Unusual drive failure - 12v line short

[Mike Tomlinson]

I have a PVR thingywotsit on my TV. Today it died, no power. Listen to
PSU - tick, tick, tick. Ah, thought I, bad caps.

But no, the cause was the hard disk - the 12v line was completely short.
I've not seen this mode of failure of a hard drive before now; has
anyone else?

Western Digital model WDC2500BB-00RDA0, 250GB.
date on drive: 25 Jul 07, out of warranty

Stuck in a 160GB IBM from my bits box and off we went.

It may be relevant that the drive is on 24/7. I cannot find any specs
for expected longevity or MTBF on wdc.com.

 

[vanyablue]

But no, the cause was the hard disk - the 12v line was completely short.
I've not seen this mode of failure of a hard drive before now; has
anyone else?

Yes. Unusual, but it does happen.
In this case, the extra current has just caused the PSU to trip out.
Most times, there is enough spare capacity in the PSU to vaporise the
part that's failed, and whatever device it is just power cycles and
comes back with the disk dead.
Cheers.

 

[Rod Speed]

I have a PVR thingywotsit on my TV. Today it died, no power.
Listen to PSU - tick, tick, tick. Ah, thought I, bad caps.

But no, the cause was the hard disk - the 12v line was
completely short. I've not seen this mode of failure of
a hard drive before now; has anyone else?

Yes, usually due to a capacitor across that rail as a filter shorting.

Not common, but not unheard of.

Western Digital model WDC2500BB-00RDA0, 250GB.
date on drive: 25 Jul 07, out of warranty

Stuck in a 160GB IBM from my bits box and off we went.

It may be relevant that the drive is on 24/7.

Nope. They get used like that a lot.

I cannot find any specs for expected longevity or MTBF on wdc.com.

The specs are pretty sparse on there now but they are typically 250K hours on consumer drives like that.

 

[Clint Sharp]

Maybe a capacitor across the 12V line died. RIP.

More likely to be a transorb but most likely to be the motor driver IC.

 

[Jerry Peters]

Maybe a capacitor across the 12V line died. RIP.

Had a Dell laptop with that problem; capacitor across the 19vdc power
input shorted, so capacitors do fail that way.

Jerry

 

[Franc Zabkar]

I have a PVR thingywotsit on my TV. Today it died, no power. Listen to
PSU - tick, tick, tick. Ah, thought I, bad caps.

But no, the cause was the hard disk - the 12v line was completely short.
I've not seen this mode of failure of a hard drive before now; has
anyone else?

It's common enough. There will probably be two TVS (transient voltage
suppression) diodes, one across the +5V rail, the other across the
+12V. You can remove the shorted diode and the drive should work
without it. Just make sure your power supply is good ...

- Franc Zabkar

 

[Robert Nichols]

:Most modern drives (within last 5 years or so) have a very
:long expected lifetime (MTBF = 250,000 hrs = 28 years
:running 24/7, and 500+k hours is more typical today).

The units for MTBF are not hours but device-hours, i.e., the product of
the number of hours and the number of devices being observed, and that
rating applies only during the device's rated service life, which is an
entirely separate parameter.

MTBF of 250,000 is almost totally unrelated to the expected service
life. It is quite possible to have a device with its MTBF 250,000 and a
rated service life of 1 hour. It just means that if you ran 250,000 of
those devices for one hour you should expect 1 failure. Once a device
passes its rated service life, the MTBF rating no longer applies.
Think: a battery used to provide power to a missle's guidance system --
built to be highly reliable for the short time it's needed, and pretty
much assured to go dead not long after that. High MTBF, short service
life.

Looking at the power-on hours and corresponding normalized SMART value
on a few fairly recent drives, it appears that the SMART warning due to
power-on hours would come at about 10 years of power on.

 

[Arno]

On Sun, 7 Jun 2009 16:17:44 +0100, Mike Tomlinson

Most modern drives (within last 5 years or so) have a very
long expected lifetime (MTBF = 250,000 hrs = 28 years
running 24/7, and 500+k hours is more typical today).

The MTFB is completely untelated to the device lifetime.
It just describes the failure probablility during the device
lifetime. Device lifetime is stated in the device datasheet
and typically 5 years.

For example, an MTBF of 250'000h gives you a failure
probability of 365*24/250'000 = 3.5%/year, in the first
5 years. It dioes not make any statement about the failure
pobability afterwards.

Arno

 

[Mike Tomlinson]

It's common enough. There will probably be two TVS (transient voltage
suppression) diodes, one across the +5V rail, the other across the
+12V.

You're right. I took the board off and there is a SMT diode on the
hidden side across the 12v line, and it is reading a short. I suppose
it could be the motor controller chip that's shorted, but it looks
intact.

You can remove the shorted diode and the drive should work
without it. Just make sure your power supply is good ...

I'll take the board into work, remove the diode today and report back.
Thanks. 250gb is still a useful capacity to have. The drive's in a
personal video recorder, not a PC, and this PVR is known to have a weak
PSU.

 

[Bob Willard]

:Most modern drives (within last 5 years or so) have a very
:long expected lifetime (MTBF = 250,000 hrs = 28 years
:running 24/7, and 500+k hours is more typical today).

The units for MTBF are not hours but device-hours, i.e., the product of
the number of hours and the number of devices being observed, and that
rating applies only during the device's rated service life, which is an
entirely separate parameter.

MTBF of 250,000 is almost totally unrelated to the expected service
life. It is quite possible to have a device with its MTBF 250,000 and a
rated service life of 1 hour. It just means that if you ran 250,000 of
those devices for one hour you should expect 1 failure. Once a device
passes its rated service life, the MTBF rating no longer applies.
Think: a battery used to provide power to a missle's guidance system --
built to be highly reliable for the short time it's needed, and pretty
much assured to go dead not long after that. High MTBF, short service
life.

