Is an expensive PSU all that - or is it some crap everyone bought onto?

[DEredita]

I been using generic PSU's for at least the last 11 or 12 years and I never once had a problem. Now everyone is like "oh don't skimp on the PSU - that's bad..."
Why? What is a better PSU going to do - it's not going to make you system run faster, it maybe quieter (but I don't care much), and they are expensive.

I've been getting people saying things like:

i'd stay away from high voltage generic psu's. they tend to blow up

The only time I've seen this happen to anyone is they were running a 3200+, 1+ GB ram, multiple hard drives, 6800GT Ultra, as well as other things like one of those audidgy cards - and their generic 350 watt PSU fried.

I feel if it is enough to run the system i.e. 450watts, then it should be fine.

What is the big thing with PSU's?

Am I missing something? I'm protected by a couple of those power strips with the surge protectors. Shouldn't that be enough? My system now has been going strong for 3 years now at about 12 - 18 hours a day and experienced brown outs, surges, and black outs while the system has been running, and only the system shutdown, I never had any components fried or get damaged.

Please explain what the deal is?

- Mike

 

[DEredita]

This is the generic PSU that I plan on using:

http://images10.newegg.com/pro...mage/11-105-113-09.JPG

What is wrong with this?

- Mike

 

[lobadobadingdong]

not much complaint here, better psu's tend to use higher quality components, giving you cleaner DC signals with less variance of voltages and less noise. (way better for OC'ing) many "cheapo" PSU's rate thier wattage incorrectly by measuring power in AC peak-peak instead of the AC rms or DC output. in which case you buy a 400W psu and it's really a 302W psu.

It's kind of a get what you pay for, you could buy a ford that runs fine with no problems and has descent performance, or you could get lexus, that runs smoother/longer, and has more features.

 

[akira34]

These days, most people that buy one of the 'generic' PSU's soon lives to regret it. I've seen generic PSU's die just after a year of service in a range of environments. I've seen quality PSU's last over 4 years in a range of environments as well. While you MIGHT get lucky with a generic PSU, you have a much better chance of having a name brand, known quality, PSU lasting for several years.

Personally, I wouldn't get a generic PSU, since the cost difference between the cheap-assed generic and quality is far less than what it would cost to replace the hardware killed by the generic PSU WHEN it fails (not if).

Also, the generic PSU's tend to be light on Amps on the 12V rails compared to quality units. Getting a 400+W PSU with less than 18A (or 22A) on the 12V rail is not a good idea. It also tells of how poor the quality really is when it can't deliver enough Amps to the line. Quality/branded PSU's also tend to have additional features that the generic's don't (another way they keep the cost down).

I tend to get PSU's from just a handful of makers that I've had good results from over the years. I also won't get a PSU from a company that hasn't been making them for at least a few years (rules out many of the newer PSU's coming into the market). I'll wait and see how they do over a 3-5 year time frame and see what they get for a reputation.

 

[Operandi]

A generic PSU is about 10x as likely to fail over the course of 1-2 years, never live up to their ratted power, and tend to loud. Buy generic if you want but you'll get burned eventually.

BTW, lobadobadingdong you analogy is flawed. My 98 SHO has 98,000 miles on it and still performs great .

 

[akira34]

Originally posted by: Operandi
A generic PSU is about 10x as likely to fail over the course of 1-2 years, never live up to their ratted power, and tend to loud. Buy generic if you want but you'll get burned eventually.

BTW, lobadobadingdong you analogy is flawed. My 98 SHO has 98,000 miles on it and still performs great .

:thumbsup:

 

[lobadobadingdong]

Originally posted by: Operandi
A generic PSU is about 10x as likely to fail over the course of 1-2 years, never live up to their ratted power, and tend to loud. Buy generic if you want but you'll get burned eventually.

BTW, lobadobadingdong you analogy is flawed. My 98 SHO has 98,000 miles on it and still performs great .

not really, I've had several fords that still run well after 200,000+ miles, but that's besides the point, the lexas's that I've been in with over 200,000 miles still run smoother and quieter, and generally the interiors hold up better (even under messy drivers). they both work, it's all about what you care more about. the lexas (or other high end automaker) has higher quality components and testing than ford (or anyother "generic" automaker)

 

[jose]

It's your money you get what you pay for, but don't complain when it fries everything in your system..

