Hard drive positioning issues?

[RebateMonger]

Drives are designed with servo data on the disks that are a position error signal. This error signal can account for a certain amount of NRRO (Non Repeatable Run Out), but over the course of time the spindle may be bent by a fraction of an inch, and for a HDD this can be much more dramatic.

I don't know exactly how disk drive bearings are constructed, so my knowledge of this topic is limited. I can see how the spindle could be effectively "bent" (maybe more of a bearing issue than a spindle issue).

But it seems like the point where the heads read the disk would remain at pretty much the same location in 3D space as the disk spins. Yeah, the disk surface would effectlvely be slowly moving "downwards" over time, but if the electronics, head assembly, and voice coil can compensate for this long-term effect, then the drive should be able to work reliably. Now if the stress of vertical orientation introduces "wobble" in the bearing (besides the "bending"), then that could certainly be a problem.

 

[Rubycon]

I hate vertical externals like WD has. Too easy to tip while running. *tip* FUUUUUUUUUUUUUUUUUUUUUUUUUUUUU

Mybooks should be placed on their side and strapped down.

 

[taltamir]

I was poking fun at WD, not you. I've got a WD external drive that came with a normal desktop drive in it.

And it stands on it's short side.

Which I believe is part of the reason external drives are so prone to failure. Which means they get to sell you a new drive to replace it later on.

I count 12 edges.

Also 8 sides if you also count the in side and the out side.

heh, they probably should have phrased it better.

 

[Tsavo]

Mybooks should be placed on their side and strapped down.

Strapped down?

Wait, I forgot you live on a giant speedboat. :whiste:

 

[C1]

BTW, if you want some bent spindles then the massive 5 platter 2TB 7200 rpm Hitachi HDDs are the way to fly. The platter assembly is so massive in those drives that you can really really feel the gyroscopic action of the unit when hand held and attempting to rotate it in the plane of the spindle axis.

 

[bryanl]

It's enough. Trust me.

I do data recovery, I see drives that have bent spindles from this every month. This isn't a "if" it is a fact. You can punch as many numbers in as you want it doesn't change the fact that drives fail when they are put in this orientation because of bent spindles.

Trust me, I wish it would go away Platter Transplants are not really what I like to do all day.

Why doesn't vertical operation damage gas turbines, which spin much faster and with much more inertia than hard drive platters do?

 

[AstroGuardian]

If the drive has some age on it, it's best to keep it in the same orientation as it's lived its life.
Jostling the HD around can possibly knock loose dust from the internal filter.
Loose dust in the HD enclosure can cause platter damage.

But for a newish HD, mount it any way you want.

This makes a hell of a lot sense to me. But, aren't hard drives assembled in zero-dust chambers? Or do you think that debris form inside the drive as it works?

 

[AstroGuardian]

OK, i tried mounting an older drive vertically and guess what... It vibrates and makes noise like a chopper. When i turn it horizontally it settles and works quietly. But it's a fairly old Hitachi 80Gb sata drive. I didn't plan on using it at all but still. I saw a white paper from Western Digital stating that drives were designed to work horizontally and vertically but not diagonally. It didn't say anything bad about mounting the drive vertically though...

But this old drive gives me the creeps. The whole table is vibrating and the spoon inside the coffee mug is shivering. This is a proof that older drives who lived most of theis lives horizontal should not be placed in other manner...

 

[Blain]

This makes a hell of a lot sense to me. But, aren't hard drives assembled in zero-dust chambers? Or do you think that debris form inside the drive as it works?

10-4 good buddy!

 

[bryanl]

If the drive has some age on it, it's best to keep it in the same orientation as it's lived its life.
Jostling the HD around can possibly knock loose dust from the internal filter.
Loose dust in the HD enclosure can cause platter damage.

But for a newish HD, mount it any way you want.

This makes a hell of a lot sense to me. But, aren't hard drives assembled in zero-dust chambers? Or do you think that debris form inside the drive as it works?

Dust forms inside the drives when the heads lose altitude during seeks and even tilt slightly, causing their head assemblies to momentarily scrap the platters and kick up metallic dust. That's why hard drives contain internal dust filters in addition to the breather filter:

Most drive filters, unlike the one in the above drawing, don't really let air pass through them easily but are merely bags to trap particles. Inside each bag is dissicent to remove and absorb moisture.

10 years ago, Maxtor introduced a line of video media drives designed to reduce dust-related failures by having a head only on the bottom bottom side of the platter. Unfortunately those drives were unreliable for other reasons.

 

[Throckmorton]

Scraping platters? Wouldn't that destroy your data anyway?

 

[Blain]

If heads are scraping the platters, the drive is on its way out.
A cushion of air (air bearing) keeps the heads from contact with the platters.

