Power Supply Review

[MixMasta]

I found it interesting that Antec is using Seasonic to build some of its units....

That said, Super Nade's comments about the article hits the nail on the head. The only reason I would read a power supply review is to find out if its junk or not. How do I know if its junk or not if I can't see the inside of the unit, and nor is it hardly mentioned? Honestly, it seems like they let anyone write an article at Anandtech. I love all the awkward sentences and repeated statements too. They'd be smart to link to the badcaps forum...

For your reading pleasure (or displeasure):
http://www.anandtech.com/casecooling...oc.aspx?i=3146

[Super Nade]

Re: Power Supply Review

I was being merely being polite

With the expensive equipment they have, they can really step it up a notch and appeal to a more technical audience without sacrificing their six-pack base. Articles can be long if they are well organized. I'd rather read a detailed article like Gabe Torres does at Hardware-secrets, Jonny of course or Paul does at [H] rather than have to skim through an incomplete one like Anand's.

The effort is there, but if you have the means, why not step it up a notch?

[gonzo0815]

Re: Power Supply Review

May be there is some revenue from web marketing at risk, if a test would be too detailed.
IF there would be honest discussion, how many people would really need all those 500w+ PSU`s ?
I know, that there are systems where a real 500w+ PSu would not hurt or may be even be mandatory, but compared with the even small enthusiast fraction, there would be not that much systems, who would fit into this category.

Personally, i have given up to relay on such reviews, unless i can see a picture of the innards and can do my own conclusions.
But ok, may be it was only due to the lacking interest of the great unwashed mass.

[999999999]

Re: Power Supply Review

The problem with web reviews is they cannot address one of our primary concerns which is premature failure. We can look back and find common corner-cutting measures like bad caps and fans which won't last the life of the system but do last for longer period than a review test period will uncover.

I don't mind very much if a 500W PSU has bad ripple when it outputs over 400W for example, I would never select a PSU within 80% rating of the actual load. I don't care if it has neon disco sheathed clear hyperbladed fan modular chromed textured voice activated caller ID with a front panel LCD water cooled overclock dial and a kickstand all inside a colorful box with a built-in handle. It just has to run for the life of the system only being touched after it's ran long enough that there's dust buildup to be cleaned out.

PSU reviewers just don't understand the needs of the majority. I only find reviews useful to see good internal pictures and to weed out units so grossly overrated that their labels are pure fiction.

[PCBONEZ]

Re: Power Supply Review

How many times have you seen a web review comment on how pretty the box it came in is.

W.T.F. is that?
(Besides padding the page with filler and wasting my time.)
.

[Brian C]

Re: Power Supply Review

I suggest we (badcaps.net) do the review. That must be interesting ! Is that possible ?

[gonzo0815]

Re: Power Supply Review

I think Jonny Guru does already a very good job in doing this. But sadly it is probably not possible to just checkout any funny PSu brand out in the world.

But i would like to see more low power to mid range test too, and i would not worry to build an office system with a 80% capacity utilisation.
There is no need for overrating in a system which is intended to last like it is and then be replaced.
Sure, a good quality unit is mandatory for such a system.

[999999999]

Re: Power Supply Review

I think Jonny Guru does already a very good job in doing this. But sadly it is probably not possible to just checkout any funny PSu brand out in the world.

Jonny Guru does provide a lot of useful information, but a lot of the conclusions drawn by readers, as well as his own, are subject to interpretation. Plus, some tests just aren't reasonable, for example if a PSU has a significantly larger % of it's transformer capacity and filtering devoted to high 5V current and the spec'd capacity accurately reflects this, it is not so reasonable to only do the same cookie-cutter load tests which subject it to cross-loading scenarios which are mismatched to that particular PSU.

In other words, some people do need PSU with higher 5V or 3.3V current capability and showing a test where a unit fails when most of the watts are coming from 12V rail, makes a PSU look bad and yet would not be an appropriate match for a system placing such a load on it, so the test is irrelevant and simultaneously suggesting a PSU should achieve a target system's needs instead of meeting it's own specifications. IMO that is backwards, what we really need is to know if a PSU meets it's spec'd %/rail output then we can make our own decisions among units that can do this, whether they are a well matched current:rail for a particular system.

