Let's take another look at capacitor aging

I see a lot of talk about "capacitor aging". It figures prominently in some of the PSU calculators. But I do not think it's as great a problem as "everybody" says it is.

Do capacitors age? Certainly. All electronic components (except perhaps inductors and transformers) age. One of the PSU calculators allows 10% a year PSU degradation under normal usage and 20% - 30% per year with 24/7 computer usage. This does not pass the "Does this make sense?" test to me.

I have a 7 or 8 year old 400 watt Antec power supply that apparently still works fine. Figuring 10% degradation a year, it should be able to produce only .90^7*400 or about 190 watts.

I have another system (recently pulled out of service) that I ran 24/7 except for when I was away on vacation. It ran for about 4 years on a cheapy 450 watt PSU. According to the PSU calculator, the PSU was producing around 180 - 190 watts when I pulled it out of service. I passed it on to a friend. It's still working in his system.

About 3 years ago, I dumped a working 10 year old P233 system (motherboard built long before the high reliability solid caps appeared on the market). It was running at 333 MHz (found an undocumented jumper setting).

I have a 25 year old Yamaha stereo amp (dead tuner) that still works well. Let's see:
.90^25 = .07% of original power.

My younger brother has a '71 or '72 vacuum tube (heat! and high voltage) guitar amp that except for tubes is original - well, the spring reverb unit doesn't work. Let's see:
.90^40 = .015% of original - really does not pass the "Does this make sense?" test.

Then there's all the 10 - 15 year old professional test equipment (mostly HP and Tektronix) in my maintenance shop.

I just do not worry about capacitor aging.
 

v1zzle

Distinguished
Mar 22, 2009
170
0
18,710
Interesting notation on that. I myself am a gamer and i've been using my Corsair TX 750 watt PSU for about 3 years now, and it still runs great. I've had a great deal of graphics crads installed on it, ranging from SLi 9600 gt to 9800 GX2 to SLI GTX 460 (current) and it's still toughing out. I think the rule applies if the PSU is running at full load. Especially today's new PSU are very efficient with the 80+ features and wattage control, if you clean and properly maintain your hardware it will last a long time. You mentioned you have a 25 year old Yamaha stereo amp, that is cool, my dad actually has an old Carver receiver that dates back from 91-92 I believe, and that thing still works great, it actually touched out his Yamaha receiver that has been in the shop several times. It can stand up to most of the hardware of today- and then some, too bad Carver changed their name to Sunfire, reminds me of the car, lol, anyhow.
 

Ten98

Distinguished
Jan 23, 2008
99
0
18,660
10% a year is just crazy. I think maybe the drop in efficiency is something more like this:

1 Year - 5%
2 Year - 2.5%
3 year - 1%
4 Year - 0.5%
5 Year - 0.25%

etc etc.
 

ferdnyc

Distinguished
Aug 26, 2011
2
0
18,510
I realize this is an old thread, forgive me for the act of necromancy. I read the original post and it brought up some issues in my head.

jsc, I see two "problems" with your initial analysis... both sides of the same coin.

(There's also a third: the fact that you end up reaching waaaay beyond the class of devices in question. PC PSU capacitor-aging computations don't apply to ancient stereo system or vacuum tube(!) instrument amps in any way! That's just silly. But let's forget that bit, and focus on the question as it does pertain to PC PSUs.)

The first issue I see is that you're applying the calculator adjustments backwards. The calculators don't function to predict the output of a PSU (initially or over time), they function to predict what size PSU will be needed to power a given setup (initially and, through capacitor aging, over time).

So, take this example...

...Except, that's not what the calculator is saying. What it's saying is (assuming your system was maximally loading the PSU from day 1 — in other words, that it was a load that even required a 400W PSU) that over time that 400W PSU will grow less able to handle that load. If you wanted, initially, to buy a PSU that would definitely be able to handle a 400W load with no problems even when it's four years old, you should get one rated for at least 560W (400W + 10%x4). That's not even accounting for headroom or worrying about efficiency, it's just saying that even accounting for its capacitors degrading over time, a 4-year-old 560W PSU should have no problems handling a 400W load. Anything smaller, you might run into problems.

And that's the other half of it — the calculators aren't telling you what they expect to happen to actual PSU performance over time. They're telling you what you should be able to trust it to handle. So, take this second example...

Again, not what the calculator is saying. It's saying that, due to the PSU's age, you shouldn't trust that PSU to power a system that requires more than ~180W-190W. And I personally think — again, given its age — that actually sounds pretty reasonable. If I had an old 450W PSU that I knew wasn't very good to start with, I'd never try to power a system that requires 400W with it! But if I had a system that I knew wouldn't tax it past around half its original rating, well, then I'd think it would probably be just fine.

