PSU tier list 2.0

Page 199 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.


And are you looking at the scoring page or the review? I don't see anything that would put this in the same class as many of the other tier 3 units. One thing the chipped off corner of the PCB shows is that this is not a cherrypicked unit, unlike many units from some other companies that are most likely cherrypicked (Seasonic, I'm looking at you). It's also not something that you'd probably see on other PCBs of the same unit. If he had gotten a unit without the chip, nobody would even know about it.

So besides the chipped corner and sleeve bearing fan, the chipped corner not really doing any harm, what makes this unit on the level of the EVGA 430W and 500W? I can think of three significantly better things: DC-DC and higher quality caps. Better ripple. Modularity.
 
I still think a review should exist before tiering. You cannot tier a unit just based off knowing the OEM, talk about welcoming criticism. Just like the EVGA NEX B. We know it is FSP, but haven't seen a review. We also cannot make assumptions that it is like the NEX G, they are completely different units and could be a completely different platform.

If there is no review it should be in the untiered section.
 



But that's where the fun stops, because the voltage regulation wasn't anywhere near excellent. At 3% average in the hot box, the word "average" definitely applies here today.

But the big deduction is, once again, the connector interchangeability between the PCI-E and CPU cables. You can plug them in wrong, and that's bad because they're not compatible.

If ever you needed more confirmation that those modular connectors should not be plugged in where they're not supposed to be, this shot is that confirmation. No way are the PCI-E and CPU connectors electrically compatible. A good design gets around that by making the connectors incompatible. This one, like the Thermaltake before it, does not do that.

The soldering gets a half point deduction too.

By the way, I don't see any vsense functionality in here at all. The unit probably does well enough before power gets into the wires, but has no way to adjust the voltage depending on load because it can't really monitor things where it needs to. This is the kind of thing you pay Seasonic extra money for.

So, I guess it's not terrible, but for the price, it could never be a good choice when there are so many others in that price range that are excellent.

 


There ARE reviews of those units, just not good or in depth ones. Considering that factoid, I'd have to maybe agree that that unit should have been un-tiered, but maybe DT has a good reason for it. Then again, it could be an oops.
 
The tier list does not tier based on price, though. If that was the case tier 1 would only have a few units. The 12V voltage regulation was pretty good, and progressive load tests for the most part are a lot less important than things like crossloading tests and transient response. For the most part, when people are not gaming and not doing anything intense, the load of the computer stays around the same general vicinity. Once they start gaming, there is a brief moment where it does go up but then it stays in that same vicinity again. Having 3% or more regulation on the 3.3V and 5V rails is no biggy.

There seems to be an NEX B review at a site called hitechlegion but that's just one of those pointless "reviews" that simply restate the specifications.
 
I don't think a psu should get judged/grading on whether it's modular or not. A non modular psu does not pretend to be modular so it should not get counted off for not being. It should be judged for what it is IMO.
 


It's still a lot better nonetheless.
 
I agree (with CT); we've commented on the modularity thing before, and also whether or not there's a Berg connector. While good to point out whether the PSU has these features, neither is something that, imho, should influence a rating. They're features; either you care about them or you don't. There's no rating scale, nor is it something about which the vendor can lie ("oops" on a marketing blurb, maybe; lie, not really; that makes it a QC issue more than one of dishonesty).
 
So what feature should and should not be considered for tier placement? The problem is that some of those features that are considered are unimportant to a good deal of users, while some of those things that aren't considered are more important. For example, should the packaging be considered in the tiering process? I can see the Rosewill Photon being tier 3 if the package is taken into account since it just comes with a think Styrofoam sheet underneath it to protect it. Though that's not strictly about the PSU itself, it is about how well it can handle rough shipping.

What about a user whose computer will only subject a PSU to about 50% of its capabilities, but the tiering process revolves around how the PSU did at 100% load? I think if the tier list is to be improved upon, what units are judged by should be strictly stated in the first post. But I'm not Dotorrent so I don't know what he looked at exactly. But even if you do list what things are and aren't tiered upon, certain features have more weight to some users than others. Like most people don't need super good ripple unless you want to do extreme overclocking. On the other hand, many people probably want a unit that is quiet.
 
