Suspended hard drives and reliability

mazir · Post by **mazir** » Fri May 21, 2004 3:31 am

Greetings,

I was just about to do Aphonos' bungee mod, when I started thinking if hard drives need to be connected to something rigid to function properly and stay reliable. Would hanging the hard drive in the air lower its life time and reliability? The drive head would have less stability.
What do you think on this matter?

Thanks in advance,
Cheers.

Post by **Ralf Hutter** » Fri May 21, 2004 6:07 am

Seems to me that there's a Seagate white paper that talks about that exact issue, and one of the conclusions they reached was that, since the drive head actually does have less stability in a "soft mounted" drive that seek times are a bit slower. Whether this decreased stability (and presumably more head movement required to access the data) would lead to increased wear/shorter life wasn't discessed, IIRC.

mazir · Post by **mazir** » Fri May 21, 2004 7:24 am

Thanks for the replies.

Maxamus, what you say doesn't make sense to me because if the drives were rigid then they would transfer their vibrations to the case. In Alphonos' mod the bungee cords allow the drives to move slightly and thus the vibrations do not reach the case.

Cheers.

mazir · Post by **mazir** » Fri May 21, 2004 12:07 pm

Maxamus, my point is that if you were to put a gyro, or some other motion detection device, inside the hard drive, or on the suspension bridge, as in your example, it would detect motion. This motion is what I am thinking of.

There is force being exerted in the hard drive, and if you know Newton's third law: "For every action, there is an equal and opposite reaction." then when the drive head seeks, or the platter spins, then force is being exerted and this is what causes the vibrations when the drive is connected to the case in the standard way.

Now when the drive is suspended by bungee cords, when force is exterted inside the drive the bungee cords allow the drive to bounce instead of passing on the vibrations to the case. Thus the drive is in motion.

Cheers.

icancam · Post by **icancam** » Fri May 21, 2004 5:10 pm

In my application of the "suspended HDD" technique, I've actually got elastic cord both above and below, so perhaps it should be referred to as a decoupled "tethered HDD". I did this because my modified Dell case is a clam-shell type rather than having a side panel that can be removed to work on the innards of the computer. I don't know if this is superior to a suspended system where the HDD or HDDs are simply hanging but it definitely gives me peace of mind when moving my computer around. It also allows me to use a level so I can precisely adjust the HDD to ensure that it is perfectly aligned on the horizontal plane (presumably lowering the wear and tear on the main bearing and thus keeping it running as cool and smooth as possible). My HDD is also actively cooled, only 3cm away from a 120mm Nexus fan. In an ambient room temperature of 29C, it is running at 34C or 16C below the manufacturer's maximum recommendation of 50C. (I was amused to see that Seagate's optimum temperature was defined as being room temperature!). I assume that any negatives associated with non rigid mounting, at least in my particular but not unique arrangement, are offset by improved life span due to consistently lowering the HDD's operating temperature.

hyperslug · Post by **hyperslug** » Sat May 22, 2004 10:42 pm

I also don't follow your logic Maxamus. A human eye would not be able to detect the difference in motion that is sufficient to cause errors in hard drive seeks. Further, the ratio of a human's mass and acceleration (F=ma) to the earth are not close to an actuator's relation to a hard drive.

From the Seagate whitepaper:

At one time, years ago, isolators were thought to be impractical for disc drive mounting; the frequent long seeks could induce excessive drive motion through isolator "wind up" and "release," which caused settling problems during seeks, which could in turn reduce drive performance and/or data integrity. Current drive designs, combined with Seagate's research and development of new isolator materials and designs, have largely eliminated these issues. In typical PC applications, over 90 percent of all seeks occur within only 30 tracks, which is less than 1 /4 of 1 percent of the drive's tracks, inducing very little motion and wind up. The materials Seagate recommends for isolators are highly damped rubber compounds, which minimize the possibilities for wind up, amplification and other unstable conditions.

