Corsair SF450 Power Supply Review

[chromakey]

@Aris_Mp - Aris I always find your PSU reviews to be very informative and thorough, but I'm confused over current draw maximums, esp. on say the 12v rail; for instance, this one says 37.5A maximum (I guess I = P/V?), but surely if my house is on say a 15A socket ring main, I'm gonna blow the RCD immediately if it spikes this high on say a high demand graphics card?

You're looking at a switching supply, so power in (current times voltage) = power out times efficiency. So if you're going from 120v down to 12v, you draw ~1/9 if the current out.

Even a basic step down transformer based unregulated power supply will allow much higher currents to flow at lower voltages. It is just a fact of how things work.

It is easy to figure out the amperage from wattage alone. This will not take into consideration efficiency because it just adds more to do.

lets use 120 watts as a base number.

@ 240 volts it would be 120 / 240 = 0.5amp
@ 120 volts that is easy 120 / 120 = 1amp
@ 12 volts it is 120 / 12 = 10amp
@ 5 volts it is 120 / 5 = 24amps
@ 3.3 volts it is 120 / 3.3 = 36.36amps

It is all 120 watts, but you would need larger wires at 3.3 and 5 volts than at 12. This is why computers use the 12 volt rail so much, older ones that did not require so much power made use of the lower rails. This is also why older power supplies may not suitable for modern systems(not enough of the overall power on the 12 volt rail).

@nukemaster - thanks so much for your reply; so using your worked example above, a 200W graphics card on the 12V may draw up to (200/12) 17A? But what is this 12V/200W draw at the socket of the PSU? Most domestic ring mains for sockets are say 15A total, so if it was 1/1 the ring would blow straight away? Is the PSU "staging" the huge current draw in reservoir caps to protect the ring?

 

[nukemaster]

@nukemaster - thanks so much for your reply; so using your worked example above, a 200W graphics card on the 12V may draw up to (200/12) 17A? But what is this 12V/200W draw at the socket of the PSU? Most domestic ring mains for sockets are say 15A total, so if it was 1/1 the ring would blow straight away? Is the PSU "staging" the huge current draw in reservoir caps to protect the ring?

You need to think of wattage alone then covert to amps.

So a 200 watt video card will draw 16.66666forever amps from the power supply.

The power supply it self is not operating off of 12 volts, it is operating off of 120 or 240 volts. 200 watts is not much of a load at those voltages. Now power supplies do waste some power(very little compared to other things like lets say a high powered amplifier or linear power supply)

So if you want 200 watts at 12 volts from a 120 volt power source you draw 200watts + whatever the power supply requires to do the conversion. For simplicity I am going to say the power supply wastes 20%(this is not a great power supply, but a nice easy number at 20 watts for every 100 it puts out). So with this number we get 240 watts of power(the power supply is wasting 40 watts in the form of heat, you can not get 100% efficiency, but we can get better than 90% with switching power supplies). So now lets take the 240 watts and make it amps with ohms law again. 240 / 120 = 2 amps well within your 15 amps limit.

You can even buy a power meter that will show you how much power your computer uses at the wall.

Now onto the capacitors(and some uses). Capacitors act like a battery they charge when power is applied and discharge when it is removed(giving what ever is powered by them a shot time of power even if the power is no longer there).

AC(Alternating current) power is a sine wave(https://en.wikipedia.org/wiki/Sine_wave). While we say 120 volts we actually have 120rms(root mean square or average). This 120 volts actually peaks somewhere around 170 volts then passes into the negative past -120 then -170 then back past -120 past 0 +120 + 170 +120 0 and so on. This happens fast enough(relatively speaking) 50/60 times a second. Now our computer runs on DC(direct current). This DC has a + that is ALWAYS positive. other devices like diodes(you can even have FET based AC to DC conversion that are more efficient, but this is getting out of hand) are used to prevent the negative part of the cycle(or will full wave flip the negative positive), but you are still left with the top half of a sine wave from 0 - 120 to 170 to 120 to 0 and may have a pause if you are not doing full wave. Enter the capacitors. They charge upto to 170 volts when the power reaches its peak voltage and slowly discharge when the voltage drops. This removes a great deal of the wave in the power. It does not take it all off and takes stupidly huge caps to fully remove it, but this is MORE than enough for the regulation in the power supply to deal with it.

Now power supplies clearly have WAY more capacitors than just some on the incoming higher voltage. In a simple form switching power supplies switch full voltage on and off very fast to keep an average voltage that is regulated(does not move much even under huge load changes). The switching still has times that the power supply is not sending power out at all. Capacitors and Inductors (coils. They kind of work like a capacitor that resists changes in current instead of voltage) are used to filter this power into very clean DC with very little voltage variation(ripple).

They also have caps for other parts and chips(bypass and other caps). I will not even pretend to fully understand all of those extra chips. I know what some do, while I have no clue about others.

