heat sink, but if thermally isolated then they can cook despite good cooling.

So I will be ditching the plastic conveniences and going with the old fashion case screw and see if the HDDs run cooler.

Having just lost another SATA out of an array, I am thinking this is a timely discovery, or is it?

Anyone have an opinion on inadvertent thermal isolation of HDDs, or is this a well worn issue that I am discovering somewhat belatedly?

Robert

 

[AdriMagnon]

Hi,

I'm not sure I fully agree with the premise that the soft plastic quick install are to blame. Not only do some very large manufacturers (both of computers such as HP and Dell but case makers such as Antec and Thermaltake) use them and I doubt they would do that if there was an issue.

I have no idea how drive manufacturers test their drives however. I will say I think that if you have two drive coolers and they are getting hot something is up. Maybe your case temps are high overall?
I'm thinking that the moving parts of the drive cause friction and therefore heat. The metal casing heats up and so does the air around the case. Maybe you need more surface area for the fans with hard drive heatsinks? I don't know if your fans blow across the drives or onto them.

I work with Dells and have been for several years. The plastic brackets they use have not, in my mind and attention, created more dead drives than normal.

I could be completely off but I think I am kinda on track. I'm thinking I just rambled here.

Adrian

 

[captaincranky]

Drive heat must be an issue or the new velociraptor wouldn't be encased in the "ice pack" mounting sled. As to whether or not heat dissipation is or isn't a factor in the manner you suggest, it would be interesting if anyone has actually tested the temperature difference between soft and hard mounting of a HDD. A soft mounting would absorb more of the drives own vibrations, perhaps making it less libel for it's own destruction.

Modern high capacity HDDs have multi platter designs and run a bunch hotter than the single platter little guys.

I think case designs like the Antec 900 have done a good deal to address the issue of HDD heat, via direct airflow over the drive cages with large front mounted fans.

Antec also hard mounts the drives (metal to metal) on this particular case. Is this an attempt at heat reduction, or a cost cutting methodology, I don't know. I do know the drive cages are being blasted directly by 2 120mm fans, which I'm sure should be given the lion's share of the credit for any thermal advantage.

 

[RKS]

I have a Thermaltake case with six installed HDDS, two 15,000 RPM SCSIs and four SATA drives that were in a striped array.
The HDD fans are two 8cm low speed types.

Soft mount will cause thermal isolation which means that the heat will accumulate around the drives. The SCSIs actually run cooler than the SATA drives.

The drive order from bottom to top is SCSI, SATA, SATA, SCSI, SATA, SATA.

I have since reasoned that the drives have been getting hotter over time - the computer is now in its fifth year. I have a separate computer used for games, so this computer mainly does documents, download, web, work related stuff hence an emphasis on storage. The computer often remains on and working for weeks on end.

I think the HDDs get hotter over time because the lubrication in the bearings dry up and this may be one of the differences between the 24/7 and regular versions of the Seagate drives. I have a dead one on the test bench ready to meet the screwdriver, so I will know more on this soon.

The fact that manufacturers go to the trouble of supplying HDD fans and the fact that soft rubber or plastic is a good thermal insulator can not be easily ignored.

Anyway, I now have a nice Antec 'Big Boy 200' case fan (arrived this morning: still in the 'playing with it; admiring it; showing it to friends' stage.)

I note that my HDD fans blow into the case. Thus the heated HDD air heads over to annoy the Mobo and other components. I will turn these fans around so that they suck instead of blow and let the Antec supply a breeze that takes over from the two side 8CM case fans as well as across the HDDs (exhausting out the two 8CM reversed HDD fans).

And it's summer in my neck of the woods. With two computers churning away in a smallish room the ambient temperature rises to 25C (77F).

Robert

 

[SNGX1275]

Your temperatures may be in a range where the drive appears to have no increased risk

Pages 5 and 6 of this google research pdf indicates that temperature, even 'extreme' (above 45C) has no statistically significant affects on drive failure until the 3rd year. And in fact, the cooler the drive is the higher chance you have of failure in your first 2 years. Annualized failure rates show a strong trend with increased failure at lower temperatures compared to high. If your temperatures were in the "moderate" range 30-40C you look to have the best shot at longevity.