The calculators aren't predictive — they aren't even MEANT to be. They're built to account for worst-case scenarios, so of course their degradation numbers are pessimistic. Because doesn't it make sense to overbuild a little bit in the beginning, to minimize the chance of problems developing down the road?
 
Well, the thread has arisen, so I'll add the remark that the impression I have is that capacitor aging is really only a factor when the capacitors are being badly stressed, such as by high heat. Kept confortably within spec, better capacitors (i.e. with no liquid electrolyte to dry up) will last indefinitely, at essentially their original rating.
 

Cygnus x-1

Distinguished
Jul 28, 2011
901
0
19,060
Look at some of the very old audio file A class amps, like Mcintosh. ten year and older from this company is still fine and expensive in some cases.
 

leon2006

Distinguished
High end equipment that i required over 15 years still works.. receivers, stand alone amps, even no name PSU for PCS that i retired because its so old... It still runs.

Branded PSU i have great experience w Antec 430 Watt (> 10 years)...

Not so great w/ Antec 850W Quatro ... 3 Failures in the last 4 years.... I phase it out. I built 3 PCS (exactly same config) and all required replacement of Antec 850 W Quadro.

I had one failure of Antec 500W

My PC PSU are loaded to 80% Capacity. I always leave 20% buffer for Start-UP surge (CMOS). My PCS are powered 24x7 except 1 day per quarter when i clean it.

Once very 3 months i stress my PC for 24 hrs that means stressing the PSU to 80% of calculated load.

That capacitor aging is for marketing purposes only. Its a question if the vendor installed reliable capacitors or not.

I just purchase a good reliable brand w/ great reviews PSU.

I also have UPS w/ surge protectors, application specific power filters on my ac line.

Sometimes Murphys Law do hit... Like a huge lightning strikes that zapped electronic equipment

To conclude i don't use the capacitor aging... I use my engineering judgement to decide the brand, capacity of my PSU for my PC.
 

bebop460

Distinguished
Jun 25, 2011
169
0
18,710
I'll awake this dead topic to say that it has eased my mind a great deal. Now I won't feel so bad about beating the hell out of my Seasonic 850 watt (80-90% load). PSU calculators and benchmarking software be damned; there is nothing like first hand experience. Truely excellent information
 

jassenjj

Distinguished
Sep 10, 2011
4
0
18,510
Well, interesting thread with many wrong outcomes.
Capacitors age differently, but aging does not change the power rating of the PSU and including such calculations is misleading.
With time the quality of the output filtering and regulation degrades (higher ripple, voltage jumps with sudden changes in the load, e.g. starting/stopping a video game that pushes the GPU's power consumption to the max).

With a good thermal design a PSU may degrade many times slower than the capacitors chart suggest as the working temperature is the most important parameter for the aging. The other parameters depend on the role of the capacitor and usually bad design or choosing the wrong types happens rarely.
Bad thermal design can be compensated if you use a power supply rated at twice (or more...) higher power as it will work on much lower thermal and electrical stress.

I'll just say that in the last 7 years I use a veteran Seasonic SS-351HT power supply (feeds CPU E8400, GPU HD4850, 4 HDDs, its fairly loaded, works 24/7). Although not a low quality product it uses OST capacitors which have bad reputation. Recently I noticed some instabilities that suggested power issues: display driver crash when *stopping* some GPU intensive task. I opened the PSU and found 2 swollen capacitors. But it wasn't capacitors' fault (I thought) as those two were placed in a position far from the cooling airflow and near to the heatsinks' base. Bad design? No! This unit survived 3 generations of CPUs and GPUs... and it will survive more. But this design lead to the failure of those particular capacitors.

I replaced all capacitors with ones with better characteristics (and reputation) and the problem is gone. Except the 2 swollen capacitors all others had good capacitance not less than 90% of the rated value and perfect ESR characteristics. So ... how long your power supply will sustain its rated power? It depends... But this does NOT mean at all that if it can't sustain anymore its rated power due to aging caps, it would be safe to load it with the half. You should either change the unit or replace the capacitors as the damage could be higher.
 

willyroc

Honorable
Jul 22, 2012
257
0
10,810
Old thread with useful information, I believe :). In my case, I've experienced a 300W rated PSU work at 250W after exactly 5 years of usage. So I'm not sure how that applies to capacitor aging. But I think capacitor aging should not be considered in the big picture.