Need some reassurance on this OCZ 700w PSU choice

[GalvanizedYankee]

Originally posted by: 1N0V471V
Go for it. An awesome PSU... I love it. Here's a review I wrote (did not use DMM but look... over ten pages of how we love it)

http://www.dfi-street.com/forum/showthread.php?t=56164&page=1

OK Uncle Dave!! Most of us realize OCZ has DFI-Street by the short hairs! Bwahahaha!

OCZ owns you guys. Members at DFI-Street HAVE to run OCZ inorder to belong to the club.

Ask the resident OCZ rep over there why they use crappy OST caps on the output side. Mmmm.

To the other posters in this thread that have doubts about jonnyGURU, Read what
the man had to say about the new Mushkin units. Mushkin are being made by Topower.
OCZ switched to FSP and Mushkin stepped into OCZ's slot at Topower.

Mushkin sent the dogs of war after jonny for that review.

...Galvanized

 

[GalvanizedYankee]

Hell of a review Uncle Dave! You used MBM and tested it with a standard system, no real
load. Then you link the reviews of others and most of those use no real load tester. Hahaha!


Galvanized

 

[1N0V471V]

Hey Galvanized, I said STRAIGHT UP in the review that I didn't have/wasn't going to test with a DMM. OBVIOUSLY I know that software voltage monitoring is WAY off. But I'm 16 buddy, so I don't have and wouldn't know how to test using a DMM.

We use OCZ because they work.

 

[1N0V471V]

To best honest Galvanized I'm tired of coming here and seeing you all tell people that 500w is good for a 7900GTX SLI/Dual core system. That's bullshit

 

[Skeeedunt]

PSU aficionado fight!

 

[Pabster]

Originally posted by: 1N0V471V
To best honest Galvanized I'm tired of coming here and seeing you all tell people that 500w is good for a 7900GTX SLI/Dual core system. That's bullshit

You've got a lot to learn, young'un.

Seeing you pimp OCZ is a real treat though

 

[Bobthelost]

Originally posted by: 1N0V471V
To best honest Galvanized I'm tired of coming here and seeing you all tell people that 500w is good for a 7900GTX SLI/Dual core system. That's bullshit

No it isn't.

The lowest wattage PSU i found here: http://www.slizone.com/object/slizone2_build.html that's sli certified (NOTE: Not the same thing as saying it's the minimum, it's saying it'll work under any reasonable circumstances.) is the FSP550PLG-SLI 550W (here http://www.sparklepower.com/pdf/FSP550PLG-SLI.pdf ). That seems to be a PSU with 36A total on the 12V line (singular).

Now, that's going to be tested to work with a P4D chip, the 965EE specifically. (if you're going to certify something then you make sure it'll work for the worst case). At idle that's 40W more than any AMD or C2D option, at full stretch it's 50W more than the top end C2D and 20W more than an FX62.

These will also be done for the highest end models that are made, see above for why.

So, assuming you find yourself a PSU with 50W less on the 12V rail(s) ie 32A (and they are flexible enough to deliver the power where it's needed) and you don't have a 965EE, you're still good to go according to a bit of common sense and the nvidia certification.

But none of that is relevant. Because people do use 500W PSUs for SLI rigs. The proof of the pudding is in the eating.

Edit 2: Oh and don't forget the A8N32, possibly the most power hungry chipset ever. Nice performer but it sucks down a lot of power compared to nearly every other board around.

 

[diablofish]

Originally posted by: GalvanizedYankee

Originally posted by: 1N0V471V
Go for it. An awesome PSU... I love it. Here's a review I wrote (did not use DMM but look... over ten pages of how we love it)

http://www.dfi-street.com/forum/showthread.php?t=56164&page=1

OK Uncle Dave!! Most of us realize OCZ has DFI-Street by the short hairs! Bwahahaha!

OCZ owns you guys. Members at DFI-Street HAVE to run OCZ inorder to belong to the club.

Ask the resident OCZ rep over there why they use crappy OST caps on the output side. Mmmm.

To the other posters in this thread that have doubts about jonnyGURU, Read what
the man had to say about the new Mushkin units. Mushkin are being made by Topower.
OCZ switched to FSP and Mushkin stepped into OCZ's slot at Topower.

Mushkin sent the dogs of war after jonny for that review.

...Galvanized

No one here doubts JohnnyGURU. It wouldn't matter is God himself tested only one unit, the same statistical analysis method of extrapolating one or two results to an entire population is the one glaring weakness of ANY of the reviews that have been done. By weighing each of the reviews, a person can get a far better understanding of the product than by simply reading one review and declaring it the gospel.

Obviously testing hundreds of PSU's is likely cost prohibitive, which is where these forums can be a valuable resource in determining which products are reliable based on multiple users reviews and feedback of multiple power supplies of the same make and model.

If in doubt, my advice of buying something else still holds.

 

[1N0V471V]

What does me being young have to do with anything? I know my stuff and that's all that should matter.

Galvanized... just because I don't want people listening to bad advice doesnt mean my review sucks... or DFIs suck... or DFI Street sucks... or I'm an OCZ fanboy... or or or

I am personally recommending the 700w OCZ GameXStream because I've SEEN (In person!) it running an x1900xtx/x6800 system with drives in RAID, etc. with NO voltage fluctuation. I know it works.

 

[imported_RedStar]

as for bad caps -- the people over at badcaps can't decide either one senior member says CapXon is well reputed ...and another senior says no they are crap. teapo..hard to say if they are in there..

Contradictions are everywhere
---
so while there are a ton of great reviews on this PSU ... i am gonna leave the ultimate resolution to OCZ ...If they answer my post. exo! If they don't, then i have the answer about ripple out of spec

well..and i want to explore whether ripple really does damage video cards, hdds and the like.

edit: ty 1N0V471V. i read all 15 pages and checked the links.

 

[jonnyGURU]

Originally posted by: diablofish
It wouldn't matter is God himself tested only one unit, the same statistical analysis method of extrapolating one or two results to an entire population is the one glaring weakness of ANY of the reviews that have been done.

I'm going to hit and run here... sorry... I'm tired of the trolls around here and what I'm going to say is going to envoke flames from someone, I'm sure...

I had three units. Two OCZ's and one FSP Epsilon. All three had the same ripple results.

And (with all due respect) other review sites that test ripple do so at the PWM and not at the load, because it takes a load tester to test ripple at the load with an oscope.

Furthermore, there's a big difference between voltage regulation and ripple. Everyone seems to get these two terms, and tolerance, confused.

ATX spec says the rails can be within 5% from their mean value (5V, 12V, etc.) That has nothing to do with load regulation. So when Antec, for example, says it's power supplies have 3% regulation, that's not better than the 5% in the ATX spec. They're two different things. The 3% is load regulation and the 5% is just a tolerance specification.

If you hook up an DMM and see only a 2% fluctuation in voltage going from idle to load, that's load regulation. Ripple are the small fluctuations that happen every ms or so that may not affect performance, but may kill your components in time. Ripple can not be measured by a DMM.

I may be my own worse critic and I often doubt myself probably more than any of you guys ever do. But the fact remains that I've tested a lot of PSU's and most of them have shown minimal ripple. Much less than 50mV in most cases. Yet three examples of an FSP Epsilon based unit has shown ripple upwards of 150mV. Even if the testing methodology is somehow flawed, how do you explain that?

Have a good night, guys.

 

[SuperSix]

Originally posted by: 1N0V471V
Hey Galvanized, I said STRAIGHT UP in the review that I didn't have/wasn't going to test with a DMM. OBVIOUSLY I know that software voltage monitoring is WAY off. But I'm 16 buddy, so I don't have and wouldn't know how to test using a DMM.

If you don't have the right tools, then you shouldn't be doing reviews.

At LEAST spend $20.00 on a DMM>

What a joke

 

[imported_RedStar]

thankyou for your response jonnyGURU. A driveby is still fine by me

Indeed this is why i remain concerned ...the sample might be small ...jG was given a random sample though..and they all showed the same behaviour.

So i have been looking for a wider sample set or confirmation from OCZ...to be sure either way.

And while there are a ton of great reviews ...only one seems to have gone truly into depth about this ripple issue. One issue i was completely ignorant of. This is why i find this the most interesting aspect of this discussion.

And while i now know that ripple can hurt components ..can someone please explain how...is this a long term thing like overclocking or more short term ...like say a year?

 

[1N0V471V]

Super, I never said it was the best, most complete, to die for review in the world.

If you don't want to read the review, don't read it. Nobody is forcing you. I clearly state that I did not use a DMM. By no means is it complete, I know that. But let's be honest, I could tell you all the effing numbers in the world about flucuations, ripples, whatever. But if it doesn't power your system with good, clean power (and with ease), then those numbers don't mean one thing. Just because it looks good on paper does not apply to the real world.

This is the point of the review. 15 pages of feedback on one PSU. Nowhere in that thread does ANYBODY complain. Ever.

It's like with graphics cards. Synthetic benchmarks mean little to nothing when it comes to actually playing the game. The same principal applies here.

 

[Bobthelost]

Originally posted by: RedStar
And while i now know that ripple can hurt components ..can someone please explain how...is this a long term thing like overclocking or more short term ...like say a year?

Ripple is where the voltage goes up and down, normally in a wave shape. As such if you've got a 12V rail with 2V ripple (this is huge by the way) at some points it will be supplying 14V, at others it will be supplying only 10V.

If you put too much voltage accross components it damages them, either by heat or by more complex processes that make my head hurt. There's a great thread on it somewhere in the CPU forum if you're interested (but finding it would be a nightmare ). There are tolerances built into the chips, unfortunatly these aren't publicised.

