(PeteCresswell) wrote:
> I just added a new drive cage to a WSH box I am using to back up my NAS
> box. Total drives to be 13 (a system drive plus 12 data drives)
>
> It only has to run intermittently - when I do my weekly backups of media
> files.
>
> Current PS is rated for 580w and, steady-state the box is pulling high
> fours, maybe into the low fives.
>
> But at startup, there is a surge that goes well into the high six
> hundreds. Coincidentally, I've been having some intermittent problems
> with the six new drives that are hanging on a couple of RAID cards.
>
> The surge lasts less than 20 seconds.
>
> Could that be a problem? Might I be setting myself up for a string of
> problems in the future?
>
> Or is it OK as long as the PS doesn't just stop working?
You could look for a means to implement "staggered spinup".
We used to be able to do that on SCSI, but I don't know
how that translates on SATA.
If you have Hot Plug capability in the driver (AHCI), you
could always fit a power switch on each hard drive. Leave
the ground path intact. Interrupt +5V and +12V with the switch.
Switch on one drive at a time, or switch on small groups of
drives.
During spinup, drives can use as much as 12V @ 3A for the first
ten seconds. If you manage to find the documentation for your
drive, that info will be available. Many years ago, the current
flow level might have been 12V @ 2A, but it's actually heading
in the wrong direction.
After spinup completes, there are only frictional losses as
the heads fly over the platter. My 2TB drive, the 12V drops to
0.51A. My 120GB IDE drive (old news), the 12V right now is 0.45A.
(Those are measured values with my clamp-on DC ammeter.)
So the steady state value tends to be less of an issue for the
average home setup.
But if you're spinning up 10 drives at 3 amps each, that's
30 amps load from 12V for the first ten seconds. For
the first 0.35 sec, the power supply does not police the
current limit. For the other 9.65 seconds, the power supply
will monitor 12V flow, and do something if you draw too much.
The trip level should be set around 30% higher than the value
on the PSU label. The supply will become "soft" near that
limit, so the voltage might not be exactly 12V as the overcurrent
trips.
If the 12V rail drops to 11V, a drive will actually spin down.
So if there is enough of a drop, the drive might take action
itself.
Don't be overly pessimistic in your power calculations.
A lot of stuff on the computer, doesn't run at full power
during every phase of operation. Even my dual core "65W" CPU,
only draws 36W flat out running Prime95 (two threads). And it
draws around 13.2W when idle. Depending on your CPU, the consumption
can be quite thrifty. You would need a monster, ancient video card,
to be hitting 400W at idle. My current processor uses a bit more,
at 43W running Prime95, out of a TDP rating of 65W. Early S478
processors, their flat out numbers were very close or a watt or
two higher than the listed TDP value.
So, yes, I'd be concerned. 13 drives @ 3 amps is 39 amps for the
first ten seconds. And you should see if there is some way to
stagger that. In a storage server, the rest of the 12V consumption
might be less of an issue (you're not likely to be running a pair
of 8800GTX in there).
The disks don't mind what you're doing, until the yellow wire
drops below 11V. I think the 5V rail is monitored as well, but
it doesn't have that big surge on it. Only the motor supply
(+12V yellow wire) has the big current draw.
Paul
|
Similar Threads
Linear Mode
