need suggestion: NAS with BitTorrent

[Mussels]

theeldest: by saying 'don't stick a 600w powersupply in this, it's just going to use tons of power' you're making yourself sound very uneducated.

PSU's dont use their rated amount for no reason... they only use whats needed. my PC Only uses 300W of a 600W PSU, you should really look into that.

 

[theeldest]

I'm finding the link to back this up (it's either on Tom's Hardware, Anandtech, or Ars).

But the efficiency of the power supply is correlated to the percent of it's max power that's being used. So a PSU with a rated power of 600 watts and 80% efficiency is going to be giving 80% efficiency when it's supplying 480w or more. As the power reduces, it's going to have worse efficiency.

I didn't mean to suggest that it'll be using 600w no matter what, just that maximum efficiency is achieved by using a power supply that's close to what you need without going far over.

I'll have the link shortly.

 

[theeldest]

Here's an image from a review of a PSU on Anandtech:

This image is the efficiency graph for a single power supply (and Enermax Revolution 85%+ efficiency 1000w PSU). The x range is the power that is being used, the y range is the efficiency that the PSU is providing.

The peak efficiency is in the 500w range, even though it's a 1000w PSU. If it's strongly underpowered, then the efficiency crashes to the 70%-80% range. Also notice, this is a 85% Plus efficiency PSU. Your's probably isn't. So subtract 5% - 15% from those efficiencys to get an idea of what yours would be.

(Here's the link to the Article)

That's what I've got for now.

If you still think I sound like an idiot, well ... so be it.

Just trying to help.

 

[theeldest]

Ok. Found the Article I was looking for.


Efficiency is the vertical axis and power used is the horizontal axis. Each line is for a different power supply. So it should be fairly obvious that some power supplies would be better for a specific computer configuration than others. (ie, some would use less power even though the computer is using the same amount of power)

Here's an image:

Mussels: by saying 'PSU's dont use their rated amount for no reason... they only use whats needed' you're making yourself sound very uneducated. (<= that actually doesn't make you sound uneducated, because that's right (they won't use their rated amount for no reason). but it's not what I was saying. if you're going to point out someone's ignorance, can you next time be sure you read their comments right?)

(also, your PSU has 82% efficiency. And if the graph for your PSU resembles those above, the 300w use range will be close to the max efficiency. So assume 82% efficiency (the rated efficiency for your PSU), your PSU is actually using 365w when your computer is drawing 300w. (300w is 82% of 365))

 

[Mussels]

i actually have a wall meter. at the ~300W that i use, mines at 92% efficiency, according to my meter.

i'm measuring what my PC uses at the WALL already, it uses around 300W, at 92% efficiency... meaning my PC is only using 260-280W, before PSU efficiency matters.


'theory' and reality are two very different things.

 

[theeldest]

How are you coming up with the 92% efficiency? What are you using to measure the power usage by everything in your computer? The only way to actually do that is to generate the load with a load generator. Otherwise it's just guesswork, and as you said, "'theory' and reality are two very different things" (so I hope you're not using wattage specs for the components).


Also, about that other thing, do you still think that someone should put a 600w PSU in a system that should have 100w - 200w power draw? (as that's what my original argument was, the one you so elegantly told me was wrong.)

 

[Mussels]

load, is generated by running a defrag on all HDD's, ATI tool, and OCCT set to CPU and ram.

Thats running my CPU, video card, ram, and hard drives all at 100% (or as close as possible to it).
My meter measures power consumed, and has a power efficiency reading as well.

Its not that i think your argument against a 600W PSU in a 100W system is wrong - its your reasons for it, that was wrong.

 

[theeldest]

Efficiency of a power supply is the power that it sends out the leads, divided by the power going in from the wall. If you can't measure the power going through the leads, you can't measure it's efficiency.

