Too many physicists here...and they cannot agree in something supposedly simple, not even the string theory or quantum mechanics!. Auch!!.

Regards

Not to go off on a tangent, but what are bad temps for a HD? Mine are between 31 and 42C...

The limit is supposed to be somewhere around 55-60ºC.

SPCR Sticky: Max safe temp for hard drives?

Cheers,

Jan

So can I mount my Seagate 7200.7 upside down horizontally? I heard that some drivers makes errors if installed upside down.

Yeah, I agree, I think they just haven't bothered to test it properly at odd angles. and to be sure, they say "mount vertically or horisontal".

AtW

Of course.

Yeah, luckily I have SPCR experts correcting my thoughts

Smart ass

Since my other thread was locked.. I've read through this thread before, but there seems to be no general consensus??

As for the poster asking about the temps, those are fine. Mine run at 45-50 degrees idle:D

There may never be a general consensus. Some of us have attempted to show that the concerns listed in this thread are not a significant risk. If there are other factors, no one has brought them to our attention. And if anything you've read here doesn't make sense, voice your question in a specific manner and it will likely be answered.

Sorry I can't help myself...

I, along with many others, have found there to be NO PROBLEM WHATSOEVER with mounting drives at any angle. There has not been a single incident reported anywhere of drive failure being caused by the angle at which it was mounted.

Bottom line: MOUNT YOUR DRIVE ANY WAY YOU LIKE

I feel much better now...

Hello,

I've read through this entire thread and have thoroughly enjoyed all of the posts, both critical and informational and am overly impressed by the minimal flaming as well. I stumbled accross this from my perusal of P180 posts and am glad that I did.

To add something to this thread that I am suprised was overlooked, but might have been on purpose due to it's relative small percentage... All of the formula's and theory here is based on the hard drives being mounted 90 degrees to the gravitational force of earth. Then the argument is about mounting the drives in a manner which is not exactly this 90 degree fashion. The part ignored however is that I'm willing to bet that if you took a level out and checked, the chances of your case being level are slim to none anyway. There in my estimation is probably a 2% variance in most cases across multiple axises. So even if you are mounting your hard drive with the recommended 4 screws horizontally or vertically in your case you do probably fall within the 5% tolerance that the manufacturers recommend, but still are not exactly at 90 degrees.

In this retrospect, the fact that all of are drives do not die rapidly and in the many many drives that I have used (mainly in RAID 5 or 10 arrays, so a large amount of drives overall) have maintained acceptable longevity, attests to the thought that it really doesn't matter too much what orientation your drive is mounted. Granted this is not your 45 degree example in two directions, but I still have 4 IBM Deathstars that perform quite well even after hanging them in free air due to a lack of space to mount in a transition case for a period of months of continuous use.

Physics aside, in real world useage, from most manufacturers that I've used (Maxtor, WD, IBM, Seagate), you'll encounter other issues before you have a drive die from mounting orientation. What issues you might ask?... In my experience, size and speed of current drives vs. what you can get now

Thanks for the excellent read to take away the doldrum of my work day.

David - penguin22

Btw... I work as an IT Manager and build our systems here. Other than the IBM Deathstars, we rarely have had any hard drive failures. And this is for our pseudo horizontal mounting... pseudo for what I said above.

One argument that I have not seen (maybe I overlooked it) is simple linearity. Classical mechanics is linear. In the absence of non-linearities, there is nothing magical about having forces along several axes at the same time. Take the gravitational force when the drive is at an angle, decompose it into three components along the edges of the harddrive. The magnitude of each one of them is smaller than 1g. As per manufacturers claims, the harddrive can handle an acceleration of +-1g in any of those directions. Be done, we're in spec.

wow, this has been an intensive refresher course in gyroscopes and physics. Now I'll have to learn to forget it all over again!

Being a scientist for my sins, every cell in me is screaming "empirical evidence!" Considering SPCR is probably one of the largest bodies of people not mounting their HDs in the manufacturer specified manner, I would think it would be perfect for a poll, simply:

Has anyone, having mounted their HD in a non-manu specified manner, had it fail within the warranty period?

Obviously, this wouldn't control for other causes of HD failure, but if the theoretical sceptics (sceptical theorists?) are correct then we should see a no. of failures. If >2% of pollists (2% being the average HD return rate, IIRC) have had failures then we can consider there to be an issue. >150 pollists would be ideal, to get a statistically useful no. of failures at that rate.