I think if I'm reading figure 5 right, one of the explanations for the high failure for extreme temp drives in the 3rd year would be that so few of them died relative to the low temp in the first 3 years. Of course the 35-40C range would be the best.

It may just be concidence that your drives failed while being held by plastic. Do you know what temps your drives are running at?

Just some info about my drives:
Maxtor 200GB 6Y200PO - 34C
Seagate 400GB ST3400620AS - 31C
Maxtor 250GB 6B250SO - 33C
Western Digital 1000GB WD10EACS - 42C (This drive is an eSATA, so the enclosure is no doubt keeping the temp up).

So I guess based on google's data I wouldn't mind seeing my first 3 drives a few degrees hotter. I have them in a Cooler Master case, 2 of the drives have a 120mm fan blowing air into the case across the drives. I would guess they are the 21 and 33C drives. The other drive is mounted up under a CD drive, I expect that is the 34C drive (well now its up to 36C), I'd check but its a pain to actually look at the drives (case is in a bad position to get to easily). The 2 drives that the fan blows air across are mounted in a metal frame, but it does have those plastic drive fasteners.

 

[RKS]

I have been unable to get any Hard Drive temperature utilities to work on my computer - tried HDTune, DTemp and HDD Temp 4.0.9 beta - each report that there are no drives with temperature data. HDtune reports temperatures on my other computer OK (IDE drives).

Perhaps it is my OS (Windows Server 2003 SP1) or a switch in the BIOS?

I can use an infra-red temperature sensor but that would give the outside case temperature and may be quite different from whatever the internal sensor registers.

On the temperature issue, I just came across this "Hard Drive Temperatures: Be Afraid"
http://www.codinghorror.com/blog/archives/000748.html
which seems to be saying much the same thing as I have been saying. Would like to know those temperatures - might have a poke around in the BIOS and see what I can see...

Robert

 

[captaincranky]

I note that my HDD fans blow into the case. Thus the heated HDD air heads over to annoy the Mobo and other components. I will turn these fans around so that they suck instead of blow and let the Antec supply a breeze that takes over from the two side 8CM case fans as well as across the HDDs (exhausting out the two 8CM reversed HDD fans).
Robert

This concept flies in the face of conventional wisdom. Since the rear case fan exausts also. If the fronts were exhausting also there would be a vacuum at or near the center of the case.

In any event, the temperature rise in the incoming air may not be as severe as you are imagining it. This is obviously is dependent on the flow volume which would dictate the actual temperature rise based on heat actually dispersed per unit of time from the HDDs. Obviously, the machine you describe presents more flow restriction than the average 2 HDD home computer. It would be interesting to hear of the results of your experimentation in the future.

IMO the ideal airflow pattern in a case is basically the one the is the current norm, where the air flow is trafficked from front to rear, rather than something that is trying to pull at odds with itself leaving the center of the case at a lower pressure.
none of this of course, takes into consideration the implementation of the Antec "Big Boy" as the mother of all side panel fans. I guess that will be the "X" factor.

Antec has polished these principles with it's P180 line of cases, which don't even have side panel windows. Agreed that these may not be the ideal gaming case., but they are considered "thermally advantaged chassis".

 

[RKS]

The original Thermaltake design for this case included 7 8cm fans.
2 on the front panel blow in across the HDDs
2 on the side blow in
2 on the rear blow in
1 on the top exhausts

I'm changing that so that the Antec blows in at the side, right into the middle of the case. This cool air then flows in two directions:
1) Across the HDDs and out the front;
2) Across the MOBO and CPU and out through the rear and through the power supply.

I expect the Antec to supply sufficient air to keep a positive pressure (relative to outside the case) in the centre of the case.

In this way there is no accumulation of heat from the front to the rear with the hottest air venting out via the power supply. Note that the amount of power that a power supply can actually deliver falls off with increasing ambient air temperature.

