Sign in to follow this  
wabbiteh

Best method for mounting hard drives?

Recommended Posts

I'm currently building a new system for myself, essentially building the case from scratch (well, modifying an old tower case). So I have some options on how I'm going to mount my 3 hard drives. I was originally going to go for hard drive suspension using elastic band of some sort, but searches of various forums have suggested that not hard mounting a drive can be detrimental towards performance/reliability, both due to head over/undershooting and the drive not being level. I also read that multiple drives hard mounted and seeking near each other can cause vibration to run through the metal they are mounted to, and cause the other drives' heads to "jump". (See: http://forums.storagereview.net/index.php?showtopic=25007 , http://forums.storagereview.net/index.php?showtopic=23102 , http://forums.storagereview.net/index.php?showtopic=13493 , http://forums.storagereview.net/index.php?showtopic=19221 etc.)

So which would be more detrimental to a drive - hard mounting three of them together, or suspending them seperately? Is it even anything to worry about? I'm not as concerned about speed as I am about reliability and avoiding data corruption/errors.

Thanks for any help :)

Share this post


Link to post
Share on other sites

From what i have read in those posts, the best way is to mount hard disks is to hard mount them and damp them using mass ala Fujitsu's test where they attached a 20kg lead weight to the drives. This basically means ANY kind of external movement is bad for your drive. Unless you are going to make a custom enclosure, this isn't really going to be possible.

The fact you are considering elastic suspension suggests you care about noise. Just buy quiet drives like Samsung's Spinpoint P series (I own a T166 and although its quiet when reading, write cycles are quite noisy) which are barely audible and mount normally in your tower case using screws. Doing this is the simplest option and will achieve 97% of what you want for very little effort. I did this and have basically had a practically silent hard disk set up. If you find you get unwanted vibrational noise (I had non, with rigid mountings) then use screws + silicone or rubber grommets.

Simple = Best

Edited by student

Share this post


Link to post
Share on other sites
So which would be more detrimental to a drive - hard mounting three of them together, or suspending them seperately? Is it even anything to worry about? I'm not as concerned about speed as I am about reliability and avoiding data corruption/errors.

Thanks for any help :)

Well, suspending them together might not be a good route to go:

http://www.dtc.umn.edu/publications/reports/2005_08.pdf

For best reliability and lowest data corruption/errors, be sure to follow your HDD manufacturer's instructions regarding mounting (and grounding.)

Edited by datestardi

Share this post


Link to post
Share on other sites

While Samsung HDDs have pleasant noise character and low acoustic noise volume, that doesn't mean they vibrate less... in fact Samsung is The Worst HDD there is when it comes to vibrating (meaning constant vibration on idle, every HDD vibrates noticeably during seeks). HDDs should definitely be soft-mounted and that recommendation goes twice with Samsung.

Wasn't there an article that claimed that very hard mounting (a la 20kg of lead) is the best, followed by very soft mounting, and that half-hard/half-soft mounting is actually much worse than either extreme? Considering typical metal chassis bends and vibrates with the drive (also creating noise doing so), can "hard-mounting" even be called hard-mounting? It's possible that not only is it the noisiest solution, it might also be the worst for performance/reliability. (This of course depends on the rigidity of the chassis.)

I think soft-mounting should be safe from reliability point-of-view. Unstable read wouldn't cause data loss (as there's ECC to notice read operation failing) but only lost performance and writing is started after HDD controller has confirmed seek to have settled. If seek overshoots, settling takes longer. No performance issue.

If on the other hand seeking "winds" the suspension and seek settles, and after that energy stored in suspension "unwinds", head may lose track following AFTER seek has settled. That could probably cause not only performance but also reliability concerns. For this to happen, "unwinding" must occur at high enough force... soft-mounting with rubber bands is way too soft to jerk the HDD after seek has ended. Thin walled computer chassis might do this and so might mounting with hard rubber pieces.

I think that if you really decouple the HDD, decoupling it well is the best. Not only will it make more sense from noise-perspective but possible from reliability perspective too.

I think the biggest reliability concern with decoupling is if your rubber bands snap and HDD suddengly drops. Use fabric strengthened rubber bands intended for clothing (consisting of several smaller bands) instead of one-piece rubben band used in kitchen. Fabric band shouldn't snap suddenly and reassuring the condition once in a while should be enough... probably best to check it every time you clean up the case from dust, be that once a month or once a year.

Share this post


Link to post
Share on other sites
Wasn't there an article that claimed that very hard mounting (a la 20kg of lead) is the best, followed by very soft mounting, and that half-hard/half-soft mounting is actually much worse than either extreme?

