Hard drives and shock
Per Inge Oestmoen
I wonder if my hard drives have been damaged.
Some days ago, a friend of mine happened to drop a backpack 25 cm down
at a vinyl-covered concrete floor. Unfortunately, in this backpack
were four new hard drives, three WD 400BB and one Mactor D740x. Their
only protection was the anti-static bags in which they came.
At the bottom of the backpack was a 0.5 mm plastic shield, but that
was the only thing that was between the hard drives and the hard
floor. There was a certain amount of dampening in that the backpack
hit from an angle so that the shield slided sideways in the impact,
but there is no question that the drives were subjected to at least
some shock. The weakest of these drives are certified to withstand a
shock of 200 G while non-operating. I am unsure what 200 G corresponds
to in terms of height of fall, given a concrete floor below.
Now I ask you what is the likelihood of the drives having been
damaged. That of course happens when the heads crash into the
platters, creating microscopic dust that will destroy the drives over
a period of some weeks or months of use. But could the aforementioned
shock be enough to be fatal for the drives, and is there any way to
tell?
--
Per Inge Oestmoen, Norway
http://www.coldsiberia.org/
Arno Wagner
Very hard to tell. The g-force on impact can be calculated as follows:
With s1: drop length, s2: deacceleration length,
a1, a2 accelerations (a1 is G = 9.81 m/s^2),
m: mass.
Kinetic energy formula e = f * s, with f = m * a
(f is force) gives:
s1 * a1 * m = s1 * f1 = e = s2 * f2 = s2 * a2 * m
=> s1 * a1 = s2 * a2
=> s2 = (s1 * a1)/s2
If we stipylate 1mm deaceleration lengt, this gives
s2 = (0.25 m * 9.81 m/s^2)/(0.001 m)
= 2452.5 m/s^2
Which is roughly 250G. So even with a clean bump the drives could be
damaged. Furthermore the schock-tolerance for cot clean falls
(turning, multiple impacts,...) is usually significantly lower.
However if you have, say, 3mm impact deacceleration, the numbers look
better, only 80G.
I would say the drives can realistically be damaged, but don't have to.
For your concrete example, deacceleration lengt would be more like
0.1 mm, giving 2500G, which should kill a drive usually.
Regards,
Arno
--
Replace "spam" with "wagner" and "komsys-pc-aw" with "tik.ee" for email.
GnuPG: ID:1E25338F FP:0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
"The more corrupt the state, the more numerous the laws" - Tacitus
Zvi Netiv
> I wonder if my hard drives have been damaged.
The simplest would be to connect them and test.
> Some days ago, a friend of mine happened to drop a backpack 25 cm down
> at a vinyl-covered concrete floor. Unfortunately, in this backpack
> were four new hard drives, three WD 400BB and one Mactor D740x. Their
> only protection was the anti-static bags in which they came.
>
> At the bottom of the backpack was a 0.5 mm plastic shield, but that
> was the only thing that was between the hard drives and the hard
> floor. There was a certain amount of dampening in that the backpack
> hit from an angle so that the shield slided sideways in the impact,
> but there is no question that the drives were subjected to at least
> some shock. The weakest of these drives are certified to withstand a
> shock of 200 G while non-operating. I am unsure what 200 G corresponds
> to in terms of height of fall, given a concrete floor below.
There are so many factors involved in the calculations of the impact, except
height of fall, elasticity of the colliding mass and surface, the distribution
of the impact over surface at the time of collision, the transmission of the
shock through the entire mechanism till it reaches the heads (the latter are
mounted on flexible arms that amortize the shock), etc., that there is no point
in calculating.
> Now I ask you what is the likelihood of the drives having been
> damaged. That of course happens when the heads crash into the
> platters, creating microscopic dust that will destroy the drives over
> a period of some weeks or months of use. But could the aforementioned
> shock be enough to be fatal for the drives, and is there any way to
> tell?
If the drives work, then I wouldn't be overly concerned about the microscopic
effects of the head-platter impact. Without going into relativistic theories, I
would guess that the platters and heads undergo more severe stresses during the
drive normal life, than that of a 25 cm fall onto a linoleum covered floor!
