Best PSUs (Archive)

  • Thread starter Thread starter Guest
  • Start date Start date
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
The 'performance' given to the power supply benchmarks do not state which benchmarks hold what weight in performance. How are we to gauge the tests if this hasn't been stated? Moreover, performance per dollar looks at the price with a base point of 0, which does not take into account the starting point of the power supplies does not begin at 0.

How seriously can the reader take 'performance per dollar' when both have these issues?
 
The tests (they are not called benchmarks in PSUs) are the ones that we conduct in each PSU review. The algorithm is made by me and I also got some input from some highly educated reviewers with huge experience on such stuff. It is optimized to reflect all major aspects of a PSU.

It is based on the assumption that 100 points are the perfect score and it deducts points for the following (except the last):
1) Load Regulation
2) Ripple Suppression
3) Advanced Transient Response
4) Turn On Transient Test
5) Overall efficiency with around 25,000 different load combinations
6) Hold-up time and Power Good Hold-up time (I use the lower value)
7) Capacity

The total score is 100 - (1) - (2) - (3) - (4) - (5) - (6) + (7)

In the case of the hold-up time, this factor can also have negative values if it exceeds the requirements of the ATX spec so it can actually add points to the total score instead of deducting. Since it is a complex algo I can say the Load regulation is weighted for 1.2 at +12V, 0.6 on the minor rails and 0.3 at 5VSB. Ripple which has higher values goes for 0.07 at +12V, 0.05 on the minor rails and 0.025 at 5VSB.

This algo's description, or at least a version of it, is provided in another site where I do PSU reviews but for obvious reasons I cannot give the link here. Thing is that I am working on it for quite some time and in Tom's I use an enhanced and more accurate version of it. For example in the other site I take as the efficiency score the 20%, 50% and 100% average while in Tom's algo I take the Overall efficiency score which is more accurate. I was able to do this because I started from zero at Tom's so I could make all the changes that I wanted in the algo without worrying about a huge database that would be incompatible afterwards.

As for the performance per dollar yes it starts at 0 but zero isn't the base point!!! How did you figured this out? The base point is the performance per dollar score of the AX1500i which currently registers the highest overall performance. So the AX1500i has 100% and everything else is compared with it.

Please, next time before you make any allegations on things that you simply cannot understand better ask first. I am here to solve every query (if I can of course).
 


We only listed the PSUs that we reviewed in the last couple of months and this Seasonic is quite old.
 
Not clear how the DarkPower 860 can be considered superior to the Corsair AX860. From the warranty all the way across the rest of the board.
 


Interesting, but it seems to me a little that the capacity doesn't really belong in there. The capacity is something that is usually considered entirely separately from the performance question, since it's basically just a number that determines whether or not it can handle the specific load of your configuration. Either it can, or it can't. For example, if my load is only 200 watts, it presents no advantage to me to buy a 1000 watt PSU instead of a 450 watt PSU. It makes the performance value fairly meaningless. What really matters is how well will it handle the load it's designed to handle, right?
 
capacity matters because load regulation is measured from 20W to full load. This means that it is unfair for a 1 kW PSU with 1% load regulation to be directly compared to a 500W PSU which scores the same load regulation. The operation range of the first PSU is double as much compared to the second.

If I didn't include capacity into the equation the low capacity PSUs would dominate the scores, since it is easier for them to have tight load regulation, low ripple suppression and higher efficiency (since as the load increases the energy losses increase as well).
 


Hmm. While I see your point, the same argument could be used against including capacity, since it means that excess capacity is lowering the bar for performance at a specific load. Or from the other direction, for a given load, excess capacity represents a compromise in performance. For example, a hypothetical 800 watt PSU and a 400 watt PSU that receive the same overall score would mean that the 400 watt PSU should actually perform far better under a 300 watt load than the 800 watt PSU would, unless I'm misunderstanding something.

Perhaps your weighting also compensates for this somehow, or makes this point negligible, but I appreciate you clarifying and discussing it!
 
I also appreciate your input in this matter! I take every user opinion seriously and I keep my ears open for suggestions or constructive input.
 
if they send me one to test I will include it in the graphs. However so far I haven't evaluated a very low capacity PSU for Tom's.
 
Evga 550w G2 vrs Evga 550w GS. Same price point, give or take a few bucks, but the G2 outperforms the GS in every department including noise, protections and warranty. Failing to see how the Seasonic powered GS rates as a best 550w and editors pick over the SuperFlower powered G2.
 
yes the overall relative performance of the 550 G2 is higher than the performance of the 550 GS. Sorry I wanted to point out this but I wrote the opposite instead.
 
Not a problem, although I do have another question, on priorities. If, as you say, the hold up time on the G2 is long, and the GS has power good past rail capabilities, wouldn't it be of much lesser importance for the unit to take longer to power up than for a unit to over load it's rails since the only protection covering the rails is overpower and at full load this can theoretically be higher amperage than what the rails are rated at?
 


The hold up time for the G2 is too short. Is it even possible to have a hold up time too long?
 
I didn't quite get your question so I will try to sort some things out before I reply.

The hold-up time of the G2 is much lower than the hold-up time of the GS (550W both).
In addition, the hold-up time doesn't have anything to do with the time that a PSU takes to power up. It has to do with how long it will keep its rails in spec in a sudden power loss.

All PSUs have over power protection, at least the ones with APFC which are the 99.5% of the branded PSUs that cost above 30-40$. Now a good question is how high this protection is set. Some have it at 110%, some 120% and some probably at 150% or even higher. This can change though from revision to revision easily and in general I prefer to educate my readers that it is a really bad idea to overpower a PSU and personally I don't like blowing stuff like some other guys do. With the conditions that I apply in my tests (45-50C ambient) nothing would survive. This is why I only test at 110%. To cut to the chase, if I included into my rating OPP then I would have to tax as fail every PSU that delivers more than its max-rated capacity. Especially the PSUs that claim 50C operating temperature.

For me the most important protection is OTP. However I chose not to add such characteristics in my relative performance algo but stick only with the rough data that I get from the measurements. Someone could ask to add also the bearing of the fan, or the capacitors' quality which are of immense importance or the warranty period. However this would result in a really complex algo which could easily deliver not accurate results. Taking everything into account and expressing this with an algo is a task easier said than done.

BUT I do take into account all factors that describe a PSU's performance, quality, reliability etc. in my final rating. I don't provide though a final rating at Tom's since it is this site's policy but I do elsewhere.

A.

 
Ok then. Now you got me totally confused. 'it has to do with how long it'll keep its rails in spec during a sudden power loss'. So you are saying that with a sudden power loss, the rails need to stay in spec to accommodate capacitance discharge? I'm failing to see how this is of such major importance that it disqualifies a psu from 'editors choice' when measured up against a psu that doesn't by your own admission perform as well, is physically much louder, has a 5yr vrs 7yr warranty, more protections. (OVP, UVP, OCP, SCP, OPP) etc. In every category the Evga G2 tops the GS except for holdup time. And to be honest, with advances in technology, design, build criteria, many have pointed out that ATX specs are somewhat outdated and often unrealistic in expectations.

I guess, in this instance anyways, I'll have to disagree with the conclusion that the Evga 550w GS is in anyway a better gaming psu than the G2. Sorry.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom