This week only PSU

[rxmac]

the last time I did maths at school, 4x18 = 72, or am I missing something here ?

Lets see I run an E8500E0, 4gig DDR2 OCZ Sli 1066 on an ASUS P5Ke Wifi Premium mobo, BFG8800GTX with a duo orb cooler(needs its own supply), 2x750gig hitachi deskstar and 1x500 deskstar along with 3 x Pioneer DVDr Sata writers, 2 x case fans as well as the tower heating control system and a whole host of USB goodies attached. Take from that what you want, but this PSU handles it flawlessly.

You assume the gfx card uses all 4 rails which it wont, what i was getting at is that the 280gtx might require more current than a single rail can supply. Hence the suggestion of using a psu with a single large 12v rail.

I think it's 225w for the 280 (and 285w for the 295 when it's available!) Need a separate PSU just for the gfx card soon.

If this is true, and i have no reason to believe it isn't then 18a from a single 12v rail might not be enough.

 

[The Deuce]

You assume the gfx card uses all 4 rails which it wont, what i was getting at is that the 280gtx might require more current than a single rail can supply. Hence the suggestion of using a psu with a single large 12v rail.
.

absolute rubbish.

 

[rxmac]

Check your facts bud, psu's with spilt rails share them out amoungst the hardware FACT, so if the drives only use 10 amps, the rest of the power on that rail is effectively wasted. High power pc's would be better off using a large single rail psu.

I could supply various links but that would be against forum rules. But here is a quote from one article.

The problem with multi-rails, the company says, is that power tends to get stuck on the individual rails. If the PSU, for example, allocates 36 amps of power from rail 1 and 2 to the CPU but the processors only consume 22 amps – the rest cannot be reallocated to the GPU or hard drive array. With a single-rail design, if the CPUs only use 22 amps of juice, the rest can be sent to the GPUs or whatever else needs the 12 volt power because it all comes from a single bucket of power.

 

[The Deuce]

Check your facts bud, psu's with spilt rails share them out amoungst the hardware FACT, so if the drives only use 10 amps, the rest of the power on that rail is effectively wasted. High power pc's would be better off using a large single rail psu.

I could supply various links but that would be against forum rules. But here is a quote from one article.

obviously a very poorly designed PSU

 

[gt_junkie]

absolute rubbish.

obviously a very poorly designed PSU

As far as i know it doesn't work like that.

You can't simply add each rails amp rating to get the total the PSU can supply.

It will most probably be a good few amps below the 72a you got to by multiplying 18a by 4.

My Tagan piperock (900w) PSU has six 12v rails rated at 20a each. Going by your math that would be a staggering 120a - however, the psu can only actually supply around 72a.

gt

 

[OpenToSuggestions]

I just got (a few days ago) the corsair 750w for my rig as it was in the this week only for about £80.

Will power this rig, and whatever I throw at it whenI next upgrade.....however with 5 hard drives in an antec 300, the cabling is a nightmare!

 

[The Deuce]

As far as i know it doesn't work like that.

You can't simply add each rails amp rating to get the total the PSU can supply.
gt

in multi rail PSU's you have 1 rail dedicated to the CPU, the rest split between the other hardware. I can appreciate that yes 4x18a.. is 72 and that one wouldn't necessarily receive a full power of 72A more along the lines of 65A (OCZ1000w has 4x20A giving a combined current of 70A)

 

[EsaT]

I dont quite understand then why my system is failing with a burning smell coming from the psu and the power connectors are getting really hot.

Have you checked that connectors aren't loose? Nothing else heating up in component whose power connector is heating up?
Extra resistance of loose connector creates heat and especially "slow short" in some component could heat connector lot because of full output of 12V being available.

Half year ago one user in Finnish PC forum had wondered why PC had shutted down during night and didn't start anymore at morning.
After looking inside he found that there was darkened spot on motherboard and CPU's ATX12V connector had partially melted.
Motherboard turned to be dead, possibly short in CPU's power circuitry had drawn full 12V output through that connector before PSU noticed that it wasn't just component asking for some power.
(Seasonic made Antec PSU was OK except for partially melted connector)

the last time I did maths at school, 4x18 = 72, or am I missing something here ?

Your math is correct.

What is wrong is all fraudulent advertising BS about multiple rails.
There aren't such thing as separate 12V rails (except two badly configured ones in Enermax Galaxys) and they're just artificial current limiters dividing one 12V output to multiple to current limited wire groups, just like circuit breakers/fuses do in houses.


And that's because some people hit their heads together and decided that you can't be allowed to have more than 20A/240VA available from single cable/connector because of some safety requirements of user installable equipment or something like that.
So instead of telling the truth about artificial current limitations BS departments deviced these fraudulent "multiple rails", which can be also used to direct attention away from weak output of old design junks and other mediocre units because most people don't have enough brain power to look for actually meaningful figures.

Intel was one of the originators of this limit but they have quietly droppped it off from their PSU design guide/specifications years ago and device which exceeds that limit is simply classified as Level 6 device by UL (Underwriters Laboratories) and because of problem caused by these "multiple rails" Seasonic has been doing such PSUs for years despite what sticker says.
And all Seasonic made PSUs with LEVEL 6 marking which have been tested for this limitation have shown that they don't have that artificial limitation.
(their newest M12Ds probably has 12V divided to two high limits like sticker says because short circuit protection being triggered in short situation would start to require perfect short when allowed current rises to full capacity of 12V)

I think it's 225w for the 280 (and 285w for the 295 when it's available!) Need a separate PSU just for the gfx card soon.

~180W and about 160W/~13A of that through external power connectors so shouldn't trigger current limiter unless something else is conencted to same wire group.
http://www.xbitlabs.com/misc/picture/?src=/images/video/zotac-gfgtx280amp/zotac_power_full.png&1=1

GTX295 will be SLI of two cards so no wonder if it draws more.