Looking at the power-on hours and corresponding normalized SMART value
on a few fairly recent drives, it appears that the SMART warning due to
power-on hours would come at about 10 years of power on.

That is total BS. MTBF is per device. Read, for example, MIL-HDBK-217.

 

[Mike Tomlinson]

More likely to be a transorb but most likely to be the motor driver IC.

They usually burn up when they fail, and this one looked fine. Took out
the shorted diode and the drive span up ok. Not yet tried it.

 

[Mike Tomlinson]

I'll take the board into work, remove the diode today and report back.

Took out the shorted diode today and the drive span up OK on a PC power
supply. Not yet tried it in the PVR but am sure it will be OK.

Thanks everyone for your suggestions.

 

[Robert Nichols]

:> In article <[email protected]>,
:> :
:> :Most modern drives (within last 5 years or so) have a very
:> :long expected lifetime (MTBF = 250,000 hrs = 28 years
:> :running 24/7, and 500+k hours is more typical today).
:>
:> The units for MTBF are not hours but device-hours, i.e., the product of
:> the number of hours and the number of devices being observed, and that
:> rating applies only during the device's rated service life, which is an
:> entirely separate parameter.
:>
:> MTBF of 250,000 is almost totally unrelated to the expected service
:> life. It is quite possible to have a device with its MTBF 250,000 and a
:> rated service life of 1 hour. It just means that if you ran 250,000 of
:> those devices for one hour you should expect 1 failure. Once a device
:> passes its rated service life, the MTBF rating no longer applies.
:> Think: a battery used to provide power to a missle's guidance system --
:> built to be highly reliable for the short time it's needed, and pretty
:> much assured to go dead not long after that. High MTBF, short service
:> life.
:>
:> Looking at the power-on hours and corresponding normalized SMART value
:> on a few fairly recent drives, it appears that the SMART warning due to
:> power-on hours would come at about 10 years of power on.
:>
:
:That is total BS. MTBF is per device. Read, for example, MIL-HDBK-217.

In a brief perusal of that document (scanned copy -- no search function
available) I do not find the term "MTBF" mentioned anywhere, so I fail
to see why you would consider that document the definitive work on the
subject.

OTOH, I spent 26 years designing high-reliability electrical and
electronic equipment and do have some idea of what I am talking about.

 

[Arno]

:> In article <[email protected]>,
:> :
:> :Most modern drives (within last 5 years or so) have a very
:> :long expected lifetime (MTBF = 250,000 hrs = 28 years
:> :running 24/7, and 500+k hours is more typical today).
:>
:> The units for MTBF are not hours but device-hours, i.e., the product of
:> the number of hours and the number of devices being observed, and that
:> rating applies only during the device's rated service life, which is an
:> entirely separate parameter.
:>
:> MTBF of 250,000 is almost totally unrelated to the expected service
:> life. It is quite possible to have a device with its MTBF 250,000 and a
:> rated service life of 1 hour. It just means that if you ran 250,000 of
:> those devices for one hour you should expect 1 failure. Once a device
:> passes its rated service life, the MTBF rating no longer applies.
:> Think: a battery used to provide power to a missle's guidance system --
:> built to be highly reliable for the short time it's needed, and pretty
:> much assured to go dead not long after that. High MTBF, short service
:> life.
:>
:> Looking at the power-on hours and corresponding normalized SMART value
:> on a few fairly recent drives, it appears that the SMART warning due to
:> power-on hours would come at about 10 years of power on.
:>
:
:That is total BS. MTBF is per device. Read, for example, MIL-HDBK-217.

In a brief perusal of that document (scanned copy -- no search function
available) I do not find the term "MTBF" mentioned anywhere, so I fail
to see why you would consider that document the definitive work on the
subject.

OTOH, I spent 26 years designing high-reliability electrical and
electronic equipment and do have some idea of what I am talking about.

No need to sling documents around. Even the name says it is
a failure probability per time, not a life time. Probabilities
are not individual counters and so a "per device" does not apply,
given statistical independence. This also means that the probability
of a device failing is not dependent on the non-failing time it
had before. (Here the "component life" comes in. It limits that
independence to a maximum non-failing operating time and says the
MTBF may become invalid what that is exceeded.)

When you actually want to measure MTBF, you run a number of devices
for a time until all have failed or the component life has
been exceeded (the latter usually done by statistical models
and/or accellerated ageing), count all non-failing hour you got
and divide them by the device number.

Now, this may be the wrong approach. It is quite possible that
device failure probability is not independent on the previous
non-failing operating time. In that case, one MTBF would not be the
right measure. Several different MTBFs could be given for
different periods of a devices lifetime so far as an approximation.
But this is not how it is done at this time. At this time
you get a failure probability per operating hour and the
only history of the device that goes into it is whether it
has already failed (and then all bets are off, it is assumed
to stay failed by the model) or whether it has exceeded its
component life time and the MTBF stated simply does
not apply anymore and is not replaced by any other value
instead.

Arno

 

[Eric Gisin]

Took out the shorted diode today and the drive span up OK on a PC power
supply. Not yet tried it in the PVR but am sure it will be OK.

Thanks everyone for your suggestions.

PVR may have a crappy PSU with poor 12V regulation.

 

[Barry OGrady]

I had a CPM computer whose 10MB HD would not spin up.
Fortunately the power supply had LEDs to show power and
the 12v LED was off. It turned out to be a shorted capacitor.

Barry
=====
Home page
http://members.iinet.net.au/~barry.og