PSU tier list 2.0

Page 152 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Why don't you make a list of all the reviewed units you can find, and then take every unit and identify what caps are used for the primary and secondaries and then list it up so you'll have a quick reference of exactly what caps are used by model number of the unit. You might also include a category for the fan model used on the unit as well, which could include specs on noise as a sub-note.

Basically, a good and bad caps by PSU model number list. That should keep you busy for a while and it would serve a purpose.
 
That might be a good idea. But I'm not going to do any "good" or 'bad". I am going to simply list the model, and then list the cap brands found in there. Nothing else. I cannot bring anything subjective into it because people get mad then. If I simply list the cap manufacturers, it is just pure raw factual information.
 
I used to program a lot, but have not programmed for nigh many many months. I don't care for high-level languages, I just want to eventually learn Assembly and call it a day (or a year, as it'll take a while).

I'll do this cap list. Sounds fun. I'm fine with repetitive. At work the other day, I stacked containers and handed them to another guy for 8 hours. kind of liked it.
 
Well those aren't really the worst of them- a woman called squintskii disregards everything I say about CrapXons in the CX's being a lot more terrible than the Chemi-cons in the S12II. All probably because she's biased towards my young age.

What a cruel woman.
 
Am I the only one who thinks it is extremely stupid that rated capacitor lifespan is an actual variable in calculating lifespan of a capacitor? I mean, c'mon, any company can lie about a capacitor's rated lifepsan easily. I don't understand why this is a value that can be trusted, especially from the Chinese brands. It is just as bad as rated wattage. Companies lie about that, so why do people not expect them to lie about capacitor lifespan? This is almost why I think brand judgement is a justification when it comes to capacitors. You can do all the math, but in the end if the capacitor company has a sucky electrolyte from the incomplete formula stolen in 2005, or if they just build cheap crap and lie, then that company should be avoided.

We should never rely on a company's word.
 
It takes a lot more manufacturing gear to make a capacitor than to make a PSU - or worse, take someone else's PSU and slap a new label on it. That means companies are going to try much harder not to get a bad name.

Only the very worst of capacitor manufacturers are outright lying - most of the capacitor messes were caused by trying to clone a competitor's product (like that only happens in the third-rate manufacturers...) and copying the datasheet rather than doing the testing themselves.

I would be very surprised if you found massive deviations from specs on genuine products - note that there's still probably a load of re-wrapped caps, but they'd be of the expensive brands anyway.
 
Too bad there's no practical way to start a PSU failure database. Have site visitors document the make/model, age, damage to other components, etc. whenever they experience a PSU failure. Talk about unbiased empirical data. I guess even this would be flawed because of things like market share or regional availability. I tend to base a lot of my personal preferences regarding PSU's on my own experience with failures I've seen, sometimes disregarding respected opinions or reviews by those more knowledgeable than me. The truth for me is that I see certain brands fail frequently and others I never or have rarely seen failures from.
 
Like you said, though, there is so much stuff that needs taken into consideration, like how many units were sold, and what volume of people whose units break come to online forums for help. I kind of want to start telling people on this forum to send me their dead PSUs so I can look inside and see what happened. Also, with dead units, how much the user used the computer and at what level of stress plays a major role.

What data may be better is how many warranties were issued out / units sold, which would be info from the company.
 
I know this forum isn't any professional source of statistics, but just browsing through the forums every day, and seeing all the instable system / broken PSU threads, they are always ones using Chinese caps. I have seen some G2 problem threads, but most of them seem to be just a manufacturing fault. I'm talking about "My system has been running fine for 6 months, but now when i play intense games my computer freezes up [or shuts down]". They always almost have a PSU with the Chinese brands, so I am not personally ready to trust them. I don't see these threads with EVGA G2 units, or Corsair RMx. I see it with Corsair RM, a lot of CX (but a lot of those sell, probably more than any other unit).

Even non-PSU issues, such as other hardware being broken, usually there is a bad PSU paired with that system. People seem to underestimate not only how easily cheap PSUs can fail, but how poor voltage regulation, ripple, and noise can affect the lifespan of other components of the computer.
 


And no company is going to give out that kind of data. Actually that brings to mind an idea if you wanted a boring repetitive task, lol. With thousands of posts regarding PSU failures, you could mine them for PSU failure data. It may not be a very scientific analysis, but you could compare the number of failures of a brand to the total number of posts and start to draw a picture of which brands have more reported problems.
 
Okay, I'm going to make this a discussion thread on Tomshardware, kind of like how this thread has become, where anybody could pop in and post a link to a thread involving a PSU failure or PSU contribution to the failure of another hardware. Then, we will record this data.

But the question is how are we ging to divide up the data? Chinese caps Jap caps? Do you think that is the easiest way? Or should there be other ways to divide up the data?

I was thinking of listing total failures, and then how many of those failures are from units with Chinese vs Japanese caps. But then again some units use a mixture of both, and then there is the issue of units with Chinese caps usually falling short in other, non-capacitor aspects of performance. So I need to figure out a way where this can be a fun community-based statistic center that, after some months, will merit interesting data.
 
Or (sorry for making so many posts) I could have multiple statistical sections. I could have a good vreg vs bad vreg section. Good ripple vs bad ripple section (keeping it simple). I would have exact values that make these units good or bad, based on reviews. Chinese vs Japanese section.
 
Once you have a database, you could probably draw all kinds of different data points from it, but I think you would run into problems trying to be too specific about individual components. Models may have design changes and use different components or even OEM's and it would be hard to tell without knowing when they were manufactured. I was thinking in a broader sense, like the percentage of failures of a certain brand or model compared to the overall number of reported failures. Even this would start to paint a picture of relative quality. For example, if model "A" had a reported failure 100 units and there were 10,000 total posts, you could infer there was a 1 percent failure rate. You could then compare it to model "B" with 50 reported failures for the same amount of posts for a failure rate of .5 percent.
 
Yes, JG works for Corsair. Has for several years.

So you're saying that if you replace the caps in a random cheapo Chinese unit, suddenly it won't have any long-term degradation issues? Silicon can fail under overload/excessive heat too. Bridge rectifiers and the switching transistors being the common ones.

Correlation does not imply causation. You can't say that the PSUs with Chinese caps failing over time means it's actually the chinese caps that are failing, and not the fact that the mfr cheaped out on other stuff like heatsinking, correctly sized components etc. At least not unless you open lots up and find bulged caps.
 
If the silicon is allowed to overload, then it's the end user's fault for overloading their PSU, in addition to the OEM's fault for not implementing proper protections. How often do we see CPUs fail? I don't think I've seen one failure ever of that silicon.

It's just a statistic, that is all. As it says in the first paragraph, it is in no way a perfect statistic. Doing all that stuff will get nowhere, just recording simple stuff is keeping it simple.
 
Over-overclock a CPU and you will start to see logic errors. Do it too much, and settings that were previously stable are no longer. The excessive current and voltage starts to eat away at the actual transistors.

Same goes in discrete parts. Too much current, or too much voltage, and they can fail in all sorts of interesting ways.

I'm not saying the user is overloading the PSU. I'm saying that the PSU may be rated for '500W', and able to deliver 500W for hours, but they might not have parts rated for high enough voltage ($$$) so being exposed to power is slowly wearing them down. So over weeks/months they don't switch as fast, or they get higher resistance, and that all leads to reduced capacity.

https://www.fairchildsemi.com/application-notes/AN/AN-9067.pdf
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom