powersupply for 16disks +dual xeon

powersupply for 16disks +dual xeon

[Stephan van Hienen]

today tried to put 16 disks (15*ide7200 +1*scsi7200) in a intel sc5200
case
(which has an 450W powersupply)

when i try to powerup the system, the powersupply goes off after 3 seconds
i think this is because off the spinupp for the disks

any tips how to fix this ?
(as far as i know there is no spinup delay for ide disks)
maximum powersupply i found is 550W, not much more than i have in my
system

(i need an 24pin xeon powersupply)

(disks are WD180GB (which asks 15W for spinup)

Re: powersupply for 16disks +dual xeon

[Ross Vandegrift]

On Sat, Feb 01, 2003 at 06:32:16PM +0100, Stephan van Hienen wrote:
> any tips how to fix this ?

THIS IS NOT A PERMANENT FIX.

Power cycle the box a few times, quickly enough to not let the disks
loose all their momentum.  Power on for two seconds, power off, power
back on for two more seconds.  Two or three cycles should have the disks
spun up, and you can leave it on to boot the box.  

I hope it's obvious why this isn't such a great idea ::-)
Nonetheless, it does work.

-- 
Ross Vandegrift
ross@willow.seitz.com

A Pope has a Water Cannon.                               It is a Water Cannon.
He fires Holy-Water from it.                        It is a Holy-Water Cannon.
He Blesses it.                                 It is a Holy Holy-Water Cannon.
He Blesses the Hell out of it.          It is a Wholly Holy Holy-Water Cannon.
He has it pierced.                It is a Holey Wholly Holy Holy-Water Cannon.
He makes it official.       It is a Canon Holey Wholly Holy Holy-Water Cannon.
Batman and Robin arrive.                                       He shoots them.

Re: powersupply for 16disks +dual xeon

[Jeffrey Paul]

On Sat, 1 Feb 2003, Stephan van Hienen wrote:

> today tried to put 16 disks (15*ide7200 +1*scsi7200) in a intel sc5200
> case
> (which has an 450W powersupply)
>
> when i try to powerup the system, the powersupply goes off after 3 seconds
> i think this is because off the spinupp for the disks
>
> any tips how to fix this ?
> (as far as i know there is no spinup delay for ide disks)
> maximum powersupply i found is 550W, not much more than i have in my
> system
>
> (i need an 24pin xeon powersupply)

You could try hacking together another powersupply to run some of the
harddisks... I'm not sure of the ATX pinouts, but it shouldn't be too hard
to hack up a 12vdc relay or array of relays to one of the 12VDC pins on a
molex connector of the existing power supply to turn on the second power
supply, kind of slaving it to the first.

Let me know if you need further data on doing this, it's a websearch away.

-j

--------------------------------------------------------
  Jeffrey Paul      -datavibe-      sneak@datavibe.net
 aim: x736e65616b   phone: 130*21*16749 or 877-748-3467
--------------------------------------------------------

Re: powersupply for 16disks +dual xeon

[Adam Luter]

The green pin on the adaptor is the signal for power on.  You need to
ground it.  I don't know how to signal a power off.

If you are curious, the reason I know this is because I have a nail
attached to that pin, and my normal power switch alligator clipped to
that nail, and a grounded portion of my case.  I use this to turn my
box on (you have to hold down for a few seconds).

The reason is that my motherboard decided (probably traumatized) not
to listen to power on requests anymore.  But it still knows how to
turn itself off through the normal hardware.

-Gryn (Adam Luter)

> You could try hacking together another powersupply to run some of the
> harddisks... I'm not sure of the ATX pinouts, but it shouldn't be too hard
> to hack up a 12vdc relay or array of relays to one of the 12VDC pins on a
> molex connector of the existing power supply to turn on the second power
> supply, kind of slaving it to the first.
> 
> Let me know if you need further data on doing this, it's a websearch away.
> 
> -j
> 
> --------------------------------------------------------
>   Jeffrey Paul      -datavibe-      sneak@datavibe.net
>  aim: x736e65616b   phone: 130*21*16749 or 877-748-3467
> --------------------------------------------------------
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: powersupply for 16disks +dual xeon

[Ross Vandegrift]

On Sat, Feb 01, 2003 at 01:30:10PM -0500, Jeffrey Paul wrote:
> You could try hacking together another powersupply to run some of the
> harddisks... I'm not sure of the ATX pinouts, but it shouldn't be too hard
> to hack up a 12vdc relay or array of relays to one of the 12VDC pins on a
> molex connector of the existing power supply to turn on the second power
> supply, kind of slaving it to the first.