Not that I recommend dropping your drives as a daily routine! ;-)
Regards, Zvi
--
NetZ Computing Ltd. ISRAEL http://invircible.com sup...@resq.co.il
InVircible Anti-Virus Software, ResQdisk and Data Recovery Utilities
E-mail sent in reply to this post will not be considered private and
may be answered in the newsgroup.
Eric Gisin
platter, except for IBM deskstart which lift to the outside of the platter.
So the heads can't crash, but they can lift and slap back down.
Just run a burn-in test for a few hours, then the drive diagnostics.
"Per Inge Oestmoen" <pi...@coldsiberia.org> wrote in message
news:3D08652A...@coldsiberia.org...
Per Inge Oestmoen
> If the drives work, then I wouldn't be overly concerned about the microscopic
> effects of the head-platter impact. Without going into relativistic theories, I
> would guess that the platters and heads undergo more severe stresses during the
> drive normal life, than that of a 25 cm fall onto a linoleum covered floor!
Well, not if we are talking about minuscule dust particles that place
themselves between the heads and the platters.
Per Inge Oestmoen
> A properly shutdown drive will have the heads parked on inner part of the
> platter, except for IBM deskstart which lift to the outside of the platter.
> So the heads can't crash, but they can lift and slap back down.
Yes, and if they do, small amounts of dust will likely be generated
from the magnetic layer. Deformation of the heads and/or their arms
are also a hypothetical possibility.
> Just run a burn-in test for a few hours, then the drive diagnostics.
Very probably an excellent suggestion. Then how is a burn-in test
performed? What kind of software is needed? As for the diagnostics,
all the large manufacturers have drive diagnostic tools for download,
I suppose.
Rod Speed
> Eric Gisin wrote
>> A properly shutdown drive will have the heads parked on inner part of the
>> platter, except for IBM deskstart which lift to the outside of the platter.
>> So the heads can't crash, but they can lift and slap back down.
> Yes, and if they do, small amounts of dust will
> likely be generated from the magnetic layer.
Nope, not with the heads parked like that. The head wont hit the magnetic
layer. Thats the whole point of parking them like that, so they cant.
> Deformation of the heads and/or their
> arms are also a hypothetical possibility.
Yes, but thats a separate issue to particles
floating around in the sealed chamber.
>> Just run a burn-in test for a few hours, then the drive diagnostics.
> Very probably an excellent suggestion.
> Then how is a burn-in test performed?
Just thrash the drive, doing lots of work.
> What kind of software is needed?
One of the benchmarks should be fine.
> As for the diagnostics, all the large manufacturers
> have drive diagnostic tools for download, I suppose.
Yes, but thats a different issue to thrashing the drive
by using it energetically. You should certainly run the
diagnostic, but its worth thrashing the drive as well.
Rod Speed
John <j...@john.com> wrote in message
news:Xns922CBC3...@130.133.1.4...
> Eric Gisin <ericg@netidea-com> wrote
>> A properly shutdown drive will have the heads parked
>> on inner part of the platter, except for IBM deskstart
>> which lift to the outside of the platter.
>> So the heads can't crash, but they can lift and slap back down.
> I thought that parking the heads away from the used surface
> is no longer done. I thought that the heads just rest on the
> disk surface when the disk is not in use. Is this so?
Thats correct, its not so. There was a
return to the use of parking the heads.
Folkert Rienstra
"Per Inge Oestmoen" <pi...@coldsiberia.org> wrote in message news:3D08D704...@coldsiberia.org...
> Eric Gisin wrote:
>
> > A properly shutdown drive will have the heads parked on inner part of the
> > platter, except for IBM deskstart which lift to the outside of the platter.
>
> > So the heads can't crash, but they can lift and slap back down.
>
> Yes, and if they do, small amounts of dust will likely be generated
> from the magnetic layer. Deformation of the heads and/or their arms
> are also a hypothetical possibility.
Yes, who cares about the imprint on the platters themselfs and what that
can do to the heads the next time they are parked (or even on startup).
Folkert Rienstra
>
> > A properly shutdown drive will have the heads parked on inner
> > part of the platter, except for IBM deskstart which lift to the
> > outside of the platter.
> >
> > So the heads can't crash, but they can lift and slap back down.
>
>
> longer done.
It has always been done. Just differently over time.
> I thought that the heads just rest on the disk surface
> when the disk is not in use. Is this so?