Our testing indicates that WinBench scores (industry-standard drive benchmarks) are not affected by mounting drives using isolators. Further testing beyond the bounds of common disc drive functions did reveal that when performing high numbers of long seeks (long seeks are becoming more rare as track densities increase), write performance could be somewhat reduced, while read performance remains unchanged. Again, most drives perform many more reads than writes, so that overall performance, even in these extreme circumstances, is not perceivably affected.

Though Seagate says it doesn't make that big a difference in real world usage, the passage in bold (emphasis mine) indicates that the drive is not rigid and that there can be a measured difference. Also note that this article addresses mounting through "highly damped rubber compounds" and not the much less rigid "just hanging by bungee cords" method.

So there's a chance of a slight performance penalty if you isolate your hard drive and also depending on how loose the mount is. It's probably not enough to worry about, and most ppl here including me are more than willing to accept that for reduced noise.

Dunno about long term life. I don't know how different laptop drives are from their desktop brethren but they seem to be hold up pretty well to shock and movement.

MoJo-chan · Post by **MoJo-chan** » Mon May 24, 2004 11:24 am

Suspension might actually increase the HDs life. When hard mounted, vibrations cause the drive to move in one direction and "hit" whatever it is mounted to.

With suspension, the vibrations are smoothed out.

Jan Kivar · Post by **Jan Kivar** » Thu May 27, 2004 6:15 am

MoJo-chan wrote:With suspension, the vibrations are smoothed out.

Unless one tips the entire computer over...

Jan

wim · Post by **wim** » Thu May 27, 2004 6:55 am

Maxamus wrote:What im trying to say is that the HDD does not know if its rigid or not, its just a figure of speech.
I have suspended my 3 HDD's and they are not moving so wheather you can class that as rigid or not i dont know..
Suspension is a form of rigidness because it is supporting the HDD, it might not be a secure rigid but is is a form of rigidness.
Think of it like this:

Is a bridge deck rigid?
Yes it is because it is held up by suspension steel wires

Now apply this methodology to suspending HDD's and you'll see that its the same principle

As for alphono's suspension, well the drives do move but that will only be when force is excerted. Inside a closed case do they move? probably not.
Now what your saying is that they are not rigid because if they were they would be sending vibrations to the case. Well they are rigid because the bungee cords are holding them togther, they just arent fastened to the case.

maxamus your physics does not make any sense at all.
by rigid he means fixed in the ground frame, i.e. the drive does not move relative to the floor/case. a suspended drive is not rigid, it hangs there in the air and vibrates. the question is whether this decreases the lifetime of the drive.

as for this comment
"the HDD does not know if its rigid or not, its just a figure of speech"
it does indeed. if you were on an ice rink trying to dig a hole with a shovel, would you be able to tell the difference between wearing skates or spiked shoes? the drive vibrates at a high frequency because the moving parts inside exert small forces on the outer casing of the hard drive. when it is screwed into the case the effective mass of the drive appears much larger (to the drive heads, the hard disk appears to weigh as much as the whole case), and since the mass is huge the vibrations are tiny. however a problem arises, the large surface area of the case then acts like an amplifier causing noise.

now the material used to suspend the drive acts like a low pass filter between the case and the drive. however the mechanical energy dissipated by the drive still has to go somewhere. less goes into noise and more goes into vibrating the drive.

it would not surprise me at all if this was bad for the drive.

Putz · Post by **Putz** » Thu May 27, 2004 7:10 am

But what do you mean "bad for the drive"? What's going to break?

IIRC, don't hard drives' head servos use worm-gear setups anyway?

wim · Post by **wim** » Thu May 27, 2004 8:49 am

the head is mounted on an arm thingy which sweeps out radially over the platter (like a record player). however the head servo is geared you can't get away from the physics of the situation, that is conservation of angular momentum: the arm spins out one way and the drive spins the other way slightly. no problem yet. but then a short time later the bungee (or whatever suspension is there) pushes back on the drive and of course you get a reaction force on the head. very slight, but also the track on the disk is very small aswell. i'm not sure, but it seems to me like it would cause more wiggly write/read operations on the platter. i don't know whether this would decrease lifetime or just make seek times worse..

also the drive runs hotter that's for sure, and heat decreases lifetime.. when it's screwed into the case it dissipates a lot of heat through conduction but when it's suspended i suppose u just have to rely on convection for cooling.

