• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Question Do PSU capacitors age/burst even when a PSU is not being used/unplugged?

Turbonium

Golden Member
Mar 15, 2003
1,819
6
81
I have some older rigs that have been in boxes/storage for around 6-7 years now. I also have a few PSUs that have been in storage on their own (one of them being factory sealed, in fact), two of which I'm trying to sell on eBay.

I'm just wondering if any of the PSUs may be an issue; could any of the capacitors (which I understand are the first things to go in a PSU) be on the way out (or already gone)? Or do PSU capacitors only age/burst at significant levels if they're being used?

(I did some quick googling, and it seems the capacitors may indeed be in bad shape - could use some clarification.)
 
Last edited:

VirtualLarry

No Lifer
Aug 25, 2001
51,985
6,909
126
Well, hmm.

I've read that capacitors last longer (electrolytics) in PSUs, if there's a slight charge on them.

So, in that line of thinking, "BNIB, unused" PSUs may go bad SOONER than used PSUs. Of course, usage at higher levels implies a certain amount of wear & tear too.

I had a pair of Antec 750W 80Plus Gold EDG (EDGE?) PSUs, that were picked up cheaply on Newegg when they were clearing them out, and stored BNIB for a number of years, and then opened up and used.

Mostly, so far, they've been OK, but I do wonder a bit.

Got one powering a 5x GTX 1660ti mining rig, each power-tuned to use only 75-80W, card + riser, plus mobo / CPU / RAM / SSD / etc.

Also have the other one in my main rig, with dual GTX 1660 Super cards (power tuned) and mining on them and a Ryzen R5 3600 CPU. Previously, I had an RX 5700 and RX 5700 XT reference card mining, power tuned to around 130-150W ea.

So not stressing them too much, they've held up, even after being in storage for probably nearly five years. Active service for at least a couple of years, I suppose.
 

Turbonium

Golden Member
Mar 15, 2003
1,819
6
81
What's the best way to check PSU health nowadays? Beyond a visual inspection of the capacitors I mean, and software monitoring of rails (I could use a refresher).

As for the PSUs listed on eBay: I'm going to make it clear that they've been in storage for years, and that they're being sold as-is, with no guarantees. Hopefully that should be enough.
 

Justinus

Platinum Member
Oct 10, 2005
2,708
823
136
What's the best way to check PSU health nowadays? Beyond a visual inspection of the capacitors I mean, and software monitoring of rails (I could use a refresher).

As for the PSUs listed on eBay: I'm going to make it clear that they've been in storage for years, and that they're being sold as-is, with no guarantees. Hopefully that should be enough.
Without being able to stress test a PSU with a dummy load to ensure its still capable of giving the rated power without dropping voltage or shutting off, there's not much you can do.

Maybe throw up some stress tests to load your PC as much as possible and make sure nothing funny happens and your voltages are OK. Software monitoring isn't the most reliable for this, it would be better to plug in a spare cable and probe it with a multimeter to check the 12v line.
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,648
105
106
Electrolytic capacitors tend to have a "shelf life" or mtbf. The clock starts ticking from the time they are manufactured. The problem is this time can vary wildly between brands and types as well as operating conditions. And the time can be much more or less than the manufacturer lists.
If a device is stored unused for a long period of time they could degrade or "dry out", in which case the initial power up might be the most likely time to observe failures.
I've heard from several sources that regular use is better than leaving them unused or seldom used.

As far as a psu goes, there is no way to really guarantee a device is good without doing a thorough round of tests on all the output rails under different load conditions. You would be checking that everything conforms the the atx specification essentially.
Many people would typically just plug an old psu in and if no instability was observed they'll just assume all is good. I guess this is ok in most cases. A good motherboard should have overcurrent/overvoltage protection and should survive most failure cases. You're likely to observe random power downs or difficulty getting the system to post.
 

mindless1

Diamond Member
Aug 11, 2001
6,365
785
126
Dodgy Chinese capacitors will rot just sitting on a shelf due to an instable electrolyte formula that breaks down. Better (still somewhat junk) Chinese capacitors won't, and their lifespan is more a matter of internal resistance causing heat during use, and the same is true for quality Japanese caps but they tend to have lower internal resistance (ESR) if a suitable part is selected for the application.

After sitting for these years, the oxide layer in the cap has degraded some and through a few hours of use will be reformed, I mean for the electrolytics, solid caps don't have this issue. Performance maintaining low ripple may be compromised during this reforming process, but these days most components in a PC, step down the voltage from the PSU to something lower anyway and excessive ripple is more a matter of a slight stability decrease rather than damaging levels.

So what I would do is turn the system on, leave it idling in windows (not bios, so at windows power managed low power state) for a few hours instead of immediately going full bore into high load stress testing or gaming, etc.

Odds are that they are fine to reuse, with their lifespan depending more on hours already used, particularly at high load, but you might want to relube the fan(s) if any are sleeve bearing.

