How does PSU efficiency affect me and do I really need an 80 Plus Gold Power Supply?

[Completely Bonkers]

But also inefficient at low loads (<15%). An 800W PSU driving an CPU idling at 50W will probably be only 50% efficient or worse.

Get a PSU to match your "idle" system, not just your "OC overload stresstest".

http://www.google.co.uk/images?hl=e...UTF-8&source=og&sa=N&tab=wi&biw=1040&bih=1060

Test your idle load. Whatever it is, take 5x this number for your PSU rating. If that isnt enough for your RAID and GPUs then obviously you will need a higher wattage PSU; but be aware that at idle you are horribly inefficient.

Let's do some numbers:

A mainboard and GPU idling at 60W

Take 5x 60W = 300W for your PSU.

Oh dear, that wont cover the SLI and RAID HDDs. OK, so we need a PSU at 700W. That's fine, but be aware that at idle you will probably be at 50% efficiency or worse.

Shame.

So for a gaming rig, you will be inefficient most of the time you arent gaming. But for a 24/7 server or NAS rig, make sure you do the numbers to ensure low heat, high efficiency, low power bills.

 

[nesco1801]

overkill yes, but PSU's are more efficient when they arent at max load

Thats what I've said. Its all about carbon footprint

Anyway PSU's are less efficient on lower end of curve also.

I'll get better and stronger PSU just to ease my paranoia, and so I can SLI one day

 

[Mussels]

some of these users fold and crunch, so they dont idle - or the systems off when they arent gaming, thus load is more important than idle.

 

[Completely Bonkers]

Agree. Depends on usage profile.

For workstations, gamers, crunchers, best to get a PSU that is 2x the average "loaded" profile, and hit that 50% "most efficient" line, as well as having scope for high load situations.

 

[newtekie1]

But also inefficient at low loads (<15%). An 800W PSU driving an CPU idling at 50W will probably be only 50% efficient or worse.

Get a PSU to match your "idle" system, not just your "OC overload stresstest".

http://www.google.co.uk/images?hl=e...UTF-8&source=og&sa=N&tab=wi&biw=1040&bih=1060

Test your idle load. Whatever it is, take 5x this number for your PSU rating. If that isnt enough for your RAID and GPUs then obviously you will need a higher wattage PSU; but be aware that at idle you are horribly inefficient.

Let's do some numbers:

A mainboard and GPU idling at 60W

Take 5x 60W = 300W for your PSU.

Oh dear, that wont cover the SLI and RAID HDDs. OK, so we need a PSU at 700W. That's fine, but be aware that at idle you will probably be at 50% efficiency or worse.

Shame.

So for a gaming rig, you will be inefficient most of the time you arent gaming. But for a 24/7 server or NAS rig, make sure you do the numbers to ensure low heat, high efficiency, low power bills.

That isn't necessarily true. If you pick a number like 85w, which is not an unrealistic point for a high end computer to idle at, a lot of modern 80+ Gold/Silver/Bronze PSUs are still over the 80% efficiency mark.

 

[Mussels]

That isn't necessarily true. If you pick a number like 85w, which is not an unrealistic point for a high end computer to idle at, a lot of modern 80+ Gold/Silver/Bronze PSUs are still over the 80% efficiency mark.

85W is actually pretty low for an idle, even systems running on IGP's tend to idle at 70W+ - few gaming systems get below the 100W mark.

 

[gourygabriev]

Also wanted to add that a good amount of Electric Providers also do their tiered billing, meaning their cost per kilowatt isn't a fixed, flat-rate but there is a premium added to people who use a lot of electricity. Like people who are using <100% of what is the allotted KW in a household will get like a lower $/KW compared to someone who is using 120% of what their allotted KW for a household. If I remember correctly, someone who is using 200% of the allotted KW per household is paying roughly 3x per KW compared to someone who used <100%. Instead of the bill going 2x the cost, they actually have to pay 6x more for using twice the amount. Last time I've checked, there were 5 different tiers being used by my Electric Provider (PG&E) in my area.

So I guess picking an 80 gold efficient power supply has its merits so people can save as much energy to keep them away from the higher tier of billing.

 

[billcat479]

I read a review on the standards, it was pretty much like what you put up there. They then focused on the marketing side, the increase in cost for a higher rated supply and the chance of it even making the ratings they claim are suspect.
The only flip side is how many computers are used in the U.S.
While the single user numbers are very small the total world use is increasingly huge in wasted power. It is important.
This is the one side where it makes any sense but the high cost going from bronze to gold rating is crazy.
Oh, Don't be putting the loss on the transformers, they are very low resistance and don't waste very much power they just convert it to equal but different voltage and current levels. The power regulators and resistors and rectifying circuits that are the bad guys. The huge heat sinks aren't on them just for kicks,
It's the Converting to DC and regulation that eats up the main loss of power. DC is a very power unfriendly source of energy. That one light bulb guy wanted to use DC power stations but they couldn't run big enough power lines to power even a small city and he was shoved aside by Westinghouse and friends. Using AC is why power stations get away with being so far away though it still is a huge waste in the lines. They do use liquid cooled superconducting lines close to the major cities to help out the line losses. Too bad they haven't got the room temp. superconducting figured out. That would be as huge as Fusion power generators.