Bergs are easy. If it's a gaming style psu, designed for modern day pc's, it shouldn't have a berg, that's a useless addition. On the other hand, if it's a psu designed to be a replacement for OEM, then a berg makes sense. There's quite a few older systems still around, especially in small business, that simply don't need an upgrade. My wife is a government employee and the software she uses is Windows 7 compatible. Not 8.1. Best bet is actually to use XP or 2000. So a berg? Sure, could be needed. Points off for having one? Sure. If it's useless, it just shows that the architect hasn't a clue. Could have saved time and money per unit by not having one there
 
The berg is fine if it is on its own adapter on a modular PSU. A non-modular consumer-oriented PSU should not have one. Other things might want to be taken into consideration also with tiering. The cables themselves. How is their quality? Are they stealth ribbon cables which are less restrictive on airflow, or are they all sleeved? (not talking about each wire, I'm talking about a sleeved cable). You also have units that throw capacitors in the cables like the G2, and while it's pretty and all for ripple, it can make the cables a but more stiff and lumpy.

Or just other things like distances apart with the SATA headers. On my 850 G2, the SATA headers are stretched a long way to reach the drives, and I wish there was a bit more distance between them. They are stretched to the maximum basically.
 
Only non-Boolean features may be rated, although there could be exceptions even there, such as your current favorite spec, hold-up time. Under some conditions, e.g. reliable power source or UPS, it doesn't matter; under others, it could arguably lead to equipment damage which is a Howling Big Deal.
This is why, as useful as it can be, the listing is not (and no reference can be) a be-all, end-all. You need a decision tree of some sort, in which an assigned tier is only one factor. How do you include the ifs, ands, and buts in a one-dimensional tier list? You can't; you need something like "Tier-1, BUT hold-up time is only 12ms" or "Tier-3, AND it's modular."

Edit: And, if the Berg connector bothers you that much, snip the damn thing off and call it a day. Otherwise its presence or absence only implies some wasted money, and cost already isn't a factor in tiering (though it is in recommendations).
 
Wouldn't it be nice if there was some kind of interactive website where people entered information like what they do with their computer, their specs, ambient temps, etc. and their budget and it automatically would generate a list of recommended PSUs based on their needs and by checking current availability at sites like Newegg, Amazon, etc. and using data to calculate the best price/performance PSUs for them?
 


You know, as soon as I put those thoughts onto text and made the post, I immediately started cycling PHP code through my head on how I could actually do such a thing possibly. This could be kind of fun actually, it's been quite some time since I've had a web project to work on. Wow. I think I might be onto something really. It'll require a lot of time and a lot of information garnering and inputting it all into databases, but if I start with a small amount of PSUs and test out some code I can maybe make this thing work! :)

I'm picturing some sort of weight system. So if the person fills out that they like a quiet PSU, that'll be a higher consideration when it generates the recommended list. Or if they don't care about noise, it'll pretty much ignore how loud the PSU is altogether, henceforth not recommending a unit based on something the purchaser doesn't need.

I think I'll start with 10 PSUs and a simple questionare.
 
You know... If you started with an existing wattage calculator and revamped it with more realistic values (don't forget to get permissions) 1/2 the battle would be over with. Even if you didn't use it, at least notating exactly how they went about the code and what wattage values they assigned the separate criteria might be a good starting point. You'll have to remain impartial, but 3 choices would be budget, better, best

I've got a couple of Lga1155's, so I'll Guinea pig.

Might want to get ahold of the guys at pcpartpicker and link up through them since they do already have a similar system in place that covers a few different areas like Australia, Canada etc. India would be good too.

Sorry, I tend to think big when faced with a good idea.
 
I was just comparing warranty periods of both the EVGA and SeaSonic power supplies lines. I just realized that the EVGA Nex650G comes with a 10 year parts and labor warranty; the same as my EVGA P2 850. Is it just me or is it that strange that two different lines would have identical warranties?
 
Nothing strange about it. Evga just has faith in its components and build being high enough quality to last at least 10 years. The warranty is set by Evga and Seasonic, no matter what OEM happens to build it. Most psus come with the same warranties, be it 3yrs, 5yrs, 7yrs or 10
 


FSP. These days, the 3.3V and 5V rails will have a maximum load in a normal system of about 2-3A. Crossloading is more important than ever. Not to mention if you want compatibility with the C6/C7 sleep states (though not that disabling it is a hassle or anything).

Just look at the 12V rail in test CL2 on the NEX 650G http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=442 it is a very realistic load. Even if the 3.3V and 5V rails have a tiny bit more load, it'll still be floating in the 11.47V-11.60V territory most likely.
 
Status
Not open for further replies.