If you wanted to be extra paranoid or cautious, you could instead power them up jumpering the PS-On pin to ground, powering a slight 12V load like an old HDD and just leave that running a few hours and measuring voltage instead of subjecting a whole system to it, but I've powered up many old systems after years of storage and never had this be a problem, mostly they just needed a new battery and a few stored in a basement needed cobwebs cleaned out.

At the same time, remember what I stated about some dodgy chinese caps failing just from sitting, which means if any are on a video card or motherboard they may have failed from sitting this long too, but otherwise will also improve from reforming the cap oxide layer from a few hours idling along. This sitting shelf rot issue was mostly seen in the early P4 and Athlon or older era.
 
Last edited:

DIY_Computer

Junior Member
Oct 3, 2021
14
2
11
Dodgy Chinese capacitors will rot just sitting on a shelf due to an instable electrolyte formula that breaks down. Better (still somewhat junk) Chinese capacitors won't, and their lifespan is more a matter of internal resistance causing heat during use, and the same is true for quality Japanese caps but they tend to have lower internal resistance (ESR) if a suitable part is selected for the application.

After sitting for these years, the oxide layer in the cap has degraded some and through a few hours of use will be reformed, I mean for the electrolytics, solid caps don't have this issue. Performance maintaining low ripple may be compromised during this reforming process, but these days most components in a PC, step down the voltage from the PSU to something lower anyway and excessive ripple is more a matter of a slight stability decrease rather than damaging levels.

So what I would do is turn the system on, leave it idling in windows (not bios, so at windows power managed low power state) for a few hours instead of immediately going full bore into high load stress testing or gaming, etc.

Odds are that they are fine to reuse, with their lifespan depending more on hours already used, particularly at high load, but you might want to relube the fan(s) if any are sleeve bearing.

If you wanted to be extra paranoid or cautious, you could instead power them up jumpering the PS-On pin to ground, powering a slight 12V load like an old HDD and just leave that running a few hours and measuring voltage instead of subjecting a whole system to it, but I've powered up many old systems after years of storage and never had this be a problem, mostly they just needed a new battery and a few stored in a basement needed cobwebs cleaned out.

At the same time, remember what I stated about some dodgy chinese caps failing just from sitting, which means if any are on a video card or motherboard they may have failed from sitting this long too, but otherwise will also improve from reforming the cap oxide layer from a few hours idling along. This sitting shelf rot issue was mostly seen in the early P4 and Athlon or older era.
I'm curious about this topic as well, after you warm the caps up by ideling in windows, will they be nearly as good as the day you bought the unit? And how long could you store a mobo/psu/gpu before you would have major problems trying to use it because caps? 10 plus years? 20?
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,648
105
106
There is no fixed time for such things, just estimates and ratings. Given 100 components with a 10,000hr rating, for example, you might have 90 of them perform perfect for 10,000+ hours and 10 might fail after only 5,000hrs.
Luck of the draw essentially. And the ratings are done under specific scenarios ie: temperatures, voltages, frequencies. Deviating even slightly from those scenarios could turn that 10,000hrs into <5,000 quickly.
Modern components are manufactured under much more consistent/reliable conditions than in the past. You might see 99% of them all perform essentially identical.
Going any deeper delves into the chemistry of the actual components and gets technical quick.
I've got a few quality electronics from the 1980's that still work fine. Although i'm sure they might be slightly off on their voltages or behavior, they still function.
I'm sure many new mobo/psu/gpu would function fine 20yrs from now if stored in a cool dry place. The failure rate is almost certainly higher, after such prolonged storage, however.
 
  • Like
Reactions: DIY_Computer

DIY_Computer

Junior Member
Oct 3, 2021
14
2
11
There is no fixed time for such things, just estimates and ratings. Given 100 components with a 10,000hr rating, for example, you might have 90 of them perform perfect for 10,000+ hours and 10 might fail after only 5,000hrs.
Luck of the draw essentially. And the ratings are done under specific scenarios ie: temperatures, voltages, frequencies. Deviating even slightly from those scenarios could turn that 10,000hrs into <5,000 quickly.
Modern components are manufactured under much more consistent/reliable conditions than in the past. You might see 99% of them all perform essentially identical.
Going any deeper delves into the chemistry of the actual components and gets technical quick.
I've got a few quality electronics from the 1980's that still work fine. Although i'm sure they might be slightly off on their voltages or behavior, they still function.
I'm sure many new mobo/psu/gpu would function fine 20yrs from now if stored in a cool dry place. The failure rate is almost certainly higher, after such prolonged storage, however.
Would using them every couple years help prolong lifespan? Or will just leaving stuff in a box for many years be fine? Some stuff is just backup parts and such so I won't use it until something breaks.
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,648
105
106
That's impossible to say for sure, but i've heard many audiophile people swear by the idea of periodically powering up old hardware (specifically mentioning the old caps). I'm sure it varies by exact type, series, manufacturer, age. I doubt the difference, either way, would be substantial for any scenario.
 