The article said that where additional mass or extreme rigidity is not practical, then damped mounting is the best (for non-mobile HDDs at least) and undamped mounting is the worst. Whether the tested non-rigid mounting schemes in the article were found to be good or bad depended singularly on the damping that the mounting scheme exhibited.

The article concluded, "The best results come from adding additional mass to the HDD. However this is not practical since.... The better solutions are to increase the resonance frequency by increasing the support rigidity of the mounting system, or to decrease the resonance gain by adopting improved dampening methods."

http://ieeexplore.ieee.org/Xplore/login.js...arnumber=753800

To the extent that a rubber band is undamped (i.e. it allows the HDD to bounce like it were suspended from a bungee cord), I believe it would not be good.

Share this post


Link to post
Share on other sites

Did someone say ANAL?! How much difference is this mounting scheme going to make to wabbiteh's life? I understand the silence issue but that can be solved really really easily just by buying the right drive.

The Fujistu article stated that soft mounting was second worst, with the worst being the way 99% of consumers mount their drives... with it in the hard drive case mounted on rails as designed.

If this issue were relevant, then every PC case manufacturer would be sued to high heaven due to data and drive failures.

Just mount it how you want and keep a regular back up regime in place.

This is like discussing whether a 1 cent piece should be 1mm smaller in diameter to save copper and ease the burden of having change in your pocket, or if spending $20,000 to get an extra 10bhp out of a 600bhp Ferrari is worthwhile.

Damn it will log off... rant over.

Share this post


Link to post
Share on other sites

Student, you could have logged off before your rant, and none of us would have been the poorer. I'm sure you weren't trying to say wabbiteh's question was dumb (it wasn't by any means, seeing as Fujitsu published an IEEE a paper on it), so I wonder what the problem was.

There is a lot of rubber-band mounting "nonsense" (my word, I'll let whiic have the last word if he wants it) out there which violates manufacturers' mounting recommendations (and grounding recommendations) and does not constitute a mounting arrangement which is necessarily tested (for reliability) by the manufacturers. I was trying to steer wabbiteh in a more proper direction. And whiic was trying to be helpful too, though I may disagree with him.

Share this post


Link to post
Share on other sites

sorry, it wasn't meant to be a personal attack... I thought it was interesting until I read the Fujitsu paper which stated that the worst method of mounting was the way 99% of consumers put their hard drives into their cases. This basically makes the whole question a non issue. If the worst way of mounting is good enough in everyday reality not to cause hard disk problems for the vast majority of consumers then wibbeteh can mount his drive anyway he wants and not face problems with data loss etc.

You could theoretically discuss what sort of liquid would best boil an egg. In a lab, you could probably measure a difference between water and molasses and alcohol etc... but you know what? Plain water is good enough in everyday reality. Its just a non issue.

Share this post


Link to post
Share on other sites
sorry, it wasn't meant to be a personal attack... I thought it was interesting until I read the Fujitsu paper which stated that the worst method of mounting was the way 99% of consumers put their hard drives into their cases.

I'd bet almost 99% of consumers don't have 3 hard drives in a case.

When you mount 3 drives together/in an array, you will do very well to think about how you're mounting them. The U.Minn article goes in to some of that.

Share this post


Link to post
Share on other sites

I had a quick read through the U. Minn article. From what I gather, YES vibrational effects are measurable, (especially affecting seek operations) but NO they don't cause data problems. The more heavy damped, rigidly contained drives in the center of a stack suffered the least from vibrational effects, confirming the findings of the Fujitsu paper.

But if there were no data problems, doesn't this go back to the "what liquid is best to boil an egg" thing? Why bother with something that in everyday reality isn't a problem? If you were running some sort of server farm that thousands of people depended on then maybe, but for single desktop use (that should have a regular backup regime in place)?

Share this post


Link to post
Share on other sites

If you are going to take things this far, shouldn't wabbiteh also isolate his entire computer case, incase someone walking in the same room causes vibrations that will affect his hard disks? Unless the floor in his room is concrete, it will vibrate significantly. Ask someone to walk across the room you are in and place your hand on a desk. You will definitely feel the vibrations cause by every foot step. He should also really ban loud music and other noises too as these too will create vibrations in his hard disk.

Do you see where this is going? All this to solve a data problem that doesn't exist.

Edited by student

Share this post


Link to post
Share on other sites
Do you see where this is going? All this to solve a data problem that doesn't exist.

