TR Forums

I've been useing a XCLIO GREATPOWER 600W for 8 years now its been through 3 mobo changes.
Is there a reason PSUs go bad over time?

Here link to newegg, http://www.newegg.com/Product/Product.aspx?Item=N82E16817189007

Is this PSU in the computer you posted a picture of in the contest? If so, yes, replace it! Burn it with fire, along with everything else.

No that computer is my dads, I think I'll scrap it and find something else for him. I kinda wanna use the case as a donor to chop down into a matx case, but not sure if i want to clean it.

The one I'm talking about is in my current pc, with a fair OC on my 7870 myst. and everything else is running normal. I haven't experienced any problems with it, I'm just wondering if is not gonna be able to push out the full 600 watts anymore.

You can count on a typical PSU losing about 20% peak power output in the first 5 or so years of its life, after which the drop tends to be more gradual (until it dies). That said, the 7870 ME (which is actually a disabled 7950) is a 200W card, so you should have plenty of headroom left.

Not too familiar with the XCLIO ODM, but they are not especially well regarded that I know of, if you do want to replace it, a corsair or seasonic unit would serve you well.

Lower quality capacitors tend to dry out over time. dust and dirt that don't just blow out (think the sticky greasy dust) get in there are start insulating parts to make them over heat.

I now prefer to replace the power supply every build iteration (and keep the old one as a backup)

In my case why spend 1200+ on a build only to put its life at risk with a 5+ year old power supply.

Thanks for the replies, I ordered a corsair cx430, to replace this one when I move it over to a case I'm gonna mod.

Make it a point to blow the dust out of your PSU, even if you don't blow out the rest of the case.

Dust collects quickly in PSUs because of the odd airflow through them. This collected dust also likes to "grow" between exposed high-voltage nodes and ground (due to the high field gradient and ionized "sticky" particles). When there's enough of a dust bridge and elevated humidity, the dust can conduct enough charge to become a problem. I've seen this sort of thing in power amplifiers with just a 50V rail, and PSUs have 300V nodes in several locations.

Interesting I did look in the back of it with a flashlight it looks pretty dusty, I'll see about cleaning it out.

Lol I just checked to see which post you guys were talking about.
The CX430 is a great choice. Its the PSU I have used to build the 3 computers for my family and friends. No failure so far and it has 80+ Bronze so thats a big plus!

Voldenuit wrote:

You can count on a typical PSU losing about 20% peak power output in the first 5 or so years of its life, after which the drop tends to be more gradual (until it dies). That said, the 7870 ME (which is actually a disabled 7950) is a 200W card, so you should have plenty of headroom left.

Not too familiar with the XCLIO ODM, but they are not especially well regarded that I know of, if you do want to replace it, a corsair or seasonic unit would serve you well.

Really, 20%? That seems like an awfully large amount. Does it depend upon component and build quality, like a good Seasonic will lose less than a crappy something-or-other?

After he mentioned that much loss I started to look around and it seems I'm closing in on the upper limits of the non-cap components, weather its really a valid point and not just some people being very cautious I don't mind as the CX430 is plenty and it hardly cost anything.

MadManOriginal wrote:

Voldenuit wrote:

You can count on a typical PSU losing about 20% peak power output in the first 5 or so years of its life, after which the drop tends to be more gradual (until it dies). That said, the 7870 ME (which is actually a disabled 7950) is a 200W card, so you should have plenty of headroom left.

Not too familiar with the XCLIO ODM, but they are not especially well regarded that I know of, if you do want to replace it, a corsair or seasonic unit would serve you well.

Really, 20%? That seems like an awfully large amount. Does it depend upon component and build quality, like a good Seasonic will lose less than a crappy something-or-other?

I believe the 20% figure I quoted came from capacitor aging, and has been used as a rule of thumb for a while. It probably doesn't take into account solid electrolyte capacitors, which would probably last longer, but since most users wouldn't know what type of capacitors go into their PSUs, is often used as a safety factor.

That and the larger capacitors in PSUs are not solid polymer ones. Is it only certain caps in the PSU design that cause this decrease?

You could run it until the wheels fall off. That said, it's never a bad idea to have a replacement on hand.