I've had my Seagate 7200.7 IDE on it's side (rotated around the long axis 90deg ~ 1yr), on it's end (rotated around the short axis 90deg ~1yr) and upsidedown (around the long axis 180deg ~ 3mths) with no noticeable difference in SMART values.

I had my old Maxtor drive (~'01) running upside down for a few years and it has survived to this day. However 180° is an even angle rather than an odd angle

didn't read the whole thread...just pointing out the popularty of vertical external storage now-adays, i-pods, walkmans and laptops (people run around with those), and vertical storage cases (antec minuet).

facinating thread!

not convinced about the gyroscopic stuff, gravity is a constant force. You need to add a force to the system to create precession.

Also, as the disk is symmetrical and perfectly balanced, tilting it wont place extra force on the bearing in order to tilt it further and cause problems with the heads hitting the platter on long seeks. It will create slightly more pressure on the topside of the bearing due to the weight of the platter though, and vertical mounting will max this difference. I think you'll still get even wear on the bearing, but perhaps a shorter life-span.

Nice analysis. I agree with D235hadow, Tibors et al.: there's no reason to reason to avoid mounting drives in (for example) 45 deg angle ...or if there is, at least it has nothing to do with gyroscopic effect.

If there is some problems with uneven wear, the uneven wear of bearings would be at it's worst when mounting at 90 deg angle (not at 45 deg). Bearing wear (uneven or even) isn't a typical problem with fluid dynamic bearings.

If there is a problem with gravity pulling the heads closer or away from the platter surface, the problem would be at it's greatest when drive is either mounted at 0 deg (PCB down) or 180 deg (PCB up). I seriously doubt it's a problem with any drives.

Someone here was afraid about mounting it at 180 deg. I wouldn't worry. At least it's not the heads crashing (unless they intentionally made different spring forces for head on the "upper" side of the platter than on the "lower" side. But that isn't the case as the drives are promised to operate in 6 orientations. Only real reason why a drive wouldn't work in 180 deg tilt would be worn out or otherwise defective thrust bearing. HDDs have two sets of bearings, one for sideways forces and also thrush bearings to counter forces "up" and "down".

Or actually forget what I said... with ball bearings, the bearings can handle both sideways and perpendicular forces. Only FDB drives have several bearing surfaces to to handle with different forces.

Hitachi white paper on FDBs (includes images of both a FDB and a BB motor, on the last page) (the BB motor presented is a bit unconvetional as it's supported only from one end where as typically they are supported from both ends and FDBs only from one end) http://www.hitachigst.com/tech/techlib.nsf/techdocs/089C4B963AEE9A6F86256D340075052F/$file/FD_White_Paper_FINAL.pdf

Hitachi white paper on 7K400's FDB motor (which is supported from both ends, unlike most FDB motors) http://www.hitachigst.com/tech/techlib.nsf/techdocs/AD40213FFFA789C186256E6600625ED6/$file/WP_MEnhance_25March.pdf

The forces applied to thrust bearings are at their greatest when the drive is mounted normally (0 deg) or at 180 deg. (Note, it's a different bearing that supports the spindle when it's tilted 180 deg.) And when it's tilted 90 deg, the forces acting on "main" bearings are at it's greatest. Both orientations are perfectly OK as stated by every manufacturer.

Mounting at 45 deg doesn't always receive the manufacturer's blessing but I don't see a reason why both thrust bearings and main bearings cannot be loaded simultaneously (both are stressed only slightly, less than mounting on any horizontal or vertical position would stress a single bearing).

The "mountable in 6 orientations" is probably a way to tell even the stupidest end-user that the drive can be mounted upside down or vertically and at the same time discouraging people of attaching drive with inadequate number of screws.

It might also mean, that to limit test scenarios, the drives are designed to operate only in 6 orientations. That doesn't mean it won't operate in any orientation what so ever, but it means they test it only in 6 orientations. To test it in all orientations it'd require INFINITE number of test. Even testing it at every 10 degree of tilting would be nearly impossible (as each testing would require hundreds or thousands of drives).

Consider: 00 deg tilt, 10 deg tilt, 20 deg tilt, 30 deg tilt, 40 deg tilt, ..., 80 deg tilt, 90 deg, 100 deg, ... 170 deg, 180 deg, 190 deg, ..., 350 deg. Then testing it tilting it over another axis 00, 10, 20, 30, ... 350. Then testing all combinations of tilting around both axis! If there's 36 tests just measuring tilting over one axis and 36 test over the other, there's 36*36 =1296 different orientations, and each requires several hundres of drives being tested over a period of several days. It's much easier to test the drive in only 6 orientations than 1296.