I Am Not an Electrican Engineer, But My Boss Is.  I did this once, and
he had a cow.  There was some problem he yelled at me, having to do with
different PSUs pulling voltage up at a different rate.  There's someway
he said that the differential in voltages would feed back through the
IDE channels, for some reason....

The short of it was that someone qualified to judge said it was a really
bad idea to use multiple seperate power supplies.

-- 
Ross Vandegrift
ross@willow.seitz.com

A Pope has a Water Cannon.                               It is a Water Cannon.
He fires Holy-Water from it.                        It is a Holy-Water Cannon.
He Blesses it.                                 It is a Holy Holy-Water Cannon.
He Blesses the Hell out of it.          It is a Wholly Holy Holy-Water Cannon.
He has it pierced.                It is a Holey Wholly Holy Holy-Water Cannon.
He makes it official.       It is a Canon Holey Wholly Holy Holy-Water Cannon.
Batman and Robin arrive.                                       He shoots them.

Re: powersupply for 16disks +dual xeon

[Gordon Henderson]

On Sat, 1 Feb 2003, Ross Vandegrift wrote:

> On Sat, Feb 01, 2003 at 01:30:10PM -0500, Jeffrey Paul wrote:
> > You could try hacking together another powersupply to run some of the
> > harddisks... I'm not sure of the ATX pinouts, but it shouldn't be too hard
> > to hack up a 12vdc relay or array of relays to one of the 12VDC pins on a
> > molex connector of the existing power supply to turn on the second power
> > supply, kind of slaving it to the first.
>
> I Am Not an Electrican Engineer, But My Boss Is.  I did this once, and
> he had a cow.  There was some problem he yelled at me, having to do with
> different PSUs pulling voltage up at a different rate.  There's someway
> he said that the differential in voltages would feed back through the
> IDE channels, for some reason....
>
> The short of it was that someone qualified to judge said it was a really
> bad idea to use multiple seperate power supplies.

I've built many SCSI based systems where the disks are in a separate
enclosure with their own power supplies. There are many commercial system
built just like this. (eg. Sun used to supply separate external CD ROMs,
tapes and hard drive storage boxes all with their own PSU) I don't see why
it would be a problem with IDE drives - cable length would be the one
thing you might have a problem with though.

It's a lot of eggs in one basket though - do you really need contiguous
filespace, or might it be better split up into separate servers for each
'volume' ?

Good luck!

Gordon

Re: powersupply for 16disks +dual xeon

[Andre' Breiler]

Hi,

On Sun, 2 Feb 2003, Gordon Henderson wrote:

> On Sat, 1 Feb 2003, Ross Vandegrift wrote:
>
> > On Sat, Feb 01, 2003 at 01:30:10PM -0500, Jeffrey Paul wrote:
> > > You could try hacking together another powersupply to run some of the
> > > harddisks... I'm not sure of the ATX pinouts, but it shouldn't be too hard
> > > to hack up a 12vdc relay or array of relays to one of the 12VDC pins on a
> > > molex connector of the existing power supply to turn on the second power
> > > supply, kind of slaving it to the first.
> >
> > I Am Not an Electrican Engineer, But My Boss Is.  I did this once, and
> > he had a cow.  There was some problem he yelled at me, having to do with
> > different PSUs pulling voltage up at a different rate.  There's someway
> > he said that the differential in voltages would feed back through the
> > IDE channels, for some reason....
> >
> > The short of it was that someone qualified to judge said it was a really
> > bad idea to use multiple seperate power supplies.
>
> I've built many SCSI based systems where the disks are in a separate
> enclosure with their own power supplies. There are many commercial system

SCSI is different as it was designed with this in mind (you drive the
lines against termination power which is supplied by _one_ source).

