Power for 11 IDE Drives

Joined
Dec 20, 2000
Messages
590
I have checked the stickies - and the link to the power supply calculator does not work :eek: :rolleyes: (for me at least)

Anyhow - Here's my situation - I have 11 drives i'm looking into putting into this system. I know that the spin-up will draw a ton of power - and continous power draw will be quite high as well. I'm running without staggered spin-up (and I doubt i'll be buying a card that supports this for this project). Also a linkie to my mainboard post...

The two WD's (WD1000JB) will draw:
2.2A on spinup 17W
350 mA on write 8W
900 mA on seek 14W

The 4 120GB Seagates (ST3120022A) will draw:
2.8A on spinup
on write 12W
on seek 12.5W

The 4 80GB Seagates will draw:
2.8A on spinup
on write 12W
on seek 12.5W

The maxtor (30GB) will draw:
on spinup
396mA (5v) 440 (12v) on write (this doesn't sound write when comparing it vs the seagates)
562mA (5v) 440mA (12v) on seek

I would have to guess the numbers which I could not find - are based on similar factors to the WD drives, except a litle bit higher. SO i'm roughly looking at:

29.6A on spin-up
4.5A on write/idle
10.2A on seek

187W on spin-up
132W on write
137.5W on seek

Now I know I couldn't provide all the numbers for each rail (couldn't find it on the manuf. sheets) - but I want to ensure I have enough power for all of these drives - hence the gurus :D Ockie? PsRage? (sp - sorry) Mods? Others?

Any suggestions? Power supply(ies) is on my buy list - but I'd like to not spend an arm and a leg... I have a 300W enermax & a crappy ($20) 500W a 250W (deer - yuck), 500W antec in my main system (not for use here).
 
My old Chieftec 360W PSU (18A @ 12V) was able to power the 15 drives in my fileserver (Sempron64 2800+, 512MB, Radeon 9200SE) without any problems, though it was very close to the limit. Unless you are running a Dual Xeon/Opteron setup or something like that i think any quality 350+ Watt PSU should be enough.
 
Somehow I doubt that a 350W power supply will do it. I have bought an Enermax Liberty 620W power supply. 80% max draw on Watts & Amps is the most I want to put my system through.
 
yup takaman's is gone Im working on a guide to replace it

err... could you list all the rest of the components including fans (size\speed) pumps ect
Id rather not hunt and peck through all your posts to find that ;)
 
heh, sure

2x 80mm LED (estimate 2,600rpm)
2x 80mm (1 right now - pretty sure will be modding the case to add second) (estimate 2,600rpm - i should put in my 80mm delta there LOL!!!!)
1x 120mm (1,500rpm, 0.08 Amp)

NForce2 Mobo (nf7-s 2.0), 1 stick ram, pci graphics card, 2 controller cards, barton 2500+, DVD-ROM, external floppy (usb)
The 11 drives listed

I think thats everything... I may have to change the 2 controller cards for a single one... (not enuff system ROM)

After I get the system up and running - is there a way to determine how many amps & watts are being drawn on each rail? (it has 2 if i'm not mistaken) (i'm by far an expert - the 5v, 3v & 12v lines also---) MBM? I know it gives voltages...
 
Dumass_Freakboy said:
Somehow I doubt that a 350W power supply will do it. I have bought an Enermax Liberty 620W power supply. 80% max draw on Watts & Amps is the most I want to put my system through.

