Electrical Engineering - Decouple fans and hard disks supplies

Author

Decouple fans and hard disks supplies

mike@spamforbidden.com

2006-06-15, 5:21 am

Hello respected engineers!
I've built myself a frame where I've put six hard disks for a RAID5
system, and six 8x8 cm fans to cool them down (they live in a quite
hot environment!). The PSU is a 500W one however I was concerned that
the fans may pollute the hard disks' supply too much.. since they're
motors after all!
I've accidentally put a finger into a fan (what a pain!), and this
stopped the fan for a moment and also caused hard disk errors, and
thus convinced me even more that I need to decouple the fans' motors
supply from the hard disks supply.

But how? I'm just a hobbyst when it comes to electronics, so I may
even intuite correctly, but I won't be really sure of my thoughts.
Right now, they are that I need some big inductors (with as low DC
resistance as possible), some (big?) capacitors and maybe also a big
rectifier as well.

The best scheme I've thought so far is to link all 6 fans together in
parallel, then apply power through a circuit like this:

6 fans--*-ind-*-ind-*-ind-*-ind-*-ind-*-ind---diode------*-----PSU +12V
| | | | | | |
cap cap cap cap cap cap 6 HDs---PSU +5V
| | | | | | |
Gnd ----*-----*-----*-----*-----*-----*------------------*-----Gnd

of course in the above scheme I don't even need them to be six,
it's simply the more the better, I just happened to have six of
them on my bench (more or less a coincidence).

But now the big question: what's the right size of the caps, also
to not overload the PSU at start? Will the diode really help the
hard disks to not get some bad bad bad back EMF?
Notice also that the wire that feeds the fans is AFTER all the wires
that feed the hard disks (otherwise the voltage drop caused by the
fans would have affected the hard disks). Yes, at least I know this. ;)

Is my scheme ok? What can I do to improve it? I really care about
the hard disks and the data they contain!

Thank you very much,
Mike

Palindr☻me

2006-06-15, 6:21 am

mike@spamforbidden.com wrote:
> Hello respected engineers!
> I've built myself a frame where I've put six hard disks for a RAID5
> system, and six 8x8 cm fans to cool them down (they live in a quite
> hot environment!). The PSU is a 500W one however I was concerned that
> the fans may pollute the hard disks' supply too much.. since they're
> motors after all!
> I've accidentally put a finger into a fan (what a pain!), and this
> stopped the fan for a moment and also caused hard disk errors, and
> thus convinced me even more that I need to decouple the fans' motors
> supply from the hard disks supply.
>
> But how? I'm just a hobbyst when it comes to electronics, so I may
> even intuite correctly, but I won't be really sure of my thoughts.
> Right now, they are that I need some big inductors (with as low DC
> resistance as possible), some (big?) capacitors and maybe also a big
> rectifier as well.
>
> The best scheme I've thought so far is to link all 6 fans together in
> parallel, then apply power through a circuit like this:
>
> 6 fans--*-ind-*-ind-*-ind-*-ind-*-ind-*-ind---diode------*-----PSU +12V
> | | | | | | |
> cap cap cap cap cap cap 6 HDs---PSU +5V
> | | | | | | |
> Gnd ----*-----*-----*-----*-----*-----*------------------*-----Gnd
>
> of course in the above scheme I don't even need them to be six,
> it's simply the more the better, I just happened to have six of
> them on my bench (more or less a coincidence).
>
> But now the big question: what's the right size of the caps, also
> to not overload the PSU at start? Will the diode really help the
> hard disks to not get some bad bad bad back EMF?
> Notice also that the wire that feeds the fans is AFTER all the wires
> that feed the hard disks (otherwise the voltage drop caused by the
> fans would have affected the hard disks). Yes, at least I know this. ;)
>
> Is my scheme ok? What can I do to improve it? I really care about
> the hard disks and the data they contain!
>
If you look at the label on the fans, it will show that the power
consumption is only about 2W and that it only takes 200mA or so. That
low power is why you went "ouch" instead of went looking for your finger..

Half a dozen of those aren't going to affect a 500W supply. They don't
need further decoupling. If you did get data errors at the same time, it
is more likely because you jarred the case or some other leads in your
haste to remove your person from the vicinity of the fan.

They also don't generate *external* back emf. The electronics in the fan
(yes there is some, unlike "conventional" motors) makes sure of that.

If you really care about your data, have a proven* backup strategy. *ie
including trying recovering the backups periodically...

You are quite right that excessive temperatures is one of the big
killers of disk drives - up there with mechanical jarring caused by such
things as attempted finger amputation. The answer to a hot environment
however, is to make things hotter, not simply shift the hot air around
faster..

Can I suggest sheets of thick aluminium as a starting point? I use
aluminium floor plates, cut to size - the same length as a drive but a
couple of inches wider. Bolt the drive to a plate, leaving a 2" wide
strip along one edge. Bolt one, or two, standard processor HSF
assemblies to the strip - but fit a suitable Peltier device between
strip and heatsink. I use 65W ones. Alternate the side you bolt the
drive on, so you have a staggered set in the complete assembly.

The Peltier device will pump the heat from the drive and plate into the
heatsink. Typically the plate will be well *below* ambient temperature.
The HSF will be well *above* ambient, so will transfer heat into the
airflow very nicely. The Peltier device, of course, also has to be
connected to a DC voltage source (eg 12v). It is, strangely enough, that
the HSF is much hotter than ambient that makes the whole system so
effective...

--
Sue

LinkBot