> built just like this. (eg. Sun used to supply separate external CD ROMs,
> tapes and hard drive storage boxes all with their own PSU) I don't see why
> it would be a problem with IDE drives - cable length would be the one
> thing you might have a problem with though.

I'd think IDE isn't designed for this (even that it works most times if
the powersupplies are close enough in terms of voltage and grounding).
So here is a risk that you would overstress something but usally it works
well.

Bye Andre'
-- 

Re: powersupply for 16disks +dual xeon

[Peter Bartosch]

Hi!


> On Sat, Feb 01, 2003 at 01:30:10PM -0500, Jeffrey Paul wrote:
> > You could try hacking together another powersupply to run some of the
> > harddisks... I'm not sure of the ATX pinouts, but it shouldn't be too hard
> > to hack up a 12vdc relay or array of relays to one of the 12VDC pins on a
> > molex connector of the existing power supply to turn on the second power
> > supply, kind of slaving it to the first.
> 
> I Am Not an Electrican Engineer, But My Boss Is.  I did this once, and
> he had a cow.  There was some problem he yelled at me, having to do with
> different PSUs pulling voltage up at a different rate.  There's someway
> he said that the differential in voltages would feed back through the
> IDE channels, for some reason....
> 
> The short of it was that someone qualified to judge said it was a really
> bad idea to use multiple seperate power supplies.

the problem with the feed back occurs when you connect the (e.g) +12V
lines of both PSU's, (and the GND offcourse) the PSU with more power
tries to supply the other and will get damaged

if you only connect the GND, and use the second supply for hdd's only it
will work. the resistance of the ICs (interface, etc) will prevent the
feet back 

Peter

Re: powersupply for 16disks +dual xeon

[Ricky Beam]

On Sat, 1 Feb 2003, Ross Vandegrift wrote:
>I Am Not an Electrican Engineer, But My Boss Is.  I did this once, and
>he had a cow.  There was some problem he yelled at me, having to do with
>different PSUs pulling voltage up at a different rate.  There's someway
>he said that the differential in voltages would feed back through the
>IDE channels, for some reason....
>
>The short of it was that someone qualified to judge said it was a really
>bad idea to use multiple seperate power supplies.

Ok, I'm gonna have to jump in here now.  People have been doing this for
DECADES with SCSI hardware.  It's a load of BS.  As long as there's a
common ground line, you won't release any smoke. (And I bet they have a
common ground through the AC plug.)

Can you say "external enclosure"?  I knew that you could.

(Note: only HVD SCSI hardware has line isolation circuits -- required per
       spec to avoid setting SE busses on fire if you plug the wrong thing
       in.)

--Ricky

PS: Yes, SCSI feeds power into the bus to power terminators. (usu. everything
    feeds power to the bus... common factory default)

Re: powersupply for 16disks +dual xeon

[Rabeeh Khoury]

>
>
>(Note: only HVD SCSI hardware has line isolation circuits -- required per
>       spec to avoid setting SE busses on fire if you plug the wrong thing
>       in.)
>  
>
Serial ATA does not have this problem since the signalling itself is 
differential.
I have about 24 drives on 3 external enclosures connected to 3 adapters 
and working perfectly

Re: powersupply for 16disks +dual xeon

[Mikael Abrahamsson]

On Sat, 1 Feb 2003, Stephan van Hienen wrote:

> any tips how to fix this ?
> (as far as i know there is no spinup delay for ide disks)
> maximum powersupply i found is 550W, not much more than i have in my
> system

You should try to fit another powersupply to drive the disks. ATX 
powersupplies can be made powermanently on by shorting ping 14 and 15 (I 
think it was, there is a FAQ page out there I found on when I was looking 
a month back). If you don't want to do this, get yourself an AT power 
supply.