Personally I'd never put a crappy generic ps in a computer w/ thousands of dollars of computer hardware..

Regards,
Jose

 

[drpootums]

What akira said about the amps on the 12v rail is the usual reason people buy better ones, but also the whole "when the crappy PSU goes so does your 1000+ dollars in equipment" thing usually has a big role in it too.

 

[Degrador]

The main reason I suggest for people to buy a good quality psu is the tighter rail voltages. More than once I've heard of people's $500 graphics cards dieing on them for seemingly no reason, until they discover their 12V rail is running at 13V. Parts are more likely to last longer if the voltages are clean and stable. A better car analogy is that you've bought the new lexus, and now decide to pump cheap ethanol mixed fuel into it - sure, the car will still run, but I'd shudder at what it's doing to the engine. Simple thing is, why skimp out on what makes it run?

 

[fixxxer0]

The main reason I'd say its not all "hype" is because if you read the boards a lot ( i know it says i joined 2 weeks ago but ive been reading on here for years), look how many problems are solved by switching out a PSU to a higher quality one. I'm not even mentioning the problems CAUSED by a cheapo one!

I think for the $100 bucks or whatever, its worth it. I have bought a $12 power supply to put in someones POS p3, but I wouldn't do it in anything someone expects to perform/last long/OC.

 

[BonzaiDuck]

I'll share my personal experience. Didn't have time to read ALL the response posts, but didn't find anything there to disagree with.

I've used "generic" PSUs since I started "building" in 1998. One went dead after about a year -- thankfully, it was not just a bad fan, either, because that would have caused my other components to fry.

I bought Enermax replacements -- a 350 watt and a 400 watt. After about three years, the Enermax 400 crapped out on me. And this time, my motherboard was toast.

For this recent build based on last-year's technology -- DDR500, OC'd 3.0C Northwood, dual RAID SATA150 and nVidia FX5950 Ultra graphics with an ASUS P4P800 mobo -- I picked a 500 watt PSU. I debated -- for weeks -- getting the Turbo-Cool 510. I was counting my pennies in the "computer-budget", because I had some other unforeseen computer costs that sidetracked the main project. I read some reviews for the ALLIED (Apex-FoxConn) AL-B500E, and bought one for $85.

Keep in mind I also had a 500W RAIDMAX I picked up for $50 atthe Pomoma Computer Fair. By the LOOKS of it, it was a "good deal", but the 3.3V rail was way out of tolerance -- at 2.92V. I eventually sent that one to the junk-yard.

The ALLIED, and a couple ThermalTake PurePower 480's I'd purchased for my family's "Christmas computers" seemed all tuned perfectly. That is, the rails didn't vary by more than 1% -- in one case it may have been 1.5%.

Eventually, I started over-clocking the current build-project with the ALLIED. With a 2.4C processor, I didn't notice a change in the 12V rail, which had settled in at something like 11.888V. When I replaced it with a 3.0C and OC'ed it to 3.67 Ghz, the 12V rail voltage dropped as low as 11.5V, although it hovered, on average, around 11.61V. I couldn't push the OC settings higher without experiencing errors in PRIME95.

With that kind of voltage drop, I settled on an OCZ PowerStream 520. I can push the CPU speed as high as 3.8 Ghz, and while it may not pass muster under PRIME95's torture-test, the voltage doesn't waver. I have it adjusted to run at an average of about 12.06V. It occasionally drops as low as 11.997V, but that's about it. I didn't spend the $210 on the Turbo Cool, but I got what I needed for $130.

And -- I figure I "broke even" on the prospect of buying the Turbo Cool. I got two power supplies for that amount of money, and the better of the two does as much as I'd want the Turbo Cool to do.