DIY_Computer

Junior Member
Oct 3, 2021
14
2
11
That's impossible to say for sure, but i've heard many audiophile people swear by the idea of periodically powering up old hardware (specifically mentioning the old caps). I'm sure it varies by exact type, series, manufacturer, age. I doubt the difference, either way, would be substantial for any scenario.
I saw these power supply testers on Amazon,https://www.amazon.com/Thermaltake-Automated-Supply-Oversized-Supplies/dp/B005F778JO/ref=mp_s_a_1_4?dchild=1&keywords=power+supply+tester&qid=1633678988&refinements=p_85:2470955011&rnid=2470954011&rps=1&sr=8-4



Would these be useful to reform caps after many years of storage? I know there's a paper clip method too but this seems simpler. So the psu warms up and dosnt immediately get hit with normal load on a mobo and bust? If the psu spec changes I will need my old psus for my old boards.
 

mikeymikec

Lifer
May 19, 2011
14,851
5,041
136
Dodgy Chinese capacitors will rot just sitting on a shelf due to an instable electrolyte formula that breaks down.
I've seen this happen with a Gigabyte Athlon XP era board. It started having stability issues that I couldn't narrow down to a part that I could replace, so I ended up selling the customer a rebuild. At the point that decision was made, the capacitors looked fine. I kept the board for another six months I think, and by then some capacitors had started to leak.
 

DIY_Computer

Junior Member
Oct 3, 2021
14
2
11
I've seen this happen with a Gigabyte Athlon XP era board. It started having stability issues that I couldn't narrow down to a part that I could replace, so I ended up selling the customer a rebuild. At the point that decision was made, the capacitors looked fine. I kept the board for another six months I think, and by then some capacitors had started to leak.
have caps gotten better since then? beacause that was a long time ago.
 

mikeymikec

Lifer
May 19, 2011
14,851
5,041
136
have caps gotten better since then? beacause that was a long time ago.
Yes, I'd say so. The newest board I recall seeing a leaky capacitor on was the other day actually :) A customer with a Core 2 Quad (so around 2008) ASRock board.

Nothing made since 2010 that I can think of.
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,648
105
106
Those seem to just verify voltages, probably nothing more than what you could do with a multimeter.
They don't appear to put load on the rails or check anything else.
Someone with a nice shop might put load on each rail ie: 10% 50% 100% and check/scope the voltages.
I guess they could be usefull, just not thorough by any means.
 

mindless1

Diamond Member
Aug 11, 2001
6,365
785
126
You don't need to load the rails of the PSU, unless the particular design requires an external load to stay turned on. It's not heat but rather voltage which reforms the oxide layer, so a paperclip, or PSU tester (small/simple type that doesn't apply much load) would be fine.

There's even a school of thought that the *best* way to do this is you take the caps out of the circuit and apply a minimal voltage then slowly raise it over time, but this is more for preserving components in vintage equipment. If I had a PSU or mobo cap out of circuit anyway then I'd just buy a new one and replace it.

There are too many variables in cap design, circuit design, equipment design to state some rule of thumb for how often equipment should be powered on, more than "every few years". Most PSU and motherboards should be fine for a decade or so if you do as suggested, just power it on into windows/other OS with it idling along at a relatively low output for a few hours. I state this from historical evidence of people reusing old systems but cap design changes, board design changes, again variables between something from yesteryear and today's components.

Caps on newer equipment can fail from bursting too, just a matter of time and heat exhausting their lifespan. Not to point fingers but the Chinese are notorious for over-spec'ing things so that one variable can make a lot of difference. Cap failure is still common in low end PSU especially, low enough end that the better advice is don't use a junk PSU in the first place... start with junk and it doesn't get better as it ages.
 
  • Like
Reactions: Leeea

Leeea

Golden Member
Apr 3, 2020
1,132
1,334
96
If they are brand name they will be fine.

If they are generic chinesium, they were crap to start with and aging them a few years is not going to help matters.
 

VirtualLarry

No Lifer
Aug 25, 2001
51,985
6,909
126
So what I would do is turn the system on, leave it idling in windows (not bios, so at windows power managed low power state) for a few hours instead of immediately going full bore into high load stress testing or gaming, etc.
in my opinion I would do is turn the system on, leave it idling in windows (not bios, so at windows power managed low power state) for a few hours instead of immediately going full bore into high load stress testing or gaming, etc.
O RLY?
 
  • Like
Reactions: mindless1

Fallen Kell

Diamond Member
Oct 9, 1999
5,663
250
126
Given that I have seen equipment with caps working fine that are 60 years old, and I have seen equipment with caps that are brand new which failed, it is really more about the manufacturing of the particular cap and what formula used for it. I would tend to agree that you should baby them for a bit on anything that was in storage for a long period of time. Give them a day or so of not doing much of anything before putting any real kind of stress on them.

You should know pretty quickly if they are still good since most electrolytic caps tend to fail spectacularly with a pop (though not always). Solid/ceramic caps will just fail to properly hold a charge and/or short open/closed (electrolytic caps can do this as well).
 
Thread starter Similar threads Forum Replies Date
G Power Supplies 8

ASK THE COMMUNITY