1,200W

[Brett Thomas]

Better hope your circuit breaker doesn't have too much other stuff on it - triple 30" LCDs and a couple NAS boxes, you'd trip the circuit breaker pretty quick.

I didn't even think of the monitor...Rob does use a 30" monitor, which is NOT on that Kill-A-Watt... haha, congratulations, Rob! You've probably used more power in one test than your entire apartment building combined!

 

[Kayden]

Somebody asked me on another forum once - "Well, if I get a more efficient PSU, will my room get colder? Because I use my system to heat my room in the winter."

I laughed, I'd never thought of that. It works pretty well as a heater.

Mine is nick named the Heat Master 4 Thousand, because I need no heater in the winter for my room just more game time to keep me warm. (c;

 

[Rob Williams]

Thanks a ton for taking the time to post, George

Those are some interesting equations... I had no idea the temperature being generated could be calculated so easily. If the HX1000W produces 198W of heat and the AX1200 produces 108W, does that literally mean that the HX1000W is running <em>about</em> twice as warm? Or does it not quite scale like that? Would there be any reason for alarm to have a PSU producing ~200W heat like that?

I like the numbers there from the AX1200 though, since that PSU will be replacing the HX1000 in our next GPU testing build. Even with the HX1000W, though, there wouldn't be much reason for concern PSU-wise, right? I am not well-versed on power, so I am not even sure what the top-limit for a socket is (I think 1800W, but I might be off), and as I'm in an apartment, I am not exactly sure if there's a dedicated 1800W to that socket, or if it's shared among others.

If I drew 1150W off that PC alone, then the monitor probably added another 100W on top of it. I guess there's still a fair amount of breathing room there. I need to make sure things are straight though, because I'd <em>like</em> to actually benchmark such a configuration in the future, but can't if it risks burning the house down

Also, since I'm not familiar with yields on power supplies, would you say that the vast majority of units that get released would perfectly match the bell curve you posted, or is there sometimes a 1 or 2% variance?

I also hate to be the one to ask what I'm sure is the most ignorant question you've ever heard, but if the HX1000 was handling only 950W, why was ~1150W being drawn from the wall socket? I can understand efficiency and how that works, but I am still not sure how the PSU could be drawing less power but the wall socket not benefiting from it. To me it's kind of like an odometer reading 100MPH while your car may only literally be going 80MPH.

Thanks again for the post!

 

[marfig]

Indeed. CorsairGeorge and everyone else.
I made heaps of copy/pastes of this thread to OneNote for later perusal. An area I know very little is an area worth learning.

 

[RainMotorsports]

Thanks George I thought I was going to have to draw out another explanation in detail after crimson went all silly on me.

Ill say it once and say it again the Killawatt reads consumption and is not the same as the PSU's output. It should be noted since we are always guessing off the total consumption to calculate the output and because of this we could be maxing one of the rails out say the 12v rail and not necessarily know it. This wouldn't be a good thing but drives spinning down, fans going wonky and bsod's are some of the usual signs. Cant say I have max'd a psu out in years.

Always nice to have a tech rep come out especially related to their own product.

 

[CorsairGeorge]

Thanks a ton for taking the time to post, George

Those are some interesting equations... I had no idea the temperature being generated could be calculated so easily. If the HX1000W produces 198W of heat and the AX1200 produces 108W, does that literally mean that the HX1000W is running <em>about</em> twice as warm? Or does it not quite scale like that? Would there be any reason for alarm to have a PSU producing ~200W heat like that?

Assuming the numbers I posted earlier, yeah, the heat being generated by the HX1000 would be almost twice as much. 200W is a lot of heat, but it's not unheard of for PSUs. Back in the day when the average PSU was 70% efficient, drawing 500W from a PSU would require around 700W from the wall. This is is a lot of heat, but unlike in a CPU or GPU, it's not all tied to one tiny surface, it's spread out over multiple devices. Also, the components inside the PSU are rated for much higher temps. For example, on our TX/HX/AX lines, the capacitors are all guaranteed at 105C, and resistors and other components could go higher than that even, and still be in-spec. So it's not like CPUs where you're struggling to cool a 200W overclocked processor down to 60C or something.

I like the numbers there from the AX1200 though, since that PSU will be replacing the HX1000 in our next GPU testing build. Even with the HX1000W, though, there wouldn't be much reason for concern PSU-wise, right? I am not well-versed on power, so I am not even sure what the top-limit for a socket is (I think 1800W, but I might be off), and as I'm in an apartment, I am not exactly sure if there's a dedicated 1800W to that socket, or if it's shared among others.