 

[Insomniak]

My rationale is like this: You just spent a ton of cash to make an awesome system...why jeopordize it by skimping on the PSU when another 30 or 40 bucks can give you much better peace of mind that your expensive hardware is not in danger? Sure, a generic PSU may work for you, but are you willing to bet ~$1000 to ~$3000 on that for 1 - 3 years?

Because when you buy a generic PSU, that's essentially what you're doing....gambling, and at the same time, decreasing the odds in your favor.

 

[JonathanYoung]

I notice a lot of people saying cheap PSUs "tend to do this" and "tend to do that." Does anyone actually have data/experience to back it up with? Personally, I think it's all hype. How do people know that the PSUs aren't made in the same factory with the same parts? Some people wanted to use cars as an analogy. Don't they make some Toyotas and Lexus in the same factories? A lot of the "evidence" suggesting that cheap PSUs are crap is anecdotal, uses circular logic, and has absolutely zero empirical data to back it up. I mean, what kind of PSUs do you think big companies like Dell, Compaq, HP, and the others use? They use generic cheapo PSUs. If generic cheapo PSUs were really as bad as people hype them up to be, those big OEMs would be in big trouble. Same thing with mom and pop stores. They often use the cheapest of the cheap components.

Now, if you're an enthusiast using all namebrand components to build your PC, then obviously you'd want to use a namebrand PSU, because all your other components are namebrand. But if you're buying the latest ECS combo from Fry's or something, then it doesn't really matter. Yes, I've seen cheap PSUs fail, but I've also seen Enermaxs fail, and I've also seen cheap PSUs power servers with 5 9s of uptime. So, unless someone comes up with a *true* empirical study of PSU reliability (instead of a 5 minute test by a tech website using MBM), don't believe everything you read.

 

[Spike]

Originally posted by: JonathanYoung
I notice a lot of people saying cheap PSUs "tend to do this" and "tend to do that." Does anyone actually have data/experience to back it up with? Personally, I think it's all hype. How do people know that the PSUs aren't made in the same factory with the same parts? Some people wanted to use cars as an analogy. Don't they make some Toyotas and Lexus in the same factories? A lot of the "evidence" suggesting that cheap PSUs are crap is anecdotal, uses circular logic, and has absolutely zero empirical data to back it up. I mean, what kind of PSUs do you think big companies like Dell, Compaq, HP, and the others use? They use generic cheapo PSUs. If generic cheapo PSUs were really as bad as people hype them up to be, those big OEMs would be in big trouble. Same thing with mom and pop stores. They often use the cheapest of the cheap components.

Now, if you're an enthusiast using all namebrand components to build your PC, then obviously you'd want to use a namebrand PSU, because all your other components are namebrand. But if you're buying the latest ECS combo from Fry's or something, then it doesn't really matter. Yes, I've seen cheap PSUs fail, but I've also seen Enermaxs fail, and I've also seen cheap PSUs power servers with 5 9s of uptime. So, unless someone comes up with a *true* empirical study of PSU reliability (instead of a 5 minute test by a tech website using MBM), don't believe everything you read.

I tend to stick with middle of the road or better brand PSU's, though that is just a personal preference. I have run computers for 4+ years on $15 PSU's with overclocking and 24/7 operation. I don't know what the voltages where as I was not monitoring such things, but I never had any crashing or power problems.

People tend to harp on generic when the truth is, most (probably somewhere over 90%) of computers have generic OEM PSU's in them. Sure, some fail, some don't, but I am willing to bet the ratio of generic PSU's failing vs Name-Brand is probably shocking as there are so MANY more generic than name-brand, but this does not translate to them all being bad. To go back to the car example, it's like saying more Ford Explorers have problems then Hummer H2's. It's kinda like a "Well duh" situation as there are WAY more explorers than H2's on the road.

Personally, I feel you don't really have any excuse not to spend at least $40-50 on a decent PSU and if you have the money, might as well spend a little more and get even better. Generics are fine for most people and perform quite well despite claims here but since you are an Anandtech reader, you are not "most people".