But i would like to see more low power to mid range test too, and i would not worry to build an office system with a 80% capacity utilisation.
There is no need for overrating in a system which is intended to last like it is and then be replaced.
Sure, a good quality unit is mandatory for such a system.

There are more low to mid PSU tests on Jonny's site than there used to be, if you consider units like an Allied 500W to be low-end.

I can't agree that even an office system should have a PSU with less than 20% margin. For example, suppose that office system used 160W, would you really put a 200W PSU in it? Normally if an OEM, shop or self-builder, builds a system with this consumption they put at least 300W PSU in it. When an OEM builds a mATX box with mATX PSU it typically has less than 120W consumption kept low with integrated video and only one HDD, etc. The other problem with only 20% margin is in recovery time, that today more than ever a video card or CPU power state change is very fast and needs faster recovery time from the PSU than ever before. Couple that with ACPI CPU idling and it means that a 10 year old system would be better running from a PSU with 20% margin than a modern one would be.

Additionally, many systems won't even POST if the consumption is within 20% margin, as the startup current is just too high. Perhaps we are thinking of two different concepts of consumption, if you were considering the peak startup current as the value against which the 80% derating or 20% margin applies then I would agree that an office system could run ok from such a PSU if the design had the right current:rail.

[Spectre]

Re: Power Supply Review

How many times have you seen a web review comment on how pretty the box it came in is.

W.T.F. is that?
(Besides padding the page with filler and wasting my time.)
.

These are retail items so the packaging and contents get written up as well, and quite frankly the boxes can often be rather entertaining. Like saying a powersupply is less than 16dBA at full load....rrrriiigggghhtttt

[Oklahoma Wolf]

Re: Power Supply Review

Plus, some tests just aren't reasonable, for example if a PSU has a significantly larger % of it's transformer capacity and filtering devoted to high 5V current and the spec'd capacity accurately reflects this, it is not so reasonable to only do the same cookie-cutter load tests which subject it to cross-loading scenarios which are mismatched to that particular PSU.

The only recent review we've done that I can think of that you could be referring to is the bargain basement roundup. The load tests employed were chosen on purpose to show what happens to these dirt cheap, older designs when the ordinary uninformed Joe Consumer decides to trust the unit that came with his case to run a modern build, which is inevitably a 12V based system. And just as inevitably, these cheap power supplies are old group regulated 5V based designs.

In other words, rather than choose loads appropriate to the power supply on that one, we decided to go all noob and test them like we knew nothing about choosing a power supply

[PCBONEZ]

Re: Power Supply Review

With all the different ATX spec varieties floating around now for both motherboards and PSU's the mother board manufactures need to start listing the power requirements for their boards rail by rail like is done for PSU's.
-
The old way "minimum ____ watt power supply" doesn't cut it anymore.

I have two Intel socket 478/533FSB boards that both say a ATX12V PSU is appropriate but one needs a full 5 amps more on the 5v than the other.

Really need to check the power requirements rail by rail now days but the info on MoBo's isn't usually provided.

How do MoBo manufacturers expect people to use appropriate PSU's when they don't provided the needed information?

.

[gonzo0815]

Re: Power Supply Review

Manufacturer don`t care much about this. And to be honest, i think it is not that hard to get the different concepts of the different ATX specs.
If you nail it down, then there are only two specs who really mater: a 5 & 3,3v centric one and a 12v oriented.

Regarding the 80% utilization, i have seen a lot of Medion systems with 3ghz p4, ATI9800 and 250w FSP MDN PSU. RAdio &TV card was included too.
And if they had not sacrificed the cooling due to noise constrains and had used better caps they would have last probably more then 3 years.

With utilization i mean the continued power with any component working at it`s limit (as far as this is possible).
For most duties this equals out to a far lower then 80% utilization.

The start up current is defined and limited in the ATX spec`s, so an system with well designed components would not have a single issue with a high utilization.
And if you put a lot of HDD`s into a rig, i think it is obvious that you have to take this into account, as ATX spec does not cover anything beyond "usual" systems.