The efficiency you're measuring is actually how efficient it is at using the power from the wall, before it even converts to DC. Most power supplies need a certain voltage, and will convert what ever they get from the wall (so from 108v in the wall (your wall's not perfect) to the 110v it needs). Power is lost during this conversion and can be measured by the back voltage created.

So, you actually need to take the 92% efficiency and multiply that by your power supplies AC => DC efficiency (since the 92% you're seeing is an AC => AC efficiency). So, .92 x .82 (your PSU's rated efficiency) is 75.4%. So, about the best you'll see since you've got poor wall power is ~75% efficiency.


Here's a link to my second article again:
http://www.anandtech.com/casecoolingpsus/showdoc.aspx?i=3413

Read the stuff on Page 3 (Efficiency Explained).


If you still don't believe me, could you tell me how you think your little magick box is measuring the efficiency of the power supply? Or what you think the efficiency is referring to? I feel like we're having communication problems (granted, message boards aren't the most fool proof way of communicating.). There are different interpretations of what we're saying and I'm trying to be as clear as possible.


Also. If I sound like a jerk, I'm sorry. (I am probably being a jerk about this). I'm just frustrated because this is one area in which I'm fairly well versed. (electronics and test equipment)


My goal is for both of us to end here knowing what answers to give in the future. (if you've got sources, I'm open for some late night reading)

 

[Mussels]

who the hell knows how my magic box measures what it does, at least i know which efficiency you're talking about here. It wasnt clear which kind you meant - i'll just call them internal and external efficiency here.

now that i get what you mean, you're saying that PSU's are more efficient at around 80% load. Using that as a general rule, you assume that a 600W PSU would waste more power, than say, a 350W, in a lower wattage PC.

My meter can contradict that - with a 350W antec in my media PC, i had 110W load. changing to a 550W antec, a newer model my load dropped to 75W, because its more efficient. going below 80% of its rated power doesnt do anything to the PSU's power draw, its going over that, and approaching its maximum that has an effect.

(I dont think we're really disagreeing too much here - we were partially talking about different things)

 

[theeldest]

I agree on us not really disagreeing.

There was another PSU article on Anandtech where they compared ATX1.3 and ATX2.0 PSUs. The 2.0 are much more efficient, even at the same rated efficiencies, because 1.3 was made for closer to 90v, and 2.0 is for closer to 120v (i think that was the reason).

It will really depend on the Antec power supplies. Realistically, I'd say the only way for a person to be completely sure will be to get the Kill A Watt and run different powersupplies. (which Antec is the newer one? I've got the Antec Earth Watts 650 (replaced an Antec SmartPower 450))

Hey, speaking of that, is anyone willing to try this in a HTPC? or Fileserver? I would be *very* appreciative. Test the same PC under full load for 1 hour on different PSUs.

 

[theeldest]

Addendum: If 3 people agree to send me old power supplies that you don't want (that have the 24 pin connector for main board), I'll buy the Kill A Watt and run the test in my server. 1 hour idle, 1 hour load.

I'll do it on up to 12 power supplies. Any takers?

 

[Mussels]

theeldest: FYI, i run on 240V.

 

[theeldest]

Well that's lucky. If you look at the top graph I posted you'll see that 240v will give quite a bit better efficiency than 110v or 90v.

 

[Katanai]

Here: http://www.shopping.hp.com/product/...hJ7XpbGpFdTvVwL5VhSX84LMDJQRJT01wW!1660157471

I know this is not cheap but if you want more from a NAS system this is the one to get.

 

[theeldest]

http://us.shuttle.com/X2700.aspx

There is another. It's the shuttle PC. Starting at $399 with an atom processor, 1GB memory, & 80GB hard drive. Get that, stick in another hard drive and it's a server. If you want more power, upgrade to the dual core Atom processor and 2GB and you're at $469.



@Katanai: That's a great suggestion ... I'm really starting to wish I had the money to buy fresh instead of using old ATX parts (that end up in a relatively huge case).