-spike

 

[Mucker]

My experience has taught me to stay with what works. Until recently, I have always used Antec PSU's and they have always served me well. I have always had solid rails with these PSU's. The size of a PSU's heat sinks is a dead giveaway to quality. Normally a PSU with bigger sinks will provide more power because heat is removed more efficiently, and the thermal protections do not kick in. They will also be less noisy if the fans are thermally controlled and that means the fan bearings will turn fewer rpm's and will most likely last longer. The wiring and internal layout is also a giveaway. Higher quality PSU's are generally laid out nicely providing unimpeded airflow, lower quality ones can be like birds nests. Heat removal is critical for optimizing a PSU's power output.

I recently thought I'd give another brand a try and bought one of those nice and shiny brand new sealed 500w Ultra-Xconnect PSU's ($50 Fry's deal) and in my experience, the thing was a major POS. Very tiny heat sinks made the fans work hard and the noise was like a vacuum cleaner. The voltages were also too high for my taste, 12.71 on the +12v, 3.68 on the +3.3v. The design was slick but all first versions have kinks to be worked out. Ultra's next version of X-connect should be much better, that is, if they are listening to user concerns. I didn't keep the first version Ultra thank goodness.

I am now back to using an Antec 550 True Control that I bought off the "for sale trade" board. The rails won't budge. They are rock solid and right on the money. The 80mm exhaust fan was chirping a bit and I'll guess that is why the seller was giving it up. I fan modded it for $6 and the chirp is gone, but I got a very good deal on it, $80. Now it is like new condition and rock steady. I'm very happy with it.

There is no question for me regarding the difference in quality bewteen PSU's, I absolutely think you get what you pay for here......my best advice is to read many reviews, the good as well as the bad before your purchase........

m

 

[bradley]

Originally posted by: JonathanYoung
I notice a lot of people saying cheap PSUs "tend to do this" and "tend to do that." Does anyone actually have data/experience to back it up with? Personally, I think it's all hype. How do people know that the PSUs aren't made in the same factory with the same parts? Some people wanted to use cars as an analogy. Don't they make some Toyotas and Lexus in the same factories? A lot of the "evidence" suggesting that cheap PSUs are crap is anecdotal, uses circular logic, and has absolutely zero empirical data to back it up. I mean, what kind of PSUs do you think big companies like Dell, Compaq, HP, and the others use? They use generic cheapo PSUs. If generic cheapo PSUs were really as bad as people hype them up to be, those big OEMs would be in big trouble. Same thing with mom and pop stores. They often use the cheapest of the cheap components.

Most people immediately blame software for computer problems, and never suspect their cheap powersupply as the culprit. Case in point, my friend was unable to run this one software program until we changed out his powersupply. We were able to duplicate these results in two separate computers.

Quality powersupplies are always going to have better capacitors (meaning less voltage fluctuation) better regulators (meaning cleaner power) higher efficiency (meaning less heat generated, and less power consumed) and better designs and layouts (meaning less failures due to heat related stress.) Heat and voltage fluctuations are the major killers of computer components, period.

Bottom line: the most important thing you could do for your computer is start with a high-quality powersupply, and motherboard.

 

[cuomoj]

Text

My Chiefmax 450w lasted almost two months before burning out and nearly taking my entire system with it. The case it came with has also been a MAJOR source of headaches.. Use with care.

J.

 

[echow87]

lets see....a generic psu cost $20-$30
a fortron 530w (very top notch psu) cost $60 + $8 shipping = $68

the price difference is not much plus the rails are dead on

 

[wkwong]

a quick check you can do is check the weight of the PSU. although this isn't a really good way to judge a PSU, you can tell if it is really cheap if the thing only weighs 2 or 3 pounds. i picked up a busted MGE PSU from my friends case and it weighed next to nothing. my antec PSU easily weighs 5 to 6 pounds.

i'm not saying a heavy psu is a good one, but it tells you something about the cheap parts (or lack thereof) manufacturers use to make cheaper PSUs.

 

[Zap]

Originally posted by: Operandi
My 98 SHO has 98,000 miles on it and still performs great .

:thumbsup: My '98 Contour SVT still runs great. Too bad Ford canned the Contours in favor of the Escape.

