oc'd e6600 and 8800GTX powered from 380W PSU - why not?

[RollingThunder]

it should have been just enough... or not. really can't say. don't have another 380w psu to play with.

still, i wonder, how many psu's out there have "true" 2x12v rails.

about that voltmeter i found. it seems to be working. not sure how accurate it is and since it's analog it's pretty pointless but just for giggles - where would be the best place to measure 12v on the psu?

MC,

Check with one of your molex connectors on the 12v side yellow wire. You can check your 5v too using the red side. Make sure you have good batteries in your voltmeter.

I think checking your 12v side under that load would be very interesting and another good evaluation of that model power supply.

 

[muhon_zen]

buy a brand new 9V duracell battery and measure it. Then you'll see how calibrated is your voltmeter (they should always measure (9V+-1%) without load.

 

[mcoleg]

Thunder, thanks man. i've done just that, burebista helped me with it.

the voltmeter is analog so there's not much precision, you understand. i also had to calibrate it against ocz520 with a known value of 12v rail. i have set it to just above 12v some time ago, now it came in handy.

all that aside, i didn't notice any large droop under load. very tiny movement on the dial when the load kicked in and then it stayed there.

i've been running the lost planet dx10 demo's test for several hours. the load from the wall was up to 330w, peaking close to 340 once or twice.

if i find a deal on a digital multimeter somewhere, i'd get one, just to know what it is exactly i am reading but as it stands, no noticeable problems.

 

[mcoleg]

muhon, good point.

still, since there's not much resolution, using this particular meter is more academic then anything. i'll show you what i mean:

 

[muhon_zen]

I'd get a radioshack 80$ digital multi. Those are good, have RS232 port for "PC Oscilloscope" which will let you record and make an average on the computer of the measurements. Also, they don't have true RMS, but if you're only measuring AC power (which is sine wave-formed) then you will have accurate readings below 1KHz.

 

[mcoleg]

connect it to the pc? that sounds pretty cool. i gotta look into it.

funny though, that would cost more than the psu in question

 

[muhon_zen]

Well, yeah xD. But it would help you very much in the future. AC measurements up to 1500V, DC to 1000V if I'm not mistaken. The RS232 (Serial) interface really helps, I'm not sure of the total bandwidth, but it will help you see, also, how your PWM's are performing and to check the ripple voltage on the 12V rails under load.

It also has continuity test (to check if cables are properly soldered/paired/joined), hFE for BJTs, Temperature measurements (not sure about accuracy here, and you need to purchase a separate thermocouple), diode voltage, logic hi-lo state, current measurement (fuse for I < 40mA and < 10A). I think there's a wattage option too, but haven't tested.

I would like to know about your Kill a watt. Where did you get it? Is it cheap? I want one :3

 

[mcoleg]

wow, that sounds like a lot that thing does. i'll check one out in my local rs next week.

the kill-a-watt is not too expensive. i got mine for $22 at amazon:

http://www.amazon.com/P3-Internatio...3?ie=UTF8&s=electronics&qid=1179582781&sr=8-1

there's a similar gadget from seasonic, angel-something. pretty much the same thing.

 

[funnyperson1]

If you just want to take basic measurements, this is a great/affordable multimeter:
http://elexp.com/tst_my64.htm

 

[mcoleg]

hmm, that one doesn't look too bad either. thanks for the tip funny.

 

[mcoleg]

update:

day 3. it's still alive .


what i have done so far:

1. disabled c1 and speedstep. that increased the load on idle by about 10-15w and on load by about 5-7w.

there's less varience on the power the system drows from the wall - that's probably what i was seeing, c1 trying to reduce power consumption and multiplier jumping between 12 and 6.

2. i've ran the lost planet demo for close to 18 hours, on and off. full screen at 2560x1600 it averages in low 20's and in a window 1920x1200 it's at 40 fps.

looks like my system has finally met it's match . what kind of monstrosity one need to build to play it with a reasonable fps? i swear...

3. i went to my local radio shack and got the multimiter muhon recommeded. it was $65 and i just couldn't resist the lure of connecting something to something else... i know it's more than i need and i am not going to use 90% of it's functions and i probably just wasted about $30 for no good reason. well, i am a guy, can't help it.

so, the voltages. 12.12v in desktop, typing. the lowest i got running the lost planet was 11.98v.

the highest temperature for psu i recorded at 99.5F. that was after running the test for 8 hours straight. still quiet. keep in mind though - it's on the bench, not inside the case so it gets cool air to work with.

here's the voltages on 12v rail:

edit: attached voltages in txt file.

View attachment mv-lostplanet.txt

 

[muhon_zen]

Damn that's a good PSU.

 

[mcoleg]

surprisingly, it is... thanks for the recommendation muhon, that's one tricked out meter .