 

[wwqzelasi]

Great post!Thank you for all the things learned from here and congratulations for the good work.

 

[Digital1]

A few days ago I upgraded from an 80+ bronze Silver Power psu(Which I've got to say has been a great psu since i've had it), to a Corsair AX 860i 80+ Platinum. I'm not clued up on rails and other technical stuff like a lot of you on here, but I did enough research to realise that if I wanted to protect my other components in my case I should maybe invest in better psu. It has eaten away at a big chunk of my Christmas money but hopefully it will save me money eventually and also save me from a lot of stress that blown components can cause.

 

[EarthDog]

People need to do the math on how much you are saving from say silver to platinum. Unless you use a distributed platform, it really is MAYBE a couple/few dollars per year.

As far as what psu to get. Unless you are clocking an AMD fx octcore to the moon under water, a quality 550w psu will power ANY single gpu system while overclocking both cpu and gpu as well as easily being able to support custom water and fans.

 

[Mussels]

People need to do the math on how much you are saving from say silver to platinum. Unless you use a distributed platform, it really is MAYBE a couple/few dollars per year.

As far as what psu to get. Unless you are clocking an AMD fx octcore to the moon under water, a quality 550w pain will power ANY single gpu system while overclocking both cpu and gpu as well as easily being able to support custom water and fans.

the money savings are tiny, unless you're coming from a generic PSU with passive PFC. the real benefits are that the quality PSU;'s have lower ripple (longer hardware life), lower temps (longer hardware life), lower noise (more sanity for owner) and a fancier sticker (more money for the manuf.)

 

[Arjai]

Awesome thread!!!!!!!!

I learned more here, on this thread, than, probably, a three credit course at college!

 

[HD64G]

If no one said it, I just want to add that when some PSUs say on their labels 500W max, this isn't the true wattage they can provide. They say another figure too about continouous power which is normally much lower. For example, my TR2 470W (which isn't mine for the last 2 years thankfully-terrible PSU) was 470 max and 310 continuous. This means that it provides only 310W and even this is not sure. It has 70% efficiency and gets hot when providing only 200W and produces coil whine at 250W. So, in cheap PSUs the max must be totally ignored. Only continuous is important and even this with a big reservation.

 

[newtekie1]

That is generally something you see on cheap power supplies, the better supplies tend to advertise the continuous load capability. Honestly, Thermaltake hasn't really put out a good power supply in a few years, they've really gone down hill.

 

[Tonduluboy]

Great Info! Thanks!
I been here longer than my registration date, but this is the 1st time i read this articles.
Now i know exactly why i bought my 80+ bronze PSU ( normally i just buy without 80+ certified psu)

 

[RCoon]

Great Info! Thanks!
I been here longer than my registration date, but this is the 1st time i read this articles.
Now i know exactly why i bought my 80+ bronze PSU ( normally i just buy without 80+ certified psu)

Bit of a necro for this thread. Almost 4 years old

 

[Mussels]

Bit of a necro for this thread. Almost 4 years old

not much has changed in PSU's since then

 

[assaulter_99]

Indeed, same old...

 

[Tonduluboy]

4 years but old thread but still informative to a newbie like me

 

[Gmr_Chick]

Thank you for this article.

Since this article/thread was created in 2010, there's since entered a new rating in the 80 Plus program - Titanium. Not that it would really matter, but would it be possible to update this with the new 80 Plus level?

 

[Ruru]

Going fine with a bronze certified PSU, what matters to me is that the PSU is filled with Japanese caps. Well, polymers can be Chinese, but electrolytics can't when I'm recommending a PSU.

 

[Ensefalon]

The transformers aren't 100% efficient is because there is energy loss, some of it is lost to heat ect. That's why there probably wont be 100% efficient because some it is always lost; unless they invent some new transformer or something else to do the job of a transformer.

Power grid transformers are 98% and above efficient, Experimental transformers using superconducting windings achieve efficiencies of 99.85% even such a small increase saves a lot of money.

EDIT: Efficiency is just that the computer needs like 250w for example but it will draw more than that because they lose energy.

A pc needing 250w will use about 294w at 85% efficiency.
It needs the extra because if it just drew just 250w from a wall it wouldn't be able to supply 250w when its converted by the transformer. It loses about 46w at 85% efficiency. So it needs to draw more power to make up the lost energy.

Stop blaming the transformer! Diodes are the ones stopping current from going backward!

The real thing is however, transformers in those power supplies are very tiny, and yet also very efficient as frequency is modulated into DC without the need for an 8 lb. iron core.
Power grid transformers are inefficient because their cores are huge and typically have a lower Fs than the current going through them.

 

[xman2007]

Stop blaming the transformer! Diodes are the ones stopping current from going backward!

The real thing is however, transformers in those power supplies are very tiny, and yet also very efficient as frequency is modulated into DC without the need for an 8 lb. iron core.
Power grid transformers are inefficient because their cores are huge and typically have a lower Fs than the current going through them.

And the award for thread necro of the year goes to........

DRUMROLL.........

@Ensefalon

Well done, replying to a 10 year old thread and even replying to someone like the thread just popped up on latest posts 10 minutes ago

 

[X71200]

That's definitely a good late night of drinking post...