Most US sockets are 110V, (well, 110V-115V). Also, most circuit breakers are 15A. So 15A * 110-115V is 1650W-1725W. That's not for one electrical outlet, but an entire circuit. In the house I live in, the fridge, microwave, stove, oven, and air conditioner are all on the same circuit because they're all on the same side of the house and near the same room. Typically circuits are broken up by large usage rooms (laundry or kitchen) or multiple mid/low usage rooms. A couple of bedrooms, a classical family room (before we all had surround sound and a 65" TV with five separate boxes plugged into it), that sort of thing.

So when you look at those 1500W PSUs, keep in mind that even if they are 90% efficient at 100% load, they'd still be drawing over 1665W from the wall to produce 1500W output. Which means that in most homes, that'll trip the circuit breaker before you even count the power consumption from your wi-fi router, NAS box, printer, monitor, speakers, phone charger, or whatever else you have plugged into that circuit. So realistically the max PSU you're going to see for sale in the US is a 1200-1300W, and even those are pushing the boundaries. In other countries, it's different. Most of Europe is 240V, and even if a house has a 10A circuit, that's 2400W of available power. Plus, PSUs are much more efficient on high-line, 240V.

If I drew 1150W off that PC alone, then the monitor probably added another 100W on top of it. I guess there's still a fair amount of breathing room there. I need to make sure things are straight though, because I'd <em>like</em> to actually benchmark such a configuration in the future, but can't if it risks burning the house down

You won't burn anything down. The circuit breaker will trip and you'll go to the panel and flip it back. I do this probably 2-3 times a week at my house during the summer, because, like I said earlier - some genius put the AC unit on the kitchen circuit which means whenever you decide to, say, microwave a burrito and run the AC at the same time, I get to go outside and flip a switch.

Also, since I'm not familiar with yields on power supplies, would you say that the vast majority of units that get released would perfectly match the bell curve you posted, or is there sometimes a 1 or 2% variance?

Good question. There's always a variance due to component variance, manufacturing, etc. But it's usually not that high. A good example is our HX750 and HX850 PSUs. We tested a bunch of them internally and they're borderline silver/gold certified. For example, let's say that 60% of them pass the gold spec, and the other 40% are high silver. We sent units to the 80PLUS.org people to gauge certification, and they said "Congrats, it's gold!" Legally, we could have sold them as a gold certified platform. But because of the variance, and because a lot of people wouldn't end up with gold units, we market and sell them as Silver certified.

I also hate to be the one to ask what I'm sure is the most ignorant question you've ever heard, but if the HX1000 was handling only 950W, why was ~1150W being drawn from the wall socket? I can understand efficiency and how that works, but I am still not sure how the PSU could be drawing less power but the wall socket not benefiting from it. To me it's kind of like an odometer reading 100MPH while your car may only literally be going 80MPH.

Thanks again for the post!

Let's use an analogy. You've got two kids, Tom and Huck, with two 1 gallon buckets each, and they've each got to fill a 100 gallon bathtub. They have to walk to the hose, fill up the buckets, and then walk back and dump all the water in the tub.

100% efficiency would mean 50 trips (100 gallon tub divided by 2 gallons per trip).

But realistically, those kids can't carry 2 gallons without spilling a bit. And also, it's a long walk and it's hot out, so some of the water will evaporate.

Assuming they walk the same speed (voltage from the wall), the math is fairly easy.

So let's say Tom is 90% efficient. He is very careful and doesn't stumble. So for every trip, out of the 2 gallons he starts with, he delivers 1.8 gallons to the tub. 100 gallons / 1.8 gallons per trip = 55.55 trips, but let's call it 56. Those 56 trips is how many times it took him to go back and forth to fill the tub. Since he loses .2 gallons per trip, and he made 56 trips, he's lost 11.2 gallons of water (heat).

But Huck is only 80% efficient. He's kind of lazy, and the buckets are heavy, so he doesn't control them well. Out of the 2 gallons he starts with, he delivers 1.6 gallons to the tub. 100 gallons / 1.6 gallons = 62.5 trips, but let's call that 63. That means he's got to go to the hose and back again 7 more times than Tom does. Since he loses .4 gallons per trip, and he made 63 trips, he's lost 25.2 gallons of water (heat).

Power supplies work the same way. You have a power load (the tub) that is on the output side, and a hose over on the input side. The PSUs efficiency is how well it converts wall (AC) power to computer (DC) power. A really efficient PSU can produce the same output power as a really inefficient PSU with less trips (power from the wall).

 

[marfig]

Quick! Someone grab CorsairGeorge and don't let him leave these boards ever again!

 

[Doomsday]

Quick! Someone grab CorsairGeorge and don't let him leave these boards ever again!

*Tries to grab to George*

Damn! Hes too fast with those Dominators!!