- Thread Starter
- #181
Thunder - i can see how it can be taken as an endorsement, easy. on the other hand, it's hardly a danger - i do want to believe that at least most people who read articles in places like overclockers.com can make up their own minds based on fact and research. so, i don't think i've convinced the populace to rush and buy puny psu's to run their monster rigs yet
now, the experts. don't take me wrong, i like big and powerful things as much as the next guy but lets face it, the question of longevity of smaller psu's is not answered. so far no one came up with hard data or calculations to support their claim except Neur0mancer (good work there man, i still want to see more data though, if you can).
i'd like to believe the experts but only if they have proof. numbers, graphs, statistics - that's what i want to see.
as for the criticism, that could have been a problem if i were a sensitive type . i don't care, the more people talk about a subject, the more we learn.
Techno - yes, good point about the rails. it's a bit more complicated though, breakers and all. i was planning to paste GlasMan's explanation here, thanks for reminding me.
here it is, GlasMan's of pcper.com nice insight into how rails in psu work:
now, the experts. don't take me wrong, i like big and powerful things as much as the next guy but lets face it, the question of longevity of smaller psu's is not answered. so far no one came up with hard data or calculations to support their claim except Neur0mancer (good work there man, i still want to see more data though, if you can).
i'd like to believe the experts but only if they have proof. numbers, graphs, statistics - that's what i want to see.
as for the criticism, that could have been a problem if i were a sensitive type . i don't care, the more people talk about a subject, the more we learn.
Techno - yes, good point about the rails. it's a bit more complicated though, breakers and all. i was planning to paste GlasMan's explanation here, thanks for reminding me.
here it is, GlasMan's of pcper.com nice insight into how rails in psu work:
Almost all have one 12v source very few have them seperately regulated, but each rail is amperage limited with a circuit breaker that will shut down the psu if you draw too much. This hodgepodge was developed to meet the ATX 2.0 and 2.1 spec. .Some recent psu's quietly let out that they are not actually current limited, but I would have to go with their published specs when trying to choose one. With the end of the ATX 20 amp limit, I hope to see more flexible ratings of the rail(s). It also seems that sometimes the only difference between 2 psu's of the same brand is the amperage limit/rail (see the TT 1200 and 1000 for one example). On a multi rail psu that didn't have them limited, the 1000 would be be more useful then the rating of a 1200 (an example using the #'s, not directly related to the TT's)
Another improvement I hope to see is information of what connectors are linked to which 12v rail.
Research Silverstone, I believe their top psu had a problem 2 pci-e connectors with one on the cpu rail so a 2nd card would overload it, while a 3rd rail was limited to regular molex's. Since corrected.
Many of the high watt psu's devote a rail per pci-e connector, so like the cpu rail, you have lost half the output. On a load tester they can load the rail to max but in real life you would have to splice in another connector to fully load the rail. This isn't tested in reviews, mostly because it is extremely hard to get a computer to fully stress the rails (you need 4 gpu's to get close).
Of course people are now used to drastically increasing power requirements due to 12v cpu's, gpu development, and sli, plus quad cores upcoming. Overkill is the word if you want your investment to last. With split rails rated 12v wattage levels have climbed dramatically, much higher than the actual amount needed, indeed faster in fact then a computer can be built to use it* and in the US more than house wiring can deliver (2000 watt psu's soon for 220v regions) *I am going to try though with my "Folding Farm in a Box" 8 cores and at least 4 gpu's.
With accurate info on how the power is distributed (for instance, which rail does the 12v delivered through the 24 pin delivered too? Why have the 8pin etx connector on a seprate rail than the 4 pin atx?) and on your power requirements you can work out which psu meets your needs with a safety margin without paying $300-500 for one that can do everything, while making up to 50% of the output unusable, more unused.
Capacitor life is effected by temp, I believe 18%/10C, not load directly.
Silent pc'ers want to have their psu at maximum efficiency at their normal power levels. Less waste heat means lower fan speeds to get rid of the excess heat. Over clockers on the other hand want to increase power needs beyond the best voltage regulation limits of such a psu and doesn't care about the heat or waste so haven't thought about it. It is only recently that it has been given any thought. There is a lot of useful information at sites like silentpcreview.com. They are really trying to solve many of the same problems, just causing them for different reasons. Many overclockers seem a bit stuffy to the uninitiated as well.