Blappo · Post by **Blappo** » Thu May 27, 2004 10:37 am

When multiple drives are in a cage, the vibrations would be transferred beteween all of the HDDs. So they are able to deal with some vibrations. Also the bungie cords are elastic, so the force that it exerts on the drive would be gradually increasing proportional to the amount of stretch, and not all at once.

Also what about static charge? Should the HDD be grounded to the case?

hofffam · Post by **hofffam** » Thu May 27, 2004 10:57 am

Maxamus - you need to study a bit of physics. The suspended bridge analogy is not applicable in this case. In a suspended bridge the primary force is exerted by gravity creating tension in the cables. The bridge is effectively "stiff" in one direction. If by chance King Kong went underneath the bridge and lifted it - the bridge would move upward.

The impact of suspension on a hard drive is likely related to head movement. If you started the hard drive from "off" there would be an initial rotation in the hard drive case opposite the direction of the platters. This is the same thing as when you are sitting in a car and you feel the car "rock" slightly in one direction when you start the engine. Once the hard disk is up to speed there is no effect.

The head of a hard drive moves much like a record player tonearm. Relative to the mass of the hard drive the head assembly is very, very light. When the head has to move, some of the initial energy will attempt to move the disk drive case in the opposite direction. If the hard drive is rigidly mounted the head will promptly move to where it is supposed to. If the hard drive is mounted softly, the hard drive will move at least a tiny bit opposite the direction the head is trying to move. The hard drive will reach the limits allowed by its suspension (bungee) and the head will move to its target.

The question is whether or not this effect is significant or not. Reasons it probably is not:

- the hard disk case is very heavy relative to the head assembly
- the head is moving in both directions constantly so the hard disk case movement is cancelled by the very next head seek in the opposite direction

I bet Seagate, WD, etc. have studied this extensively. If it were a significant problem they would document that rigid mounting is required for warranty protection.

hofffam · Post by **hofffam** » Thu May 27, 2004 11:46 am

Wim - the head on the Maxtor drive I took apart is not geared. I'd be very surprised if a gear were used because gears have "slack" and the head movement must be extremely accurate. The head on the Maxtor is essentially a "motor" that is reversible. It has a limited range of motion. The voltage to the motor coils swings from plus to minus to make the head move left or right.

hofffam · Post by **hofffam** » Thu May 27, 2004 12:43 pm

Maxamus - your analysis of a suspension bridge is surely more accurate than mine. The compression you mention is the downward force exerted by the cables on the pillars. The bridge is not equally stiff in both directions. It would not be trivial to lift the bridge from underneath - but the primary resistance to lifting would be the horizontal tension in the road surface itself. I defer to your CE expertise....

In any case I hope you agree the bridge example is not particularly useful considering the impact of elastic suspension on hard disk vibration.

matt_garman · Post by **matt_garman** » Thu May 27, 2004 2:13 pm

Ignoring the arguments about vibration for a minute, what about the effects of not having your drive perfectly vertical (or horizontal) when using various forms of suspension mounting?

For example, in the Shuttle Zen SFF:

...The inclusion of two small aluminum cones that screw into the front corners of the base, prop the front up by about an inch... the slight slant puts the hard drive at an angle that's not a multiple of 90 degrees from level, which has traditionally been frowned on by hard drive manufacturers as contrary to long bearing life.

That's one of my hesitations to suspension-mounting my hard drives: it's hard to get the drive as level as you can when mounting the drive to a precision machined case. Surely it can be done, with patience and a level. But over time, the elastics will age and sag at different rates, requiring the drives' levelness to be re-checked periodically.

Thoughts?

Matt

hyperslug · Post by **hyperslug** » Thu May 27, 2004 2:33 pm

This might be relevant to the discussion:
(PDF)The Effects of Using Isolators for Mounting Disc Drives in PC Systems
Frank Bernett, Personal Storage Product Engineering, Seagate
It's an interesting read, this fellow knows his stuff. The only trouble is, he makes no indication about what kind of isolation he used other than they were "properly designed".

hyperslug · Post by **hyperslug** » Thu May 27, 2004 2:44 pm

matt_garman wrote:Ignoring the arguments about vibration for a minute, what about the effects of not having your drive perfectly vertical (or horizontal) when using various forms of suspension mounting?