I run 10 disks off of a 350W power supply (dual 1.13GHz pIII board) and it 
has worked fine, I think you're not too far off the limit. Perhaps that 
550W will do the trick.

-- 
Mikael Abrahamsson    email: swmike@swm.pp.se

Re: powersupply for 16disks +dual xeon

[jlewis]

On Sat, 1 Feb 2003, Mikael Abrahamsson wrote:

> You should try to fit another powersupply to drive the disks. ATX 
> powersupplies can be made powermanently on by shorting ping 14 and 15 (I 
> think it was, there is a FAQ page out there I found on when I was looking 
> a month back). If you don't want to do this, get yourself an AT power 
> supply.

So can this trick (connecting 14 to 15) be used as a work-around for brain 
damaged motherboards that won't power-on after power-loss to keep a system 
always on?
 
> I run 10 disks off of a 350W power supply (dual 1.13GHz pIII board) and it 
> has worked fine, I think you're not too far off the limit. Perhaps that 
> 550W will do the trick.

We have a dual PIII server with 2 9gb and 8 36gb drives running on a 450W 
supply (IIRC), but those are all SCSI drives with various spin-up delays.

On another system we were using to test a large RAID array, we just rigged 
up an AT power supply to half of the disks.  That works fine on the bench, 
but is a little sloppy if you're trying to build a production server.

----------------------------------------------------------------------
 Jon Lewis *jlewis@lewis.org*|  I route
 System Administrator        |  therefore you are
 Atlantic Net                |  
_________ http://www.lewis.org/~jlewis/pgp for PGP public key_________

Re: powersupply for 16disks +dual xeon

[Mikael Abrahamsson]

On Sat, 1 Feb 2003 jlewis@lewis.org wrote:

> So can this trick (connecting 14 to 15) be used as a work-around for brain 
> damaged motherboards that won't power-on after power-loss to keep a system 
> always on?

I don't know, but I wouldnt suggest it and I wouldnt do it on a production 
system without extensive testing first. I haven't seen the behaviour you 
suggest above in motherboards since a PPRO200-board way back, all newer 
boards I have worked with have an option in Power Save-settings as to what 
to do after a power loss and the following power return (that is usually 
default to the desktop setting, ie do not power on after power return).
  
> We have a dual PIII server with 2 9gb and 8 36gb drives running on a 450W 
> supply (IIRC), but those are all SCSI drives with various spin-up delays.

I have 9 IDE drives all powering up at the same time (mostly IBM 7200 rpm 
units) without any problems on my 350W).
 
> On another system we were using to test a large RAID array, we just rigged 
> up an AT power supply to half of the disks.  That works fine on the bench, 
> but is a little sloppy if you're trying to build a production server.

I know people that have done this as well when they have lots of drives. 
It's wa cheaper than trying to find a power supply with twice the output 
of a regular power supply.

-- 
Mikael Abrahamsson    email: swmike@swm.pp.se

Re: powersupply for 16disks +dual xeon

[jlewis]

On Sun, 2 Feb 2003, Mikael Abrahamsson wrote:

> > So can this trick (connecting 14 to 15) be used as a work-around for brain 
> > damaged motherboards that won't power-on after power-loss to keep a system 
> > always on?
> 
> I don't know, but I wouldnt suggest it and I wouldnt do it on a production 
> system without extensive testing first. I haven't seen the behaviour you 
> suggest above in motherboards since a PPRO200-board way back, all newer 
> boards I have worked with have an option in Power Save-settings as to what 

Unfortunately, alot of cheaper boards have this problem.  Some of the
lower end (either Tyan or Asus, don't remember which) have this.  I was
looking at turning some IBM PC300 (Celeron) systems (really small desktop
cases)  into small dedicated DNS servers, but they also apparently can't 
be set to power-on after power-loss....But I guess I'm drifting way off 
topic now.