If you put too little voltage accross componentes then they don't work properly, and your computer crashes.


Now, how important is it? Not sure. The ripple probably won't fry the parts quickly, after all some people put .2V more (in addition to ripple) through their chips for years without problems, but it's not going to be good for the parts. I can't give you a number here, i just don't know. It'll be long term and it won't help overclocking to have large Vripple.

 

[imported_RedStar]

ty, Bobthelost
GalvanizedYankee:
"OH! Some review sites check for ripple AFTER the mainboard caps have filtered
the PSU's output NOT BEFORE. Ripple should be checked at the PSU"

so if these caps on the MB filter the power, then how does ripple damage the important components?
Someone else (edit: Navid) asked the same question on this thread.

edit2: OCZ has replied:

" Tony
here to help
--------------------------------------------------------------------------------

Join Date: May 2003
Posts: 2,514

--------------------------------------------------------------------------------

We don't believe there is an issue, we have many test units as well as retail units in the field and no one has reported any issues thru ripple on any of the lines.

Out of all the reviews we have seen only one reports ripple to be outside specs... take from that what you will. "

http://www.bleedinedge.com/forum/showthread.php?p=189618#post189618

 

[jonnyGURU]

Originally posted by: Bobthelost

Ripple is where the voltage goes up and down, normally in a wave shape. As such if you've got a 12V rail with 2V ripple (this is huge by the way) at some points it will be supplying 14V, at others it will be supplying only 10V.

Errata, Bob. That would be 4V of ripple. 2V of ripple would be 2V total, either up, down or in both directions, like 11V to 13V would be 2V of ripple.

 

[Bobthelost]

The capacitors clean up the power, if you were to use smaller ones then you'd get much more ripple, if you don't have any then things go horribly wrong (think i set fire to a lab that way once, happy days). (Background information)

Now there are more capacitors and bits on the motherboard that turn the 12V power into 1.35V or whatever you need to drive your CPU, as such it's like washing something twice in different washing machines. If the second machine (motherboard) is good then it'll clean up the failings of the first machine (PSU) but you don't want to rely on this as other parts might not be so lucky (graphics card).

 

[Bobthelost]

Originally posted by: jonnyGURU

Originally posted by: Bobthelost

Ripple is where the voltage goes up and down, normally in a wave shape. As such if you've got a 12V rail with 2V ripple (this is huge by the way) at some points it will be supplying 14V, at others it will be supplying only 10V.

Errata, Bob. That would be 4V of ripple. 2V of ripple would be 2V total, either up, down or in both directions, like 11V to 13V would be 2V of ripple.

You're probably right on that, i could check my notes but since there's a good chance i'd be shown wrong i'll take your word for it. Just one thing i'd like to say:

I'd have gotten away with it if it weren't for you darned kids!

*waves fist as police drag him away*

 

[Adul]

Why not this

or this

 

[imported_RedStar]

well my store i will be getting my computer from is canadacomputers

so..i have only so many choices.



btw: it only took me about 20 min to pick an enermax 350 5 years ago.

now with sli --the whole power supply issue has become much more complicated.

But at the very least ..this is all very educational )

edit: Adul..your second choice has 1% ripple!

 

[SuperSix]

Originally posted by: 1N0V471V
If you don't want to read the review, don't read it. Nobody is forcing you.

Believe me, I won't be reading it - it's useless, all opinion with no valid data. Why would I want to waste my time?

There's enough opinions - I want facts.

 

[GalvanizedYankee]

RedStar. small CapXon are used in Zippy/Emacs PSUs, the bigger guys are UCC and RubyCon. Teapo are OK for PSU use but don't fair well on mainboards. The big complaint about them is they fail with no visinle sign. Dirty current from the PSU works board caps harder than clean current and can/will shorten board cap life. Thank god and green apples that caps normally fail in the open state, not shorted.

Spouting off half truths, by any of us, don't get it

Read this thread in total. http://www.badcaps.net/forum/showthread.php?t=1659
Yes it's a sound card....a very expensive sound card. Bwahahaha! trodas got his first soldering iron at 6, he is now in his early 30's. He knows!


...Galvanized

 

[imported_RedStar]

heh..there is no way i have the skill to replace capacitators. But i am surprised that expensive X-fi card is using bad caps.

And i now understand the link between alleged bad caps and ripple.

Searching in the proc forum for old ripple posts ...i did find a link to an anandtech review (2003) that stated bad ripple turned out to be the leading cause of ram failure .

So if i understand this correctly, the caps in the PSU filter the ripple ...bad caps in the PSU thus lead to more ripple.

But i still don't have a time scale ...as we all know overclocking shortens the life of components..but it takes more time than we care about

 

[Bobthelost]

You're about right there yes. Bad capacitors might also explode on you, which is obviously a bad thing.

Ripple won't do the parts any good but it's negligable as long as it's under 50mV. More than that isn't a great sign.