Intel X25-M SSD -- no protection from overvoltages?

[Franc Zabkar]

I seem to be answering about three questions per week, in several
storage forums, in regard to damage sustained by hard drives after
they have experienced an overvoltage. Sometimes an internal PC PSU is
to blame, other times it is the result of a miswired USB-SATA/IDE
adapter, but most often it is the result of using the wrong AC adapter
(eg a 19V laptop adapter) to power an external drive. In most cases
the damage is restricted to a shorted TVS (transient suppression
diode). The solution is simply to cut it out with a pair of flush
cutters.

Here are some examples:
http://www.users.on.net/~fzabkar/HDD/

By way of comparison, here is a recent thread involving an Intel SSD:
http://forum.hddguru.com/intel-x25-ssd-160-burned-t18945.html

Strangely, this expensive device has no overvoltage protection.
Furthermore, the onboard DC-DC converter ICs have maximum input
voltage ratings of 5.5V and 6V. This makes me wonder just how many
SSDs are failing due to power supply issues. BTW, the aforementioned
ICs are available for $1 and $4 in single quantities from places like
Mouser and Digikey, as are the TVS diodes. I wonder how many SSD "data
recoveries" (costing $3K+) are nothing more than simple chip swaps
that your local TV repair shop could handle?

- Franc Zabkar

 

[Arno]

I seem to be answering about three questions per week, in several
storage forums, in regard to damage sustained by hard drives after
they have experienced an overvoltage.

Simple: If it still works, don't trust it anymore, it can break
completely without warning. If it does not work, look for
a fuse and transil overvoltage supressor. If one is there,
replace the fuse, pull off your data and do not trust the
disk anymore. If there is just an unfused transil,
remove that. The disk can still have suffered major
damage and break at any time without warning.

Sometimes an internal PC PSU is
to blame, other times it is the result of a miswired USB-SATA/IDE
adapter, but most often it is the result of using the wrong AC adapter
(eg a 19V laptop adapter) to power an external drive.

Hehe. Being able to read and having not slept through shool
physics will help here.

In most cases
the damage is restricted to a shorted TVS (transient suppression
diode). The solution is simply to cut it out with a pair of flush
cutters.

Not a permanent repair. These suppressors are typically sized so
that the overvoltage is somewhere above the absolute maximum
ratings of the rest of the circuit, but typically low enough
that it still works. For a time. Also, using a TVS without
fuse or 0-Ohm resistor as circuit breaker is not professional.
In case there is a fuse or 0-Ohm resistor, you typically
have to replace that, and the suppressor diode should be intact.

Here are some examples:
http://www.users.on.net/~fzabkar/HDD/

By way of comparison, here is a recent thread involving an Intel SSD:
http://forum.hddguru.com/intel-x25-ssd-160-burned-t18945.html

Strangely, this expensive device has no overvoltage protection.

Well, overvoltage protection is somewhat optional in this case.
Any good PSU already has effective overvoltage protection.
If you combine an ElCheapo PSU and an expensive SSD, you sort-of
deserve what you get. I think Intel's reasoning is sound.

Furthermore, the onboard DC-DC converter ICs have maximum input
voltage ratings of 5.5V and 6V. This makes me wonder just how many
SSDs are failing due to power supply issues.

Probably not a lot, see above.

BTW, the aforementioned
ICs are available for $1 and $4 in single quantities from places like
Mouser and Digikey, as are the TVS diodes. I wonder how many SSD "data
recoveries" (costing $3K+) are nothing more than simple chip swaps
that your local TV repair shop could handle?

Most data recovery is actually software-only recovery. Most people
cannot tell the difference between a broken SSD and one where
the filesystem is corrupted.

Side note: Effective overvoltage protection is a bit more
difficult than adding a transil. For 5V circuitry, the very
least is a trisil and a fast fuse. The transil typically
goes way above the 7V maximum most 5V circuits can stand.
But the trisil will short out the 5V line even on short power
spikes and likely generate a lot more service calls, so
manufacturers put in nothing or a transil, sometimes without
fuse.

And second side-note: Overvoltage protection is the task of
the PSU and not the disk or other device. My guess is the
many, many badly designed or manufactured PSUs out there
are the reason for some manufactueres including protection
in the drives. But the solution is to get a good PSU, not
to do anything on the disk side.

Arno

 

[Franc Zabkar]

Well, overvoltage protection is somewhat optional in this case.
Any good PSU already has effective overvoltage protection.
If you combine an ElCheapo PSU and an expensive SSD, you sort-of
deserve what you get. I think Intel's reasoning is sound.


Probably not a lot, see above.

As I said in my prologue, I'm anwering several questions related to
overvoltages every week. I can't see why SSDs would be any less prone
to damage than HDDs, especially once they become prevalent in external
storage.

TVS diodes have been standard equipment in HDDs for the past decade. I
see no reason why they should not be used in SSDs. Furthermore, I see
no reason why the designer could not specify a PWM controller with a
voltage input range that could at least tolerate a 12V supply. Other
consumer equipment such as TVs and DVDs have PWM controllers with OVLO
and UVLO (over/undervoltage lockout) features. This means that the
oscillator is disabled whenever the input voltage is either too high
or too low.

In fact here is one such IC:
http://www.onsemi.com/pub/Collateral/NCP1212-D.PDF

BTW, I expect than none of the logic on the SSD would see the +5V
supply directly. Instead I expect that they would be powered from the
Vcore and Vio DC-DC converters.

- Franc Zabkar

 

[Arno]

On 11 Apr 2011 14:41:13 GMT, Arno <[email protected]> put finger to
keyboard and composed:

As I said in my prologue, I'm anwering several questions related to
overvoltages every week. I can't see why SSDs would be any less prone
to damage than HDDs, especially once they become prevalent in external
storage.

If you put an expensive SSD into a cheap enclosure, then you get
what you deserve. With an expensive enclosure, the enclosure will
have overvoltage protection. The problem is that the disk is the
wrong place to do this. In any sane design, overvoltage protection
is in the power circuitry, not the storage device. Of course, there
will allways be people that do not get it an buy cheap PSUs. For
those the Transils will at least give data revovery a chance.

Side note: What happens to the disks that, following your
advice, do not have a Transils anymore? Do you at least
instruct these people to replace them and get rid of their
PSU and get something with reasonable protection circuitry?

TVS diodes have been standard equipment in HDDs for the past decade. I
see no reason why they should not be used in SSDs. Furthermore, I see
no reason why the designer could not specify a PWM controller with a
voltage input range that could at least tolerate a 12V supply. Other
consumer equipment such as TVs and DVDs have PWM controllers with OVLO
and UVLO (over/undervoltage lockout) features. This means that the
oscillator is disabled whenever the input voltage is either too high
or too low.

In fact here is one such IC:
http://www.onsemi.com/pub/Collateral/NCP1212-D.PDF

BTW, I expect than none of the logic on the SSD would see the +5V
supply directly. Instead I expect that they would be powered from the
Vcore and Vio DC-DC converters.

I think your expectations are unrealistic. True, you could insulate
the electronics completely, at least with an sata interface. But
then you need a larger converter which is more expensive and the
consumers are not willing to pay for that.

A good quality 6W DC-DC converter with wide range input
runs something like $40 (more expensive for single units).
It is alos rather large and does not fit a normal SSD.

Arno