It'd also be pretty useless to test them in 1296 orientations as there's no real reason to expect a reduction in reliability if the bearing load is distributed over more than one bearing. It's less load on a single bearing, more bearings loaded, the total load being the same.

So I consider this a case closed. I won't be afraid to mount my drives in ANY orientation imaginable, as long as the drive doesn't change it's orientation during operation. (Elastic suspension is a whole other thing, discussed in other threads.)

___________________
Aaargh... way too many typos. Editing to fix the typos for the 4th time. (Logical content unchanged.)

whiic, suberb all encompassig post... However.....


With all due respect... you 'failed' to address this one last burning issue:...





Have you any thoughts?...





Regards
Pete
(I think personally, that the glue used to secure the magnetism on would make it OK... )

OK, peteamer, let's address that as well (even though it's no doubt just a comic relief).

The "glue" can hold the magnetic attached to the platter with a certain force. This is similarish to static friction ("stiction").

To slide a bit off we need centrifugal "force" (which isn't a force, but just inertia, i.e the particles tendency to conserve it's speed and direction).

The platter of a 3.5 inch drive is actually slightly bigger than 3.5 inches. The 3.5" was given to that form factor HDDs because it's the size of floppy drives. Floppy drives are bigger than 3.5" but the floppies themselves (or actually the flexible magnetic disc inside it) is about 3.5 inch in diameter.

Well, I don't know what's the diameter of a 3.5 inch HDDs platter. Probably around 4 inch, which would translate into around 10 cm. The radius is about 5 cm. (10000 krpm and 15000 rpm 3.5 inch drives have much smaller platter even though they have similar size casings. But let's just consider 5400 rpm and 7200 rpm drives.)

The accelerating force on the particles on the media is towards the rotational axis and is created by the stiction alone. The speed of media on outer diameter is 120 rps (7200 rpm divided by 60) times 0,05 meters =6 meters/second. Thus acceleration towards the center is V^2/R = (6 m/s)^2/(0,05 m) = 36 m^2/s^2/m = 36 m/s^2. Since m/s^2 is the correct unit of acceleration, I probably used the right formula. 36 m/s^2 divided by 9.81 gives 3,67 G acceleration.

So, the stiction has to overcome 3,67 G forth of acceleration when the drive is horizontal. When vertical, the stiction has to overcome 2,67 G when the media particle is in it's uppermost point and 4,67G when it's in the lowermost point.

I don't think the variation of 2 G between minimum and maximum causes vibration of the particles in the media (the particles are just too small to have any room to vibrate) so I don't think this play any mentionable role in particles breaking free of the media. The maximum force could play. But I think that if it's designed to withstand 4 G, it can handle 5 G.

Thus the media will remain on the surface of platter... unless one of the following occurs:
- head crashes onto media, scraching media off the platter with tremendous force per area (the force might be small but the area of collision is tiny).
- corrosion or other chemical reaction breaks the stiction and causes the media to flake of the platter. This is one of the numerous suggested ways a Deathstar can die. Deathstars have glass platters and it's speculated the media didn't stick to it properly.

But to answer StarfishChris's question: yes, your bits might slide off the platters if you happen to own a Deathstar. Otherwise I would worry of other ways of HDD dying than of media flaking off. (There include: motor controller burn-out, corrupted firmware, head crashes, formation of bad sectors, bad servo control sectors, mechanical wear-out(*), etc.)

(*) where as FDBs aren't suspectible to bearing wear, drives with "fluid dynamic bearings" have just one fluid dynamic bearing... and the actuator assembly which uses ball bearings! Why? Simply because fluid dynamic bearing requires constant motion and don't like when the direction of the motion changes. Fluid dynamic bearing may have physical contact when they spin up/down while being virtually immune to metal-to-metal contact while operating. FDB bearings in actuator assembly would wear itself out very quickly without external "oil pump" to keep the "oil pressure" up when actuator stops.

Correct me if I'm wrong.

_____

EDIT: I made a stupid mistake as I was too tired to actually realize how silly results I got from my calculations. I forgot 2*Pi from the equation when calculating the speed of media and that really made the results very very far from the truth.

The speed of particles on the outer diameter is 120 rps * (2*pi*R), R being the radius, 0.05 meters. 2*pi*R is the circumference (0.314 meters). Thus the speed is 37.7 meters per second (136 km/h or 84 mph).