Yes, in the head parking area where on some drives they will
be held in place by clamps or magnets. IBMs lately have head
parking ramps that lift the heads off the platters alltogether.
Folkert Rienstra
"Per Inge Oestmoen" <pi...@coldsiberia.org> wrote in message news:3D08D6D9...@coldsiberia.org...
> Zvi Netiv wrote:
>
> > If the drives work, then I wouldn't be overly concerned about the microscopic
> > effects of the head-platter impact. Without going into relativistic theories, I
> > would guess that the platters and heads undergo more severe stresses during the
> > drive normal life, than that of a 25 cm fall onto a linoleum covered floor!
>
> Well, not if we are talking about minuscule dust particles that place
> themselves between the heads and the platters.
Yes that is very bad.
Who cares where they came from and what that would do to
the drive, would the drive heads happen to pass those places.
Btw, ever heard of centrifugal forces?
Zvi Netiv
Wouldn't it be zeroed by the Coreolis acceleration as north at Norway? ;)
Rod Speed
> Folkert Rienstra <see_Re...@myweb.nl> wrote:
>> Yes, in the head parking area where on some drives they will
>> be held in place by clamps or magnets. IBMs lately have head
>> parking ramps that lift the heads off the platters alltogether.
> That's interesting. Do you need to do
> anything to enable the parking mechanism?
Nope, its automatic at power down.
> ISTR years ago there was a command in DOS.
An addon actually.
> IS that sort of thing necessary anymore?
Nope.
> I recently bought two hard drives to use in
> removeable trays and maybe I am not parking the
> heads after taking them away from the machine.
Its automatic as the power is removed.
Folkert Rienstra
"John" <j...@john.com> wrote in message news:Xns922D646...@130.133.1.4...
> "Folkert Rienstra" <see_Re...@myweb.nl> wrote:
>
> > Yes, in the head parking area where on some drives they will
> > be held in place by clamps or magnets. IBMs lately have head
> > parking ramps that lift the heads off the platters alltogether.
>
>
> mechanism?
No, it's automatic.
>
> ISTR years ago there was a command in DOS. IS that sort of thing
> necessary anymore?
No. I guess, that that is what the parking cylinder parameters
were for in bios setup, to tell the parking utilities to where to
steer the heads to. I don't recall any parking command in ATA.
>
> I recently bought two hard drives to use in removeable trays and
> maybe I am not parking the heads after taking them away from the
> machine.
No problem as long as you allow them to fully spindown before physically
removing them.
Mike Tomlinson
<pi...@coldsiberia.org> writes
>Well, not if we are talking about minuscule dust particles that place
>themselves between the heads and the platters.
Hard disks have filters inside to filter out microscopic particles
created when the heads land on the platter surface. The rotation of the
platters drags air through the filter, which traps any particles carried
in the airflow.
I'd do what Zvi says - connect the drives up and test them. Run a test
program which writes to every sector (manufacturer's diagnostic is
probably best.) Leave running overnight.
If any drive fails, do a warranty claim on it and don't mention you
dropped it.
--
Security-wise, NT is a server with a 'Kick me'
sign taped to it. -- Peter Gutmann
Arno Wagner
> In article <3D08D6D9...@coldsiberia.org>, Per Inge Oestmoen
> <pi...@coldsiberia.org> writes
>>Well, not if we are talking about minuscule dust particles that place
>>themselves between the heads and the platters.
> Hard disks have filters inside to filter out microscopic particles
> created when the heads land on the platter surface. The rotation of the
> platters drags air through the filter, which traps any particles carried
> in the airflow.
The two 2 year olf Fujitsu hdd's I dismantled did not have them.
Furthermore I have not heard from these filters so far.
True, there are filters, but they cover the "breater hole" that
serves to equilize air pressure in the disk and outside of it.
> I'd do what Zvi says - connect the drives up and test them. Run a test
> program which writes to every sector (manufacturer's diagnostic is
> probably best.) Leave running overnight.
Good advice. If this reports failures you know the drives where
damages. However, if it does not, you might still have significantly reduced
reliability and drive live.
Mike Tomlinson
<sp...@komsys-pc-aw.ethz.ch> writes
>The two 2 year olf Fujitsu hdd's I dismantled did not have them.