U.Minn was not trying to test disk drive reliability. Their testing didn't last long enough and/or measure the most indicative parameters for that. Yes, they found no evidence of data corruption in their tests, a tribute to the robust design of todays HDDs.

Seagate put it best, perhaps, in their 2000 isolator white paper, when discussing the impact of vibration/mounting on disk reliability:

"Statistically speaking, there is a higher chance of a data-integrity problem occurring when a higher write retry rate occurs in a disc drive."

I think you can see write retries manifest themselves in the U.Minn paper.

Seagate also described other problems that soft-mounting can cause at page 5 of that paper:

http://www.seagate.com/docs/pdf/whitepaper...oustics_qrk.pdf

Do you see where this is going? All this to solve a data problem that doesn't exist.

Todays HDDs are robust and most can tolerate sub-optimal environmental conditions much better than yesteryears drives. However, the effect of user handling on disk drive reliability today is not so much the "non-issue" as you might think it is.

Share this post


Link to post
Share on other sites

Wow, this got more of a response than I expected. While datestardi and whiic were perhaps the more helpful of the three, student did manage to put things in perspective :P (Being anal isn't *all* bad... But seriously, it was helpful.)

Based on everything I've read so far, it doesn't look like bungee-cord hard drive suspension is going to be any real danger to data - the risk is there, but so long as not every single write operation is going crazy, it should be okay. That said, being as neurotic about my data as I am, I'm going to give extremely rigid mounting a shot first, and if the noise is absolutely unbearable, I'll suspend them and just try to ignore my worries :rolleyes:

(Random unnecessary info: The plan for mounting them is something along the lines of: aluminium square tubes set up like this: [tube][==drive==][tube] (front view), each drive getting it's own set of tubes in order to quell vibration. While seperate tubes for each drive aren't strictly necessary, it's actually easier to set it up that way, and by some voodoo magic this all fits perfectly into the design I had.)

Thanks for the help.

Share this post


Link to post
Share on other sites

datestandi: "The better solutions are to increase the resonance frequency by increasing the support rigidity of the mounting system"

So, the "better" solution than 20 kilograms of lead (or same weight worth of feathers, since 1 kg of anything weighs the same), is to increase resonance frequency higher, there human hearing is more sensitive.

So, 7200rpm HDDs (idle) vibration occurs at 120 Hz (first harmonic, fundamental frequency). There's also upper harmonics like 240 Hz (second harmonic, first overtone), 360 Hz (third h, second o.t.), etc. (multiplies of lowest frequency). So, increasing resonance frequency will reduce amplitude of vibration (because upper harmonics are less intense) but not prevent it from happening. Human hearing is very poor at 120 Hz (which is around where subwoofers operate: you feel it, not much hear of it) but it gets more sensitive not too much above that. You can hear upper harmonics as humming tonal voice. Having just the HDD vibrating is much less irritating than having a computer case capable of amplifying upper harmonics.

student: "I understand the silence issue but that can be solved really really easily just by buying the right drive."

Sorry about the swearing but I'd better say it frankly: that is utter, complete bullshit.

HDDs that vibrate the less don't necessarily have the best acoustic frequency properties and those that don't produce much acoustic noise may not be the least vibrating ones. Samsung is a perfect example. I believe there's quita a bit of concensus on SPCR that Samsung is The Drive for nearly all silent/very_quiet builds. Only when decoupling is not possible or if Samsung is not available locally (since there's people from all over the world in SPCR) it's reasonable to recommend a WD... or Hitachi (latter mostly for some special purpose such as media storage, due to APM).

If you go for WD because they generally vibrate less than Samsung (in expense that it causes more acoustic frequency noise!) you may still receive a bad sample of WD. You can't cherry pick your HDD since they are sold in electrostatic bags! While acoustic noise typically varies very little (aside from really bad or even faulty samples), vibration is a hit-or-miss thingy. Nothing is guaranteed or even likely.

datestandi: "There is a lot of rubber-band mounting "nonsense" (my word, I'll let whiic have the last word if he wants it) out there which violates manufacturers' mounting recommendations (and grounding recommendations) and does not constitute a mounting arrangement which is necessarily tested (for reliability) by the manufacturers."

Sure, they are not necessarily tested in perfectly decoupled rubber band cradle (aside from some studies like that on made by Fujitsu). At least it's not part of factory burn-in and testing process that all HDDs goes through. But in lack of a real reason to be afraid I wouldn't say it's dangerous either. It's just something unknown.

Before you correct my opinion, I do recognize that HDA accelerating in opposite direction than actuator is a factor, but what I mean is that even with "hard"-mounted HDDs (to a thin steel chassis), they will vibrate. It's tension and release with steel chassis acting as a relatively stiff spring load. I see "hard"-mounting potentially dangerous too because steel chassis may "release" after seek has settled and cause an off-track write and data loss. Obviously this is impossible with extremely soft mounting (i.e rubber band cradle) because resonance frequency is down at around 1 Hz.

datestandi: "When you mount 3 drives together/in an array, you will do very well to think about how you're mounting them."

All the more reason to decouple the drives because when (individually) decoupled, they only suffer from self-cause vibration: a harmonic and predictable vibration from spindle inbalance and vibration from actuator movement which is also predictable because it occurs at the same time as the seek itself.

Vibration caused by other drives cannot be predicted before they occur. Individually decoupled drives perfectly eliminates this problem so I wonder why this subject always comes up when talking about decoupling... and usually it comes up as a threat instead of as a solution to the threat. I'm certain it's the latter instead of the former.

Decoupling a HDD tray with multiple HDDs... that's a potential increase of risk, I have to admit.

wabbiteh, "Based on everything I've read so far, it doesn't look like bungee-cord hard drive suspension is going to be any real danger to data"

The absolutely biggest threat is sudden, catastrophic physical damage due to severe impact to the HDD if it falls down either due to breaking of the suspension, sliding off the cradle (only if you didn't secure it properly) or because computer case was transported in non-upright position. That's the absolutely biggest threat I see.

Not only a threat but it's also an inconveniance matter if you have to move you computer... having to remove HDDs from the system first. But I think it's worth it if your ambient noise level is low enough and you don't have other very noisy components in your case (stock CPU HSF comes to mind).

If you are interested in silencing, http://www.silentpcreview.com is the best source of information. They don't intend to be jacks-of-all-trades so they don't do any performance testing (except for cooling performance of heatsinks). They'll leave performance testing for SR, Xbit, iXBT... and unfortunately for people like Tom's Hardware too.

And Heatsinks aren't tested with SilenX Ixtrima or better yet, 200cfm Delta, like some other's do but instead using a Nexus reference wan undervolted to lower rpm than stock. Nothing is more stupid that trying to obtain ultimate overclocks with HS intended for passive or otherwise ultra-low-airflow cooling.

I've been chatting there lately but I might do a comeback here. At least to some degree. I'm still waiting for that 1TB Samsung for my new quiet Q6600 G0 build (OC @3.15 GHz @1.35V). It's taking forever to become available... :(

Share this post


Link to post
Share on other sites
student: "I understand the silence issue but that can be solved really really easily just by buying the right drive."

Sorry about the swearing but I'd better say it frankly: that is utter, complete bullshit.

I don't get it? You say its utter complete bullshit and then prove my point?

Only when decoupling is not possible or if Samsung is not available locally (since there's people from all over the world in SPCR) it's reasonable to recommend a WD... or Hitachi (latter mostly for some special purpose such as media storage, due to APM).

If that isn't you recommending the "right drive" for the quietest operation, I need to go back to school.

Edited by student

Share this post


Link to post
Share on other sites

I recommend Samsung decoupled. WD if it can't be decoupled. Hitachi if harddrive is infrequently used. There's no single silent drive for all purposes. Choosing the right drive is important but you said:

"I understand the silence issue but that can be solved really really easily just by buying the right drive."

Right drive is necessary but that's far from all that is needed to consider. You simplified it too much.

Samsung hard-mounted (bad) -> WD hard-mounted (good) -> WD decoupled (better) -> Samsung decoupled (best)

So, what's the "best" drive? I can only tell what's the best combination or best suited drive for certain mounting method.

To be accurate, I cannot really tell what's the best combination as there's some new drive series that I have neither heard or read a review of. There simply is no general "WD noise profile" that all WDs would match. Same applies to all other manufacturers.

WD's 5400rpm 1TB GreenPower certainly stirs previous simplifications about "WD (hard-mounted) / Samsung (decoupled)". Now with GP, it may already be "WD (hard-mounted) / WD (decoupled)".

Share this post


Link to post
Share on other sites

I have drives running almost every single day that where made when their 8 gigs of space was a common item on store shelves. What I am getting at is even if the standard drive install is the worst for drive or data longevity, some of them survive quite well. And for years too. Conversely, I have had brand new drives fail in a few days. Was their failure due to evil mounting practice?

Share this post


Link to post
Share on other sites

Interesting discussion...

I can't hear my drives at all, but there is two reasons for that. One is a solid built case that is thick aluminum side panels and steel frame, two is that it has so many fans that's all i hear anyway! lol

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this