The commonly agreed 10% ageing per year is fairly accurate in my understanding - meaning that a 600W PSU might only deliver 350W after five years; It's cumulative, not additive - so 600(0.9^5)

I read a couple of whitepapers from Teapo that explained how capacitor ageing comes from ripple current generating heat; the heat accelerates the absorbtion of the dielectric by the paper layer, reducing the ability of the capacitor. The 10% is more like 7-12% based on the quality of the mains supply fed to a typical Active PSU but given that we all use similar grids, it's not going to vary by too much.

I would assume that more expensive, higher quality 80+ Gold supplies use better/bigger components for the same rating, meaning that an 80+ Gold PSU rated for 600W might actually be capable of over 800W when new, and still function at close to 800W when several years old.

YES!

XClio is not bad, they use CWT for most of their units last I checked. Of course the Corsair is better, but the Greatpower should work fine for years to come.

Chrispy_ wrote:

The commonly agreed 10% ageing per year is fairly accurate in my understanding - meaning that a 600W PSU might only deliver 350W after five years; It's cumulative, not additive - so 600(0.9^5)

I read a couple of whitepapers from Teapo that explained how capacitor ageing comes from ripple current generating heat; the heat accelerates the absorbtion of the dielectric by the paper layer, reducing the ability of the capacitor. The 10% is more like 7-12% based on the quality of the mains supply fed to a typical Active PSU but given that we all use similar grids, it's not going to vary by too much.

I would assume that more expensive, higher quality 80+ Gold supplies use better/bigger components for the same rating, meaning that an 80+ Gold PSU rated for 600W might actually be capable of over 800W when new, and still function at close to 800W when several years old.

So if capacitor aging is based on temperatures, it means that 105C rated capacitors age less in a given environment...that's good to know. I just find this a little worrisome because I have used moderately-rated PSUs (450W) my my single CPU/single mid-range GPU systems for some time, and while the peak power draw of my current system is only ~225W, losing 10% per year means that the PSU would be running much nearer to it's maximum capability after some years.

Keep in mind those aging figures for capacitors are based on a defined duty-cycle. If the computer spends significant amounts of time online running at high power in a hot environment, the PSU lifespan will shorten significantly. If it spends most of its time off or in sleep mode in an air-conditioned house, the PSU may last far longer.

Regardless, keep the dust blown out and change a PSU out every few years on any system you consider important. Switchmode PSUs occasionally have some spectacular failure modes that may include fire or the electrical destruction of connected system components, so it doesn't hurt to spend another $80 now and then to keep $800-worth of system components running smoothly.

I am skeptical about this whole loss of capacity over time issue. I can buy that there may be some loss over time, but I have trouble believing that it is on the order of 20% per year.

In my work facility, we have video equipment that has been running 24x7 for 10+ and in a few cases 20+ years. In addition, both there and at home I've run computers for 8 to 10 years without problems (other than the ancient computer itself).

So, does anyone have some actual links to this research showing degradation over time?

And to the original poster, I'd take the view of running your supply until it died. After all, you wouldn't replace the engine in your car after 4 or 5 years just because it wasn't 100% of its original capacity yet was still getting you around town, would you?

videobits wrote:

I am skeptical about this whole loss of capacity over time issue. I can buy that there may be some loss over time, but I have trouble believing that it is on the order of 20% per year.

I tend to agree. Not because of any scientific measurements I've made, but just from observation over the years. The wording should probably read "Can be as much as xx% per year" (crappy hardware in the worst possible conditions). I also think that in all likelihood, capacitor aging starts to taper and then plateau over time.

The Egg wrote:

videobits wrote:

I am skeptical about this whole loss of capacity over time issue. I can buy that there may be some loss over time, but I have trouble believing that it is on the order of 20% per year.

I tend to agree. Not because of any scientific measurements I've made, but just from observation over the years. The wording should probably read "Can be as much as xx% per year" (crappy hardware in the worst possible conditions). I also think that in all likelihood, capacitor aging starts to taper and then plateau over time.

It definitely plateaus because I can still run my decade old dell and its low powersupply. There is no way it is going to hit 0, you know?

I wonder if it has to do with the application of the capacitor in a circuit and what kind of loads it experiences, whether it's in a high or low-voltage circuit etc. Anecdotally I have seen plenty of very old equipment that runs fine too, but who knows how out of spce it's running, or whether the circuit is designed to account for capacitor performance change over time. I think that might be the case, after all manufacturers rate PSUs up to 5 or even 7 years, which would clearly be long enough for the caps to degrade significantly according to what is in this thread.