Do you have the model numbers? I have a number of dead Fujitsus - will
have a look inside them.
Most hard drives I've opened have had a microfilter. It looks like a
small wad of cotton and is aligned to sit in the flow of air created by
the spinning platters. Perhaps it's been left out in more recent models
as a cost-saving measure.
>Furthermore I have not heard from these filters so far.
>True, there are filters, but they cover the "breater hole" that
>serves to equilize air pressure in the disk and outside of it.
Yes, I know about those. Usually there's a label nearby warning you not
to cover it.
M.
Rod Speed
Mike Tomlinson <mi...@NOSPAM.jasper.org.uk> wrote
in message news:1eaezdAN...@jasper.org.uk...
> Arno Wagner <sp...@komsys-pc-aw.ethz.ch> writes
>> The two 2 year olf Fujitsu hdd's I dismantled did not have them.
> Do you have the model numbers? I have a number
> of dead Fujitsus - will have a look inside them.
> Most hard drives I've opened have had a microfilter. It looks
> like a small wad of cotton and is aligned to sit in the flow of
> air created by the spinning platters. Perhaps it's been left
> out in more recent models as a cost-saving measure.
Or more likely because its a pretty useless approach. If you've got
loose particles of media floating around in the chamber after a head
crash, you've got buckleys of it all being collected by any filter like that
and none getting between the head and the media before the filter.
That approach wasnt even viable with low tech drives and hasnt
got a hope in hell of working with modern drives where the heads
fly very close to the surface to get the modern data densitys.
Its very unlikely that modern heads would be in very good condition
after the initial head crash anyway, quite apart from what happens
to any fragments of the media after the initial head crash.
Arno Wagner
> <sp...@komsys-pc-aw.ethz.ch> writes
>>The two 2 year olf Fujitsu hdd's I dismantled did not have them.
> Do you have the model numbers? I have a number of dead Fujitsus - will
> have a look inside them.
Sorry, I threw them away after opening them and I do not remember.
I fried both in one go with a misconnected Y-cable.
> Most hard drives I've opened have had a microfilter. It looks like a
> small wad of cotton and is aligned to sit in the flow of air created by
> the spinning platters. Perhaps it's been left out in more recent models
> as a cost-saving measure.
Could be.
Per Inge Oestmoen
> Previously Mike Tomlinson <mi...@nospam.jasper.org.uk> wrote:
> > I'd do what Zvi says - connect the drives up and test them. Run a test
> > program which writes to every sector (manufacturer's diagnostic is
> > probably best.) Leave running overnight.
> Good advice. If this reports failures you know the drives where
> damages. However, if it does not, you might still have significantly reduced
> reliability and drive live.
I will certainly follow that advice, but I wonder what basis there is
for assuming that also in the absence of demonstrable damage the
drives might still have reduced reliability and shortened life.
If the shock has not reached above a certain threshold where the heads
start to hammer the platters' surface, one might infer that no
permanent deformation or damage has taken place either since the arms
are unlikely to be deformed unless they have made such a movement as
to slam into the disk surface with violent results.
Or am I wrong?
Arno Wagner
> Or am I wrong?
I think so. There is a range of impact strength, where the heads get
somwhat damaged, but continue to function. In that case the
heads may, e.g. fly unevenly, and have a higher chance of sustaining
further damage.
It is really difficult to say. I would think that either no damage
or noticable damage is the usual outcome from shok, but I really
do not know.
Per Inge Oestmoen
> I think so. There is a range of impact strength, where the heads get
> somwhat damaged, but continue to function. In that case the
> heads may, e.g. fly unevenly, and have a higher chance of sustaining
> further damage.
With today's high-precision instruments, one would assume that any
misalignment would result in detectable malfunction. Which means that
any damage and the resultant lack of precision will be easily detected
by a diagnostics program.
> It is really difficult to say. I would think that either no damage
> or noticable damage is the usual outcome from shok, but I really
> do not know.
Are there some who know more?
Folkert Rienstra
"Arno Wagner" <sp...@komsys-pc-aw.ethz.ch> wrote in message news:aei093$6ond2$1...@ID-2964.news.dfncis.de...