 

[Drew@PSU]

True story, I want one...maybe I can pass it off as a work expense and have them pay for it
Good stuff mcoleg, glad to see someone dedicated enough and in the right position to run a test like this. Maybe you can deflate the Watt inflation factor in the industry! But definitely keep us posted, I may not get the 750 silencer, although its a pimp looking PSU

 

[mcoleg]

thanks Drew.

it's a prof of a concept more than anything else at this point. i do like my 750 quad and i mean a lot, it's a really good piece of machinery. it's just now it looks like i could have gotten away with something much smaller.

i haven't realized that i should put some numbers together and see what my system really needs untill recently.

i do hope the experiment is a success. i need to make sure the psu keeps working for at least several days without blowing the cups before i can recommed it with no hesitation.


edit:

update:

on burebista's recomendation i've posted a link to this thread on silentpcreview forums:

http://www.silentpcreview.com/forums/viewtopic.php?p=342167&highlight=#342167

 

[RollingThunder]

thanks Drew.

it's a prof of a concept more than anything else at this point. i do like my 750 quad and i mean a lot, it's a really good piece of machinery. it's just now it looks like i could have gotten away with something much smaller.

i haven't realized that i should put some numbers together and see what my system really needs untill recently.

i do hope the experiment is a success. i need to make sure the psu keeps working for at least several days without blowing the cups before i can recommed it with no hesitation.


edit:

update:

on burebista's recomendation i've posted a link to this thread on silentpcreview forums.

MC,

An interesting thread to say the least. Without having seeing one before, how does your Radio Shack multimeter connect? What did you say you wattage consumption was while running this test? I am impressed by the steady voltage this power supply put out.

The only issue I have a question with is how well this power supply or any power supply will last running at near peak capacity with day to day use. I know you can't answer that and don't expect anyone can without leaving it run for however long it would take. Hypothetically, I relate it to running your car flat out versus at 50 mph. It has to take a toll over time.

Most of us here, including me, realize that the online wattage calculators are just plain wrong and the assumed perception of a power supply needed for a new build are exaggerated. I know the Corsair I have probably exceeds my needs by close to double.

Nevertheless, a most interesting series of posts and nice photos. Thank you.

 

[mcoleg]

Thunder, the power from the wall was at 340w max. it stays on load on average at 330+/- now that i disabled c1.

good point about the wear and tear. who knows, really...
i am going to run high power consuming tests on it for at least a week.
i could probably connect something else to it to increase the load, a fan or two, m/b.

any requests? i can run a benchmark or a test someone is interested in.

"how does your Radio Shack multimeter connect?" - do you mean how it's connected to the pc? if so, it's through a serial port. i know. had to install one just for this

 

[RollingThunder]

Thunder, the power from the wall was at 340w max. it stays on load on average at 330+/- now that i disabled c1.

good point about the wear and tear. who knows, really...
i am going to run high power consuming tests on it for at least a week.
i could probably connect something else to it to increase the load, a fan or two, m/b.

any requests? i can run a benchmark or a test someone is interested in.

"how does your Radio Shack multimeter connect?" - do you mean how it's connected to the pc? if so, it's through a serial port. i know. had to install one just for this

MC,

So you're series of tests run the PS at about 87% (330w) capacity if efficiency is 100%. That's interesting. And even if the power supply gives out after a week of hard tests, it still says a lot of good things about it.

I know that if Muhon's friend can bring him home one of these EW 380s from the USA, he will be very happy with it!

 

[funnyperson1]

MC,

An interesting thread to say the least. Without having seeing one before, how does your Radio Shack multimeter connect? What did you say you wattage consumption was while running this test? I am impressed by the steady voltage this power supply put out.

The only issue I have a question with is how well this power supply or any power supply will last running at near peak capacity with day to day use. I know you can't answer that and don't expect anyone can without leaving it run for however long it would take. Hypothetically, I relate it to running your car flat out versus at 50 mph. It has to take a toll over time.

Most of us here, including me, realize that the online wattage calculators are just plain wrong and the assumed perception of a power supply needed for a new build are exaggerated. I know the Corsair I have probably exceeds my needs by close to double.

Nevertheless, a most interesting series of posts and nice photos. Thank you.

I know that analogy is used often in this scenario (and I have used it myself), but where it falls apart is in that an engine is a mechanical being with moving parts, gaskets crack, oil gets sludge, there is no real equivalent to that with a power supply.

As long as all the components are within temperature spec, the first thing to fail should be the fan. The only other parts that might fail before the fan are the electrolytic capacitors.

The rest of the chips, mosfets, transformers, etc. should be fine.

At the same time with 340W max and ~80% efficiency, his system is only using ~270W of power so he's not even pushing it really. Not to mention that at idle he was only using 250W from the wall which means that most of the time he should be around 50% of the PSU's rated capacity.

It still is nice to know that there is some breathing room though, and its fine to overcompensate a bit, but I think that xomething like your Corsair 520 will be good enough for 98% of the people on this forum.

 

[mcoleg]

well, actually it's 330w from the wall. with 80% efficiency that's 264w the system is actually consuming. that means that the psu is only operating at ruffly 70% of it's rated capacity. it's not really as simple as that, since we have to calculate rail loads separately but it's a good enough approximation, i think.

sorry if i didn't make it clear - i just record what kill-a-watt shows, i don't convert it to the actual power consumed by the system.


edit: funny, you beat me to it