I can assure you it (my 360W chieftec) survived lots of stressed situations, but i've recently replaced it with an Akasa 460W because i'll add some more drives soon (up to 20). 620W is a 'little' bit too much power but there isn't anything wrong with playing it safe. ;)
 
Dumass_Freakboy said:
heh, sure

strangely NjO is correct

if the 350W is distributed in proportion to the amps required by the config
and its actually rated at a reasonable test temperature
and in his example the server doesnt see alot of on off, so any instability at startup when it might be heavily stressed to spinup all those drives is a truely rare thing and wont be wearing on the supply and or components day in and day and day out as an on off desktop would


however
a modern configuration wont be asking for a "balanced" proportion of amps across the all rails
its going to be heavily biased to just the +12V, and thus with the exception of say an Antec Phantom 350 (an unusual supply) most 350W supplies will be woefully short on the amps where youd need them. Conversely a modern monster watt ATX12V v2.0 will have the +12V amps you need however its artificially hyped up on its total wattage becuase alot of it is on the +3.3V and +5V rails for backward compatibility and simply unavailable and uneeded in a modern config, were it to be deducted your still near 350W total

consider this 25A x 12V = 300W

which is why exactly what is getting powered makes all the difference

I'll work up your configs amp needs in the next half hour ;)

Dumass_Freakboy said:
After I get the system up and running - is there a way to determine how many amps & watts are being drawn on each rail?

yes send it to my house where I'll remove all the connectors run the wires for the main rails through my 27 DC current sensors add the connectors back, attach the sensors to my Keithley 2700 Data Acquisition System and plot you out a realtime amp draw through a power cycle and benchmark run. :p

short of that not really
you can measure what its drawing from the AC outlet pretty easy, but what that breaks out to as amps per rail requires the above, and AC figures also include energy lost as heat do to inefficiency
 
As an eBay Associate, HardForum may earn from qualifying purchases.
Ice Czar, maybe this is a topic for another thread, but how would a professional audio interface be for monitoring sensors like that? I've got access to an interface that samples at 96 kHz in 24-bit resolution, so I think you could get a really good picture of what your power supply is outputting. It's a little higher bit-depth (although I think it's only really sampling at 20 bits), but a whole lot more samples/sec - 96000 as opposed to a max of 3500 for the Keithley linked. I'd probably need to use a resistor of some sort to bring the voltage from +12 or +5 down to 1.23 vrms. Do you know if this would be a reasonable way of checking power supply, or is it a hopeless cause?

 
brother Im fumbling around in the dark myself

ask gee :p

as far as the sample rate, those DAQs are employed explicitly for that and the higher sample rate Id say isnt really needed for basic current measurement, if you were going after AC Ripple on the DC line youd use an Oscope

as of right now Im employing mine for thermocouple logging

Logging.jpg
 
HAve youu considered selling your drives and buying 4 big drives in the 250-300GB range?

For a little under 400$ you could tranfer the data to the bigger drives and then sell the drive on e-bay and get 250-300 bucks.

I was lucky enough to take advantage of the fry's thanksgiving weekend deal on 300GB SATA drives for 89$ and actually made money selling 6 250GB SATA drives

Ansd previously because a Silverstone TJ-06 case won't work with 6 IDE drives (cabling is a mess I sold those, replaced them with SATA drives and made a profit selling the IDe drives.

YMMV
 
renegade44e said:
I was lucky enough to take advantage of the fry's thanksgiving weekend deal on 300GB SATA drives for 89$ and actually made money selling 6 250GB SATA drives

Was it you who posted the pic of the pyramid of boxes?

I was an idiot for not taking advantage of that deal...
 
Ice Czar said:
as far as the sample rate, those DAQs are employed explicitly for that and the higher sample rate Id say isnt really needed for basic current measurement, if you were going after AC Ripple on the DC line youd use an Oscope
Digital to Analog Qonverters? :p Sorry, I'm just too used to audio stuff. At 96k, though, you get samples every 10 microseconds (more or less) - wouldn't that be enough resolution to see the ripple? Audio interfaces only go up to 384k (3 us) that I've seen, so if it's not enough it's time to get friendly with the people in the physics labs on campus :D

Who's gee? I live in my own little world named Disk Storage. I seem to recall from another thread in a galaxy far far away that he's one of the electronics gurus... makes sense.

 
My thread has been hijacked :(

Ah well, I have a nice power supply now that I can use in an SLI rig sometime next year hopefully.
 
Back
Top