MadManOriginal wrote:

So if capacitor aging is based on temperatures, it means that 105C rated capacitors age less in a given environment...that's good to know. I just find this a little worrisome because I have used moderately-rated PSUs (450W) my my single CPU/single mid-range GPU systems for some time, and while the peak power draw of my current system is only ~225W, losing 10% per year means that the PSU would be running much nearer to it's maximum capability after some years.

I would assume that a manufacturer offering a 450W supply with a 3 year guarantee will build it from components that will still meet 450W after 3 years of typical duty cycle. They ought to be factoring in capacitor ageing already. It's cheap manufacturers that you probably need to take ageing into account from the purchase date. Decent 80+ units are far more likely to meet spec for the full duration of the product warranty so you could probably start deducting 10% per year from year 4 onwards.

Capacitor ageing isn't a myth, but I am simply guessing that cheap manufacturers with 12m warranties cut every corner they can to get the price down. Reputable manufacturers (ODM's like Seasonic and CWT) with 36m warranties probably make sure that their products meet spec for the duration of their warranties, meaning that they overspec their caps to provide extra headroom. A standard E-Cap used in a PSU will degrade over time because that's the nature of the technology. Better quality caps will degrade slower, which is partly where the 7-12% range comes from even though everyone's using similar AC power with the same phase/ripple characteristics.

Okay, I think a few things need to be cleared up. First of all, a capacitor's loss in capacitance (if it happens) does not necessarily impact a power supplies wattage rating. If one of the main filter caps loses some capacitance, the first thing you'll see is in an increase in ripple, but no loss in output power. In fact, a well-designed supply will take this aging into account and will meet all of it's output and ripple and other specs up to and beyond it's warranty period.

Second, power supplies do not lose power over time. I have no idea where this concept came from, but if it's an extrapolation of the capacitor aging process, it's faulty. I have 30-year old linears that meet every spec as shipped. I have several 10-year old switchers that, according to this theory, should be putting out maybe 20 watts but are still in service and working fine.

With regard to the aging of the caps, here's two aging specs from Teapo:
- aluminum electrolytic (radial): after 1000 hours at rated voltage at 85 degrees C, capacitance = rated +/- 20%
- aluminum conductive polymer: after 2000 hours at rated voltage at 105 degrees C, capacitance = rated +/- 20%

The first thing to notice is the improved temperature spec for the conductive polymer designs. Pretty impressive. The second thing is the variability of the spec over time - the plus OR minus 20%. Why would it be plus OR minus? It's because the capacitors WHEN SOLD are spec'd at +/- 20% . So the manufacturer is guaranteeing that after operating at full temp/voltage spec for either 1000 or 2000 hours, the caps will meet their original specifications.

videobits wrote:

And to the original poster, I'd take the view of running your supply until it died. After all, you wouldn't replace the engine in your car after 4 or 5 years just because it wasn't 100% of its original capacity yet was still getting you around town, would you?

If your car's engine had ordinary wear-and-tear failure modes that included the catastrophic destruction of every component in the vehicle, you might. But in any case, car engines require both routine and failure maintenance in order to continue using the vehicle. Other than maybe a fan change-out there isn't any such thing as routine or failure maintenance in a PSU; if it's getting old and thus at risk for a failure, replacement is the only option.

Most PSU failures simply involve the system going dark but there are some less-desirable possibilities, ranging from subtle and inexplicable instabilities on a system that otherwise seems fine, to catastrophic domino failures. If a PSU has been with the same system for 5+ years, that's probably an acceptable risk, since the hardware has also aged out. If a PSU has been re-used across multiple system builds, then the attached hardware may be worth protecting with a PSU replacement.

videobits wrote:

And to the original poster, I'd take the view of running your supply until it died. After all, you wouldn't replace the engine in your car after 4 or 5 years just because it wasn't 100% of its original capacity yet was still getting you around town, would you?

A more appropriate analogy would be a car that hasn't been serviced in any way for 4 or 5 years; no oil changes, belts never checked, etc. You see, there is no such thing as "maintenance" for a PSU. It degrades as you use it, and at some point it fails. Running it "until it died" might mean that one day your machine doesn't start up anymore; or it might mean system instability that leads to file corruption, or even actual damage to your hardware.