A better car analogy is that you've bought the new lexus, and now decide to pump cheap ethanol mixed fuel into it - sure, the car will still run, but I'd shudder at what it's doing to the engine.

IIRC ethanol is fine for most engines up to 10% or so without retuning. With retuning, most engines can run on pure ethanol. Methanol can be corrosive and isn't good for engines.

Is an expensive PSU all that

Back to the topic... Expensive power supplies are just expensive. Crap power supplies can be crap. What I get are the good yet reasonably priced power supplies. Fortron units are good ones. Don't even have to buy the 500W+ units, 350-400W of quality power is enough for most systems. There are plenty of other "brands" that are just rebranded Fortron units. I think many of those have model numbers that match Fortron numbers starting with "FSP." Sometimes cases will come with those "FSP" labeled power supplies. I've also had a case (I think my Compucase) come with a Heroichi power supply. I got the case years ago from Newegg for pretty cheap and just by looking at it the PSU doesn't look too special, but Heroichi is supposed to be decent. Antec SmartPower seem pretty decent. No dedicated 12v rail like their TruePower series, but if you buy an Antec case, may as well use the PSU it came with.

I also like TTGI "SuperFlower" power supplies. Love the one I got from Directron for $30. 350W, 14cm fan, quiet, 22A on 12v.

 

[Salvador]

I'm soured on Antec power supplies. I've had a couple fail me recently and so has a friend. I think that there are better power supplies out there for the money and I don't think that I'll be buying Antec again.

Sal

 

[BonzaiDuck]

It is true -- no truer words were ever said -- that "you get what you pay for." It is also true that consumers who are rational should have multiple objectives: "Get what it is you want or need", "Get it at the lowest price possible", and "Be a market force that pressures sellers/manufacturers to become more and more efficient and price-competitive."

If you are building a PC for your 10-year-old niece to use with her schoolwork, and possibly play a few games that are not terribly demanding on a PC, the system is not going to be "over-clocked". Same for a system that someone needs primarily to run spreadsheets and accounting software. In these cases, you could probably settle for a PSU in the $45 to $80 range. You could also do a little investigation to determine the odds of getting a "bad" one or a unit that had rails that showed more than a 2% deviation from the spec.

Some people here on these forums have suggested that the ThermalTake Purepower PSU's are worthless, but I've used three so far for machines that are not going to be over-clocked -- machines for business and only casual gaming. The rails are all within 0.05V of the spec, which in the worst case means a 1.5% deviation from the 3.3V rail requirement. And that's "three out of three." Those are $65 PSU's.

The question then remains -- "How long will they last?" or "If they don't last indefinitely, what is the chance they may cause damage to another component?"

You cannot be sure of the answer to these questions with a low-end PSU, but many low-end PSU's ship in OEM computers, and many low-end PSU's have been known to last past the obsolescence threshold of the computer.

Advice? Buy low-end -- take some chances; buy high-end -- at least try to be a miser.

 

[acx]

I have a Antec True430W in my rig with a GeForce 6800 Ultra and 5 HDD and a DVD burner on a DFI LanParty 250Gb with 2 sticks of Ballistix. At post, my +12V is at 11.3V. Running prime95 drops the rail down to 10.8V. Running a 3D game drops my +12V rail down to 10.5V!!! :Q

Luckily I am getting a Enermax EG701AX this week. Hopefully no more random system crashes!

 

[slurmsmackenzie]

Originally posted by: lobadobadingdong

Originally posted by: Operandi
A generic PSU is about 10x as likely to fail over the course of 1-2 years, never live up to their ratted power, and tend to loud. Buy generic if you want but you'll get burned eventually.

BTW, lobadobadingdong you analogy is flawed. My 98 SHO has 98,000 miles on it and still performs great .

not really, I've had several fords that still run well after 200,000+ miles, but that's besides the point, the lexas's that I've been in with over 200,000 miles still run smoother and quieter, and generally the interiors hold up better (even under messy drivers). they both work, it's all about what you care more about. the lexas (or other high end automaker) has higher quality components and testing than ford (or anyother "generic" automaker)

i thought a lexus was a ford?