AFAIK there are no facts that confirm or contradict the "Mounting at angles wears out the bearings" theory.

This is right on up there with "Cracking your knuckles leads to arthritis."

greeef · Post by **greeef** » Thu May 27, 2004 4:59 pm

right.

number one - if you want to know, test. There are plenty of hard drive benchmarks out there, and if one of you doesnt test it soon, i may well do, as i have suspended my drive.

number two - there has been absolutely no evidence saying that cracking knuckles is realted to the onset of arthritis. There has been evidence that suggests it may actually help clear it up.

http://www.newscientist.com/lastword/ar ... p?id=lw490

that explains what the noise is.

It has been suggested that this redistribution of the fluids can actually help clear up calcium deposits that do form. I know the only joint that i have that gets a little arthritic is the one i cannot crack (it's a family trait, got it off me dad)

number three - maxamus i'm afraid the physics here is quite obvious. The moving weight of the reading head needs to push against something. Usually it has the mass of the whole case, but in the case of suspension it can only react against the inertia of the hard drive. This must slow it's accelleration and increase seek time.

Perhaps big heavy heatsinks improve performance not only in extending hard drive life, but in speeding up seek times too!

finally i have to say i hadnt even considered making efforts to mount the drive at 90 degrees, perhaps i should.

griff

Post by **MikeC** » Thu May 27, 2004 6:37 pm

Here's a user report:

I've soft-suspended dozens of HDDs with clothing elastic that's even softer than bungee cords over the last 2 years. I've generally tried to get them level -- ie 90 degree angles to floor -- and shown owners of some of these PCs to check that from time to time. They have invariabley been fan cooled -- by a Panaflo at 5~6V blowing outside air over the drives. During this 2 year period, there has not been a single HDD failure I know of -- other than one or two of my own killed in action (ie, too much handling, accidents in the lab...).

Hard-mounting is probably safer -- but are you willing to pay the noise price? Not me. I am willing to take the risk of suspension shock mounting -- and have never paid any price.

Putz · Post by **Putz** » Thu May 27, 2004 8:32 pm

Exactly, MikeC. With all this Physics theory floating around, and claims that soft-mounting could reduce durability and increase seek times, it's easy to lose sight of the fact that nobody has ever noticed a performance or reliability drop (pun intended) from suspending their hard drives. So let's stop trying to figure out why drives should be failing or underperforming, and start trying to figure out whey they aren't.

hyperslug · Post by **hyperslug** » Fri May 28, 2004 4:30 am

greeef wrote:number one - if you want to know, test.

Exactly. Benchmarks beat speculation any day of the week, but we'd need to have like 50 hard drives mounted in cages, 50 by bungee cords and let them undergo typical user access patterns for workstations for 5 years and then note the survivors. We could speed up by the test by making it more intensive and shortening the time, but I wouldn't believe anything less than 1 year. So, .....who wants to fund this?

In the end, I'm betting it's not enough of a difference to matter, especially considering the noise.

number two - there has been absolutely no evidence saying that cracking knuckles is realted to the onset of arthritis.

Wish you were around in elementary school to get some teachers off my back about that.

hofffam · Post by **hofffam** » Fri May 28, 2004 5:40 am

I think the best evidence that suspended mounting doesn't harm drives is the language in hard disk warranties. They don't require rigid mounting.

I suppose a contrarian view is that warranties are now so short that the drive makers know that soft mounting has no impact on the first few years of a drive's life.

The Seagate white paper mentioned above shows the drive makers can track seek delays/misses caused by drive movement so if they wanted to they could expose that data if necessary. Maybe SMART could show this?

greeef · Post by **greeef** » Fri May 28, 2004 7:24 am

you know, in retrospect i reckon all the shocks that a hardmounted hard drive experiences (case bumping etc) would reduce the life of a drive far more than a soft mount.

griff