----------------------------------------------------------------------
 Jon Lewis *jlewis@lewis.org*|  I route
 System Administrator        |  therefore you are
 Atlantic Net                |  
_________ http://www.lewis.org/~jlewis/pgp for PGP public key_________

Re: powersupply for 16disks +dual xeon

[Andre' Breiler]

Hi,

On Sat, 1 Feb 2003 jlewis@lewis.org wrote:

> On Sat, 1 Feb 2003, Mikael Abrahamsson wrote:
>
> > You should try to fit another powersupply to drive the disks. ATX
> > powersupplies can be made powermanently on by shorting ping 14 and 15 (I
> > think it was, there is a FAQ page out there I found on when I was looking
> > a month back). If you don't want to do this, get yourself an AT power
> > supply.
>
> So can this trick (connecting 14 to 15) be used as a work-around for brain
> damaged motherboards that won't power-on after power-loss to keep a system
> always on?

No, this won't work in most cases (as the board doesn't know that it
should start up).
What I usally do is an really small mono flop + trigger combination which
waits for 10s to switch an relais (you can draw power from the standby
power) which simulates the atx power switch. Note, that for some boards
to function correctly it's needed to release the relais after 5V are
present on the normal power.

Bye Andre'
-- 

Re: powersupply for 16disks +dual xeon

[Stephan van Hienen]

On Sat, 1 Feb 2003, Mikael Abrahamsson wrote:

> You should try to fit another powersupply to drive the disks. ATX
prefer not to put 2 psu's in my system (not really much space (and
connecting it, can be difficult))

> think it was, there is a FAQ page out there I found on when I was looking
> a month back). If you don't want to do this, get yourself an AT power
> supply.
>
> has worked fine, I think you're not too far off the limit. Perhaps that
> 550W will do the trick.
system runned ok with 11 disks (10ide7200 +1scsi 7200) for a few months
also disks spinup ok if i remove 2 disk (so 13*ide7200 +1*scsi7200)

so looks like the 450 has just a few W's short for powering all disks
and i should look for a 550 (or maybe 650 (do they exist?))
any tips for brands to look for ?

i found these 2 :

http://www.enermax.com.tw/eg851ax-vh.htm
http://www.enermax.com.tw/eg651pvefm24P.htm

(why is the 851 talking about +12V1 +12V2 and +12V3 ?)

Re: powersupply for 16disks +dual xeon

[Peter L. Ashford]

Stephan,

> today tried to put 16 disks (15*ide7200 +1*scsi7200) in a intel sc5200
> case
> (which has an 450W powersupply)
>
> when i try to powerup the system, the powersupply goes off after 3 seconds
> i think this is because off the spinupp for the disks
>
> any tips how to fix this ?
> (as far as i know there is no spinup delay for ide disks)
> maximum powersupply i found is 550W, not much more than i have in my
> system
>
> (i need an 24pin xeon powersupply)
>
> (disks are WD180GB (which asks 15W for spinup)

Someone already suggested a 550W PS.  The alternate path is the 460W
redundant from Zippy.  These are a bit pricey, but they have three 230W
modules.  The system should be able to run on two modules, but would
require all three for initial power on.

Adding a slave power supply should work, but you'll have to make sure that
all voltages have adequate load.  Most switching power supplies require a
minimum load to function properly.

Good luck.
				Peter Ashford

Re: powersupply for 16disks +dual xeon

[Alvin Oga]

hi ya

On Sat, 1 Feb 2003, Peter L. Ashford wrote:

> Stephan,
> 
> > today tried to put 16 disks (15*ide7200 +1*scsi7200) in a intel sc5200
> > case
> > (which has an 450W powersupply)
> >
> > when i try to powerup the system, the powersupply goes off after 3 seconds
> > i think this is because off the spinupp for the disks
> >
> > any tips how to fix this ?
> > (as far as i know there is no spinup delay for ide disks)
> > maximum powersupply i found is 550W, not much more than i have in my
> > system
> >
> > (i need an 24pin xeon powersupply)
> >
> > (disks are WD180GB (which asks 15W for spinup)

225W      15 drives * 15
50-75W ?  dual xeon 
30-50W    motherboard

seems like there's plenty of power ... just not at 12V to spin the
drives ...