> Previously Mike Tomlinson <mi...@nospam.jasper.org.uk> wrote:
> > In article <3D08D6D9...@coldsiberia.org>, Per Inge Oestmoen <pi...@coldsiberia.org> writes
>
> >>Well, not if we are talking about minuscule dust particles that place
> >>themselves between the heads and the platters.
>
> > Hard disks have filters inside to filter out microscopic particles
> > created when the heads land on the platter surface. The rotation of the
> > platters drags air through the filter, which traps any particles carried
> > in the airflow.
>
> The two 2 year olf Fujitsu hdd's I dismantled did not have them.
A Micropolis Tomahawk and an IBM DCHS both have them.
Nomai/SyQuest cartridges also have them.
> Furthermore I have not heard from these filters so far.
> True, there are filters, but they cover the "breater hole" that
> serves to equilize air pressure in the disk and outside of it.
>
> > I'd do what Zvi says - connect the drives up and test them. Run a test
> > program which writes to every sector (manufacturer's diagnostic is
> > probably best.) Leave running overnight.
>
> Good advice. If this reports failures you know the drives where
> damaged.
> However, if it does not, you might still have significantly reduced
> reliability and drive live.
Right. The drive's error recovery process may conceal any problems
that are there and may get worse later on. Slight damage to spindle
bearings may get worse with time. Platter excentricity may wear out
the bearings over time. A harder working drive may get hotter than it
would if it were working under normal conditions. It takes a lot more
than a (relatively) simple software test to find those type of problems.
Folkert Rienstra
"Per Inge Oestmoen" <pi...@coldsiberia.org> wrote in message news:3D0F448A...@coldsiberia.org...
> Arno Wagner wrote:
>
> > Previously Mike Tomlinson <mi...@nospam.jasper.org.uk> wrote:
> > > I'd do what Zvi says - connect the drives up and test them. Run a test
> > > program which writes to every sector (manufacturer's diagnostic is
> > > probably best.) Leave running overnight.
>
> > Good advice. If this reports failures you know the drives where
> > reliability and drive live.
>
> I will certainly follow that advice, but I wonder what basis there is
> for assuming that also in the absence of demonstrable damage the
> drives might still have reduced reliability and shortened life.
>
> If the shock has not reached above a certain threshold where the heads
> start to hammer the platters' surface, one might infer that no
> permanent deformation or damage has taken place either since the arms
> are unlikely to be deformed unless they have made such a movement as
> to slam into the disk surface with violent results.
>
> Or am I wrong?
Well, that depends on what you call demonstrable, doesn't it.
Folkert Rienstra
"Per Inge Oestmoen" <pi...@coldsiberia.org> wrote in message news:3D0F6597...@coldsiberia.org...
> Arno Wagner wrote:
>
> > I think so. There is a range of impact strength, where the heads get
> > heads may, e.g. fly unevenly, and have a higher chance of sustaining
> > further damage.
>
> With today's high-precision instruments,
Oh, where can I get those?
> one would assume that any misalignment would result in detectable
> malfunction. Which means that any damage and the resultant lack
> of precision will be easily detected by a diagnostics program.
Not the usual diagnostic program. There are no provisions in the ATA
specs to allow for that in standard ATA. However, S.M.A.R.T. may pass
drive internal diagnostic info to the outside world. Like flyhight and shock.
Per Inge Oestmoen
> "Arno Wagner" <sp...@komsys-pc-aw.ethz.ch> wrote in message news:aei093$6ond2$1...@ID-2964.news.dfncis.de...
> > However, if it does not, you might still have significantly reduced
> > reliability and drive live.
> Right. The drive's error recovery process may conceal any problems
> that are there and may get worse later on. Slight damage to spindle
> bearings may get worse with time. Platter excentricity may wear out
> the bearings over time. A harder working drive may get hotter than it
> would if it were working under normal conditions. It takes a lot more
> than a (relatively) simple software test to find those type of problems.
Then the question remains: what is the likelihood of this scenario? If
it is probable that a shock that does not cause immediately detectable
damage has still left the drives with reduced life span and less
reliability, then drives that has been subjected to any shock just
have to be thrown even if they happen to function flawlessly - if you
are unwilling to take risks.
Do you know positively that relatively minor shocks can lead to
deformation of spindle bearings, arms and other parts even if
demonstrable misalignment or malfunction is not detected and the shock
did not kill the disk(s) in question?