[Klug-general] eSATA

Karl Lattimer karl at nncc.info
Fri Jun 9 19:12:26 BST 2006


> 
> Like everything that particular issue can be overcome by good design.
> In some parts of the country the company I work for have shelves with 14
> SATA drives without any issues from heat.  Granted they are in well
> maintained data centres.  Those are well beyond my budget to replace NAS
> setup at home though.

external enclosures imply just that, an enclosed space where the drive
resides. drives cannot be effectively passively cooled because of how
hot they get, if you use passive cooling i.e. hot air rising draws in
air from the bottom of the chassis as in the Apple G4 cube there will be
localised points of heat which will damage the drives eventually. If you
use an active cooling system as simple as fans it will be noisy, if you
use a heat pipe between the drives you'll probably get away with it. if
you use a peltier then again could work, depending on the type of
peltier, an air circulatory peltier is impossible, but a diode type may
succeed. Basically you need a simple heat exchanger, if that is as
simple as replacing the air once every second or as complicated as a
heat pipe moving the heat elsewhere for cooling you need some cooling if
you're gonna pack them tightly. If you do find an enclosure in that case
I'd suggest ensuring it has adequate cooling.

My grandfather used to design cooling systems amongst other things for
the MoD so I know a thing or two about it. Fortunately I was privy to
some of his work after he died, suffice to say the circulatory peltier
systems that are in use in our naval vessels are very efficient at
converting tepid air into hot and cold, enough so that the hot air is
used to heat the water and the cold is used to cool their galley. 

> 
> I'd be surprised if SATA drives are any hotter than PATA or SCSI drives
> but I can't say I've ever bothered looking into it.

It isn't necessarily SATA disks that get hotter, its the disk rotation
that dissipates as heat. As the disk spins in its almost 0 resistance
baring the air around it gets hot, this is the same effect as using a
cylinder with holes in its length rotating inside another cylinder with
water, the friction of the air in the case of disks rotating against a
flat surface generates an increase in entropy which can't dissipate in
any other way. Before anybody tries to smash my statement there by
saying drives are created in a vacuum they are not, in fact hard disks
are manufactured in a class 100 clean room, at approximately 0.5atm of
pressure, the air inside is unable to expand as it gets hotter and even
if it did it has plenty space to expand, the kinetic energy generated by
spinning of oxygen/nitrogen molecules from the disk is thrown against
the metal plate at the top of the drive, thus the energy is transferred
away, the hotter the metal gets the hotter the disk will be, if you cool
the top of the drive because of the rules of entropy the disk itself
gets cooler. FYI, most PATA drives are 6.4/7.4kRPM SATA1 drives are
generally the same, SATA2 is going toward the 10kRPM speed of scsi, scsi
drives are fairly hot when running at full read/write and so are the
10kRPM SATA2 drives PATA/SATA1 get about as hot as each other.

As a side note, Western Digital drives are manufactured at around 0.7atm
and because of this get noticeably hotter, well, the last time I checked
thats what they said this was around the PATA ~20GB era so don't hold me
to that, seagate are built at around 0.3atm and have two sets of
barrings one at the top and one at the bottom (well scsi do but I'm not
sure about SATA even though I have a Seagate 300Gb SATA2), the seagate
drive i have runs fairly cool ~40*C in comparison to my old maxtor 300Gb
which kaputted a couple weeks back.

I hope this overzealous brain dump of hard drive information is helpful,
even if its just a single sentence in amongst the rest.

K,





More information about the Kent mailing list