All these parts are sized(selected for the desired result) by an engineer who knows WAY more about this stuff than I ever will.

One more interesting use of capacitors is to block DC. When a cap is used on + and - terminals of a power source(DC) it will charge and help clean up voltage changes, but when it is inline with DC it will actually block it. This is used to allow AC to ride a DC signal in some audio amplifiers and other devices that have to remove a DC component of an AC signal. The have to be selected properly or else they will actually filter part of the AC out(they add some distortion so higher end audio gear will avoid as many signal path caps as they can), but again this is just a forum post.

You may be interested in this as well since it covers the AC to DC process(a very basic one compared to the high performance high efficiency one happening in a computer power supply). You can skip the 3 phase part if you want. it just adds more fun. Down lower you will see a section on Rectifier output smoothing to see what capacitor does in AC to DC conversion.
https://en.wikipedia.org/wiki/Rectifier

 

[turkey3_scratch]

@nukemaster: Hopefully I won't sound like a nitpicker, but just to clear confusion for chromakey, if by wasting 20% you mean 80% efficiency, it'd be 250W. If you mean 20% of 200W (240W), it'd be 83.3% efficiency.

 

[nukemaster]

@nukemaster: Hopefully I won't sound like a nitpicker, but just to clear confusion for chromakey, if by wasting 20% you mean 80% efficiency, it'd be 250W. If you mean 20% of 200W (240W), it'd be 83.3% efficiency.

I am willing to take that. I want easy fast numbers

Kind of depends when you factor the waste. I just want 20 watts wasted for every 100. I did not want a 192 watt video card. so 200 watts + 20 percent waste = 240

 

[chromakey]

@nukemaster - thank you so much for the detailed and thorough explanations you've given; you're very kind... i profess to know a little about electronics (my father was an electronics engineer) but it never hurts to have a 101 refresher.

i'm really sorry to labour a point, but i still can't get round this 12V draw capacity; Aris states that for this PSU above, the 12V rail is rated at 37.5A, so if that rail is going to draw anything like that current, the ring will blow... this seems to tie in with (in your example) 200W / 12V = 16.6A; and this is a 450W PSU so the 12V rail would max out at say 450/12=37.5A... i understand 200 or 240 / 120V = around 2A, no sweat, but this dratted 12V in there surely means the current draw jumps for that given wattage. it's crazy i know, logic dictates the PSU would never draw anything like 16 or 37A at the wall, and i'm sure if the components tried that, everything would go up in a puff of smoke (i've seen the gauge of cables needed to carry 30 odd A and it sure ain't PC PSU and card cables) - jeez it's driving me nuts trying to figure this, i must be really thick, why would a stat be given by Aris that the 12V rail can support 37.5A?

 

[turkey3_scratch]

37.5A of 12V load does not equate to 37.5A in your AC wall socket. It translates into much less. 37.5 * 12 = 450W.

Efficiency = DC / AC. So let's say efficiency is 80%. 80% = 450 / ac so ac = 562.5W

current * voltage = power

current * 120V = 562.5W

current = 562.5W / 120V
current in your wall = 4.69A

 

[nukemaster]

37.5amps @ 12 volts is only 450 watts. So you draw 450 watts + power supplies losses. Even if you had 100 watts of loss(made up number here) you would only be pulling 550 watts from the wall. 550 watts / 120 volts = 4.583forever amps.

Think of wattage as the limiting factor in this case.

A 120 volt plug with a 15 amp breaker and 14 gauge wire(the normal standard here) is good for 1800 watts.

Wire size/length(prevent losses) IS limited by current[amps] and voltage so to draw 1800 watts @ 12 volts you would need wire rated for 150 amps(but the incoming 120 would still be fine on 14 because it is 1800 watts. Not looking at losses in this example).

No single wire or trace in the computer should ever have to carry that much current.

For example a 6 pin PCI-E connector is rated at 75 watts(6.25amps). 8-pin ones are rated at 150 watts(12.5). They also use multiple conductors. Multiple wires are used to ensure power is delivered safely inside the computer. The short runs(compared to other cables) should help as well.

 

[turkey3_scratch]

Another way to think of it chromakey: power is the rate at which energy flows. Hardware does not run on "watts", it runs on energy. Electromagnetic energy is composed of magnetic and electric fields. High voltage and low current is a strong electric field and weaker magnetic field. High current and low voltage is a stronger magnetic field and weaker electric field. Same amount of energy, just a different configuration.

A power supply takes the alternating current and voltage and makes those electric and magnetic fields into the proper form for hardware.

 

[grantguy]

As a followup to @alastairab's comment, and to help any future visitor in a similar circumstance, I am doing a build using this PSU in the Fractal Design Node 304 which takes a standard-sized PSU. I ordered Silverstone's PP08 bracket from Amazon.com and shipped to Canada without trouble, and inexpensively. The mounting holes I think are standardized, and I had no issue mounting the bracket on the PSU nor fitting the combination into the case.