 

[taltamir]

Scraping platters? Wouldn't that destroy your data anyway?

it does, bad sectors are most commonly caused by this, or by dust particle created by this being trapped between the platter and read head and gouging a a scratch in it. The more dust there is, the more new dust is created via scraping the disk.

Drives all come with spare area, when they detect bad sectors they remap them to the spare area, so you need to lose more then a certain amount of sectors before you actually "see" bad sectors. If you see them, it means you have to many that you ran out of spare area.

That is why you should not move a HDD While it is working, since you can cause the heads to impact the platters.

 

[Blain]

Can you remap a HD to select out slow responding sectors?

 

[taltamir]

Can you remap a HD to select out slow responding sectors?

The only way a sector is "slow responding" is if it fails to read the first time. If a sector makes a habit of failing to read it will be marked as bad automatically. I know of no tool that lets you manually mark sectors as bad.

 

[C1]

I know of no tool that lets you manually mark sectors as bad.

SEATOOLS, after HDD testing, will ask if you want to repair (ie, do a remap) of any bad sectors which were determined to have been found.

 

[taltamir]

SEATOOLS, after HDD testing, will ask if you want to repair (ie, do a remap) of any bad sectors which were determined to have been found.

which is completely different than manually marking a sector of your choice as bad.

Anyways, it would certainly be odd if it behaves as you said, because it serves no purpose for it to ask if you want to mark as bad sectors that it has found to be bad. If a sector is bad it should be remapped.

 

[bryanl]

Scraping platters? Wouldn't that destroy your data anyway?

Not if they're scraped *gently* enough to not puncture the overcoat.

The platters of any drive that had been in service for several years will show faint tangential scratches all over it because the trailing edges of the head assemblies usually touch down momentarily on every seek. This is why platters have a protective coating (I believe only carbon is still used, but Teflon was another coating) with lubricant over it.

 

[VirtualLarry]

Not if they're scraped *gently* enough to not puncture the overcoat.

The platters of any drive that had been in service for several years will show faint tangential scratches all over it because the trailing edges of the head assemblies usually touch down momentarily on every seek. This is why platters have a protective coating (I believe only carbon is still used, but Teflon was another coating) with lubricant over it.

Those are "tail-dragging sliders". Unsure if those kind are still in use. Perhaps Mark R or Russ could elaborate.

 

[Blain]

SEATOOLS, after HDD testing, will ask if you want to repair (ie, do a remap) of any bad sectors which were determined to have been found.

which is completely different than manually marking a sector of your choice as bad.

Anyways, it would certainly be odd if it behaves as you said, because it serves no purpose for it to ask if you want to mark as bad sectors that it has found to be bad. If a sector is bad it should be remapped.

True, I wasn't asking about "bad" sectors, they're always detected.
I was asking about remapping to weed out slow responding sectors, say above 250ms.

 

[taltamir]

a 7200 RPM drive takes 8.3ms for a full rotation.
the only way a sector will take an extra long time to read is if it fails to read the first time.
If it does fail to read the first time, then it will try again and again until giving up on the data it contained. 250ms / 8.3 = 30. The first one might take slightly longer due to the need to position the head first. But the bottom line is that if a sector failed to read 30 times in a row before succeeding, then it will probably be marked as bad anyways.

If you are looking to figure out why you have occasional hiccups where the drive takes a while to get something, then my guess would be that either you have a green drive (goes to sleep mode quickly), or you have OS power saving setting set to put the drive into sleep mode (AFAIK its after 20 minutes of inactivity on windows by default)

 

[Blain]

If you are looking to figure out why you have occasional hiccups where the drive takes a while to get something, then my guess would be that either you have a green drive (goes to sleep mode quickly), or you have OS power saving setting set to put the drive into sleep mode (AFAIK its after 20 minutes of inactivity on windows by default)

Neither.
Occasionally I run across HDs that take a long time to boot into Windows even though they have had the clutter cleaned out, defragged and very few programs are loading at startup.

 

[taltamir]

Neither.
Occasionally I run across HDs that take a long time to boot into Windows even though there are very few programs loading at startup.

who said its the hard drive's fault? and that it is caused by "slow to read sectors"? Because I seriously doubt its the HDD.
There could be an at boot program running (common with certain windows updates) with a lot of random writes (which are very slow on HDDs) or any number of issues that cause such an at boot slowdown.

 

[Blain]

Nobody is blaming the HDs. I'm simply pondering a possible hardware related component in the slow boot situation.

 

[bryanl]

But the bottom line is that if a sector failed to read 30 times in a row before succeeding, then it will probably be marked as bad anyways.

What about 18 times (150ms)? In my experience, the 1TB and larger S-brands always have a few sectors needing at least 18 spins to read, yet they're never marked bad, while the W&H brands never have sectors that slow (or the were marked off during manufacture). This is with MHDD, a DOS diagnostic.