	-
	- maximum power/current is used to spin the drives up
	- and if the ps is designed incorrectly... it will take 
	- current away from +3.3 and +5v ... screwing up the mb
	- ( measure the voltage w/ a storage scope to be safe )
	-

	- if you can add a separate 12V power supply for powering the
	disks.. you'd be in biz..

	- you'd have to power up the disks first ..
	and for shutdown, you'd have to shut off the mb first

	- or add a simple RC circuit to delay drive spin ups
	and than allow the soft "power-on" to the motherboard

- simple resister + cap ... for each 2 or 4 hard drives  
  ( y-power cables )   - say 10K + 470uf   and another one at 470K + 470uf
	- probably allow for 5-10 second delays of disk power ups

	and until the drives is powered up.. the mb should wait
	befor it powers up

- you'd need a separate resistor/cap/transistor for the
  soft "power-on" signal
	- the last thing to do ... for the system to start its
	real boot process

lot cheaper than the $600 dual power supplies
	and marginal 500W supply that can't handle the 12 drives


-- and for motherboards that does NTO boot up by itself
   add a 470uf capacitor across the power-on sig and now it boots by 
   itself


-- and there's places that advertise free PCB for 3-5 pieces ( 2 layers )
	-- never called um though

c ya
alvin

Re: powersupply for 16disks +dual xeon

[Mikko Saukkoriipi]

Stephan van Hienen wrote:

>today tried to put 16 disks (15*ide7200 +1*scsi7200) in a intel sc5200
>case
>(which has an 450W powersupply)
>
>when i try to powerup the system, the powersupply goes off after 3 seconds
>i think this is because off the spinupp for the disks
>
>any tips how to fix this ?
>(as far as i know there is no spinup delay for ide disks)
>maximum powersupply i found is 550W, not much more than i have in my
>system
>
>(i need an 24pin xeon powersupply)
>
>(disks are WD180GB (which asks 15W for spinup)
>-
>To unsubscribe from this list: send the line "unsubscribe linux-raid" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>  
>
More important than getting larger wattage powersupply is to find one 
that can provide enough power on the 12V rail. All powersupplies are 
really not built the same. I have a cheap and shitty 450W Q-tec 
powersupply in my server with 7 harddrives. It kept crashing when 
copying between harddrives. I found Firingsquad's powersupply guide(1) 
and calculated that my server requires 20A on the 12V rail, but the 
powersupply could only deliver 12A/15A Max. Now I have 2 AT 
powersupplies for the harddrives, but this is not a good solution. My 
friend recently bought a 465W Enermax that can provide 33A in the 12V 
rail. What I've seen, Enermax PSUs provide most power to the 12V rail. 
Fortron-Source top-end models would be another possibility, like 
FSP550-60PLG.

Miksa

Re: powersupply for 16disks +dual xeon

[dean gaudet]

On Sun, 2 Feb 2003, Mikko Saukkoriipi wrote:

> powersupplies for the harddrives, but this is not a good solution. My
> friend recently bought a 465W Enermax that can provide 33A in the 12V
> rail. What I've seen, Enermax PSUs provide most power to the 12V rail.
> Fortron-Source top-end models would be another possibility, like
> FSP550-60PLG.

zippy p/s also seem to be in the ballpark of enermax... i.e.
<http://www.zippy.com.tw/2_power/2_product/2u/ac_input/P2G-6460P.htm> has
32A on 12V.

that's what i've got in a box with 9 ide disks.

i'm really happy with the 3U chassis i've got too --
<http://www.rackmountpro.com/8productpage.cfm?prodid=558&u=RM3U3Q-IDE-P300W>.
since the p/s is to the left of the mobo, my 3ware card is more in the
center of the box and that lets me use 18" cables to reach all drives (i
might have got away with 12" for some drives, didn't try though).

-dean

Re: powersupply for 16disks +dual xeon

[Stephan van Hienen]

Ok received an Enermax 550W (650W peak) yesterday
(which has 36A on the +12V)

works ok now with 15*ide7200 and 1*scsi7200