Shock rating specs for HDDs

[Franc Zabkar]

The high shock ratings for HDDs have always puzzled me, but recently I
got to thinking about them in another way. Specifically, I asked
myself the question, how much shock does a HDD sustain when it is
dropped from a particular height?

Let's say that the HDD is dropped from a height H and hits the ground
with a velocity V.

We have ...

V^2 = 2g.H

Let's now say that the HDD is decelerated to rest during a very short
impulse of duration T seconds, and let's express this deceleration (S)
as a multiple of g, where g is the acceleration due to gravity (9.8
m/s^2).

So ...

S = (dv/dt) / g
= (final velocity - initial velocity)/(duration of impulse)/g

= - (V/T) / g

= - sqrt(2g.H) / T / g
= - sqrt(2H/g) / T

AIUI, most shock ratings assume that T = 1 ms.

Therefore, if an object is dropped from 0.6m (~ 2 feet), then ...

S = 1000 x sqrt(1.2 / 9.8) = 350G

For reasons of simplicity I have assumed that the deceleration is a
constant, and that the object does not bounce.

- Franc Zabkar

 

[Arno]

The high shock ratings for HDDs have always puzzled me, but recently I
got to thinking about them in another way. Specifically, I asked
myself the question, how much shock does a HDD sustain when it is
dropped from a particular height?

A lot if the surface is hard. I think I posted some calculations
similar to yours here some years ago. Short summary, on a
hard surface, it is enough for a HDD to fall over when
placed on a small edge or to drop something like 10cm.

Arno

Let's say that the HDD is dropped from a height H and hits the ground
with a velocity V.

We have ...

V^2 = 2g.H

Let's now say that the HDD is decelerated to rest during a very short
impulse of duration T seconds, and let's express this deceleration (S)
as a multiple of g, where g is the acceleration due to gravity (9.8
m/s^2).

So ...

S = (dv/dt) / g
= (final velocity - initial velocity)/(duration of impulse)/g

= - (V/T) / g

= - sqrt(2g.H) / T / g
= - sqrt(2H/g) / T

AIUI, most shock ratings assume that T = 1 ms.

Therefore, if an object is dropped from 0.6m (~ 2 feet), then ...

S = 1000 x sqrt(1.2 / 9.8) = 350G

 

[Franc Zabkar]

A lot if the surface is hard. I think I posted some calculations
similar to yours here some years ago. Short summary, on a
hard surface, it is enough for a HDD to fall over when
placed on a small edge or to drop something like 10cm.

I found this old post of yours (June 2002):
http://groups.google.com/group/comp.sys.ibm.pc.hardware.storage/msg/c6db5684abf8d16f

You approach the problem by considering the small, but finite distance
over which the impact occurs whereas I have approached the problem
from a time perspective. In both cases we have made assumptions which
are difficult to quantify in practice.

That said, I think your example probably gives a more intuitive feel
to the shock spec than mine. It's probably easier to visualise a
physical deformation than it is to imagine an impulse.

- Franc Zabkar

 

[Arno]

On 13 Sep 2012 15:37:59 GMT, Arno <[email protected]> put finger to
keyboard and composed:

I found this old post of yours (June 2002):
http://groups.google.com/group/comp.sys.ibm.pc.hardware.storage/msg/c6db5684abf8d16f

Time flies ;-)

You approach the problem by considering the small, but finite distance
over which the impact occurs whereas I have approached the problem
from a time perspective. In both cases we have made assumptions which
are difficult to quantify in practice.

Indeed. But when viewed from the point that the hdd should
survive, it is a bit easier.

That said, I think your example probably gives a more intuitive feel
to the shock spec than mine. It's probably easier to visualise a
physical deformation than it is to imagine an impulse.

Impules is notoriously non-intuitive. But it does work.

Anyways, we both arrived at the conclusion that these
300G ratings are not nearly as good as they look, and
that is the important thing.

Fun fact: I had a HDD that I droppen from arund 80cm
onto thick carpet (at least 5mm deacceleration). That
should give no more than something like 150G, but the hdd
had massive problems afterwards and I just got the data off
before it died. (Fortunately, getting the data was just
to see whether it was possible.) The 300G figure is
too optimistic for real drops anyways.

I think "treat like raw egg" is still the best advice.
A rag egg, btw, can realistically take something like 70G,
according to a brief survey of the extensive literature ;-)

Arno