If I can't get the S.M.A.R.T. system to work I'll have to check drive temperatures with an infra-red temperature sensor. This will very accurately give the temperature of the rear of the HDDs.

As two of my HDDs have died, there are now only four left in the case:
15,000RPM Seagate SCSI, 18gig, used as a boot drive and system only;
15,000RPM Seagate SCSI, 36gig, used as a programs drive and page file server;
Two 200gig Seagate SATA, 200gig, used for data (was 4*200gig in striped array-impressive in 2004 when installed).

Robert

 

[seanc]

RKS, what Seagate hard drives were these?
If they're 7200.11 - check out this on Seagate http://seagate.custkb.com/seagate/crm/selfservice/search.jsp?DocId=207931

Edit: Never mind I see your new post above mine.

 

[RKS]

Thanks. I have SeaTools but didn't think to have a poke around the Seagate website.

My HDDs are
ST300822AS (two of four remain)
ST318453LW
ST336753LW

But I think the ol' SATAs have done a great job and due for retirment anyway. Will replace them with NS types (24/7)

Robert

 

[SNGX1275]

On the temperature issue, I just came across this "Hard Drive Temperatures: Be Afraid"
http://www.codinghorror.com/blog/archives/000748.html
which seems to be saying much the same thing as I have been saying. Would like to know those temperatures - might have a poke around in the BIOS and see what I can see...

Robert

The codinghorror site says " According to our research, increasing HDD temperature by 5°C has the same effect on reliability as switching from 10% to 100% HDD workload. Each one-degree drop of HDD temperature is equivalent to a 10% increase of HDD service life." But they don't give any information on 'their research', so I have to be a bit skeptical. Surely they didn't have anywhere near the sample size as Google.

And on to google, they also have data on page 4 and 5 of that report I linked about HD failure and utilization. They did not break it down in percent utilization, they just have low, medium, and high. The only time High utilization is more than 4% higher failure rate compared to Low is at 0-3Months and 5+ years. 1 to 4 years old High utilization seems to fail on average 1% more often than Low (as in if 2% of the Low usage drives failed, you might expect 3% of the high use ones to fail). Their error bars on that data are really tight, so stastically it is significant. But to me, if a drive has a 1% higher chance of failing than another drive I wouldn't consider that a big deal.

 

[RKS]

The Google data gives utilisation as the number of read-writes to a drive that is running. But just running wears a drive down. I assume that Google had the drives running constantly as this is the environment their own drives operate in.

If, instead, we look at switch-on time per day then we note that the longer a drive is on the quicker it will age and the older the drive is the more sensitive it is to temperature.

OR

Those drives with a three year warrantee die after just over three years because the bearing dries up causing the unit to heat up. Drives with a five year warrantee probably show the same trend two years on.

Thus we see a rise in heat+failure at about three years in figure 5 and two peaks at 3 and 5 years in figure 2.
[Add-Edit: the first of my drive array failed at 3.0 years, second at 4.5 years (3y warantee), SCSIs have 5 year warrantee]

So I think that keeping the temperature down to the target area of 40C +/- 5C is what we're looking for.

On my No.2 computer, where S.M.A.R.T. does work, I have 37, 35 and 40C for the three IDE drives (200, 200 and 250gig).

I'll be able to compare those temperatures with the measured case temperatures to get some kind of calibration for measurements on my No.1 computer. That is the plan...now to find the time...I will be fixing other people's computers this weekend so it will have to be next week.

But I now have some good background thoughts to consider whilst I have the computer on the test bench - thanx

Robert

 

[captaincranky]

You guy do understand that "1,000,000 Hours MTBF" equals 117 Years don't you?

Do you think it's possible that the drive manufacturers are Bee-eSsing you?

 

[RKS]

I have my new Antec fan in and the computer is much cooler and quieter. The Antec pushes air into the side of the computer and all other fans exhaust (top of computer and at the front and back).

The plastic HDD slides appear to allow transient heat changes more readily, as when one defrags the HDD or does other HDD intensive activity. Having metal to metal appears to damp down these transients, though my computer does not seem to recognise SMART data from the HDDs so this is partially conjecture based on direct observation.

Robert