V^2/R = (37.7 m/s)^2/(0,05 m) = 28400 m/s^2 = 2900 G.

So the force when the particle is at it's lowest point 2900 G + 1G = 2901 G. And at it's highest point 2900 G - 1 G = 2899 G.

I also made another miscalculation. When I got the "6 m/s" instead of "37.7 m/s", I place it into the following formula:
"(6 m/s)^2/(0,05 m)" which would be correct if the 6 m/s was correct (but it's not). The I got " = 36 m^2/s^2/m" which isn't the result of the previous quote. The "correct" result would be 6*6/0.05 = 720 m/s^2 which is approx 73 G, so I basically forgot to divide it by 0.05 (i.e multiply it with 20). Gee, I must have been sleep writing.

Any way, I don't think there's any reason to worry about bits sliding off the platter.

I have an 40 gig Seagate drive that worked great for about a year. I then installed it upside down and after a couple of weeks it would not spin up when I started my pc. Tapping the drive with my finger during startup worked most of the time.

It's now installed the right way up and works fine again.

Years ago,I did testing at Seagate,drives were "right side up" but drives are much more sophisticated now (and we just tested that they worked,not MTBF. I have used drives right side up,upside down and sideways-have not seen much difference.

just to be correct
it wont be a problem if i mount my 2 WD HDD's so the spot where the power and sata connections will be facing the bottom of the case
and the hdd itself will "stand"
any problem with that?

Ok has anyone run a benchmark to test any PERFORMANCE difference between mounting positions?
Perhaps in some mounting positions gravity causes the disk to become "heavier" and spin marginally slower?
Just a HD tach between both regulation mounting and 90 degree would be great, possibly repeating a few times to iron out errors.
Im on a laptop this week

Would it be a problem to change the HD from horizontal to vertical after a longer period?

Actually it is an odd angle. 180° = 1π (1*pi)

"Would it be a problem to change the HD from horizontal to vertical after a longer period?"

Badly worn-out ball-bearings might no like that, but for drives with fluid dynamic bearings, it shouldn't be a problem as they don't wear out that way. (Bearing fluid may lose it's properties or evaporate under extreme condition such as high humidity or temperature, but it wouldn't develop excessive free play like BBs that make physical contact with metallic or ceramic bearing balls.)

I have a ~4GB Maxtor mounted upside down since when it was first installed by HP and a Seagate U5 that has been runnin for years vertically, IDE connector pointing straight down. Not odd angles but not very common either. They operate in other orientations as well, even though they are getting old and have ball-bearings.

Thanks for the reply. I have a WD 2500KS SataII HD. I think that's FDB.

Leaving aside for a moment the possible gyroscopic effect, I'd like to look at the device in question as if only purely gravitational effect was operative.

In this case, any orientation but case-horizontal would result in an uneven load spacial distribution on the bearing(s).

Whether mechanical or not, uneven force is a definite contributor to the shortening of bearing life.

In mechanical (ball) bearings, uneven force causes uneven wearing of the ball-track surfaces as they pass across each other. The interior trace interaction cycles must look really beautiful when mapped out. I believe that while operating, ball bearings must make quite a number of shifts in their relative motions, but a persistently off-balance load will create an 'attractor' in the phase space of the dynamic system within which they operate.

Since the phase space of the system is three or less it can be directly represented without dimensional disard or remapping, ie. in 3D space. The 'attractor' will have a geometric shape composed of the first primary load distortion and its polycyclic rotational harmonics (generated by the bearings, runs, support structures).

Keeping a bearing running a long time means keeping those attractors from forming, ie. maintaining symmetric loads at all times, among other things (such as lubrications if necessary).

In the FDB case, fluid will normally form higher-resistance pools in areas of lower-than-average load, and become thinner in areas of high. At high pressure and with certain designs it will behave differently, helping instead of harming the shaft's equilibrium.

However, nonsymmetric forces will introduce changes in fluid flow pattern, whose effects, in particular the formation of recurring low/high density zones, will result in a fraction of the fluid being circulated in the close contact areas, while the majority will be inactive in the low-contact 'gap' areas.

Thus the working fluid which interacts in the 'high wear' zone will have slow cycling through the low-interaction 'storage pool', the 'actually working' pool of fluid will be much smaller and thus age faster.

Undoubtedly many more effects of non-optimal loading have been documented already, for both bearing cases.