To what extent do power supplies "age"?

GotNoRice

[H]F Junkie
Joined
Jul 11, 2001
Messages
11,962
One paradox that I've always had trouble being able to reconcile is the short lifespan of many power-supplies. It's not uncommon at all to have a PSU fail after only a year or two, and a PSU that is, let's say, ~8 years old is already considered to be both old and have had a long, productive life.

When we compare that to electronics in just about any other aspect of life, things are dramatically different. I have numerous stereo components that date back to the mid 70's, most of which are 100% original and have never been serviced or repaired. I know many people who have appliances - microwaves, refrigerators, washing machines, etc that date back to the 70's or 80's also.

Typical reply when this comes up in conversation seems to be that power-supplies are often used 24/7, where most audio equipment, appliances, etc are only used occasionally, actually being turned off or not used most of the time. That doesn't really fly in terms of a refrigerator, but none the less, let's go with that.

That brings me to my ultimate question - To what extent do power-supplies "age" independent of usage? As in, I'm talking about a PSU sitting new in the box that has never been used, or maybe a PSU installed in a system that is only turned on briefly once every few months. To what extent does the PSU age, primarily due to the passage of time? I have read posts on here from people who say that the capacitors age even if the PSU isn't being used and ~8 years is about as much as you can expect. I don't personally see how that could be true unless the PSU is built with sub-standard components.

One reason I'm asking, I was just offered an amazing deal on a number of new in box old Antec powersupplies. The powersupplies are brand new, still wrapped in plastic and everything. Problem is they are 10 year old models. They are EPS12V units with 24+8 main power cables, and plenty of SATA and PCIE power cables, so they should still be compatible with new builds and adequate for office systems, etc when I get called in to replace a dead PSU. What I don't want however is to be installing a PSU that is basically a timebomb due to it's age.
 
Two things I would be concerned about:

1) If there was a capacitor problem going around at the time, you probably have bulgy/leaky caps already in the box; you could maybe open one, inspect the capacitors and do some sort of load test.

2) Efficency: ">70%" efficiency isn't great these days. Not sure if that's really a big concern for you.
 
My thoughts is be fine for replacement in older pc's....I wouldn't use them in new builds and i would also not sell them for same price as newer unites...in other words i would pass on the savings to my customers within reason
Like toast said check them out ahead of time before putting them in..should only take an extra 5 min's to see if anything is visibly bad (not like they have a warranty anyway)
 
The capacitor "plague" is largely a thing of the past at this point. That said, all electrolytics will age eventually. In truth, that vintage stereo gear is probably due for some replacement caps.

I am in the process of recapping a Hallicrafters SX-100 HF receiver from the early 60s...it ran fine until about 2 years ago when the 60Hz started getting into the audio pretty bad.

I am running a couple of recapped Antec supplies here that are at least 10 years old.

I don't personally see how that could be true unless the PSU is built with sub-standard components.

Because the liquid, or gel, electrolyte compound used in their construction dries out. The heat of operation will dry them out faster, but they will dry out on the shelf eventually.

Go to digikey and price out replacement caps for these supplies, if it is still a great deal go for it. I would not put them in clients machines without new caps personally.
 
Last edited:
One reason I'm asking, I was just offered an amazing deal on a number of new in box old Antec powersupplies. The powersupplies are brand new, still wrapped in plastic and everything. Problem is they are 10 year old models. They are EPS12V units with 24+8 main power cables, and plenty of SATA and PCIE power cables, so they should still be compatible with new builds and adequate for office systems, etc when I get called in to replace a dead PSU. What I don't want however is to be installing a PSU that is basically a timebomb due to it's age.

Are you paying like $3 for each one or something? The Truepower I and II series as well as the Smartpower series were well known for their above average death rates due to a poor choice in caps and cooling design. In more specific terms, Antec/CWT specified/used caps (Fuhjyyu caps) that did not respond well to heat. Unfortunately, Antec made the decision to have the PSU fan only ramp at really high temps in an effort to keep the PSU quiet. However, just medium to high temps were enough to screw up the caps so really high temps were murderous. Over time, this degraded the quality and performance of those PSU series.

So when you mentioned "sub-standard components", those Fuhjyyu caps are a clear example of that. As jojo69 mentioned, you would need to recap these PSUs before butting them into people's systems.

With that said, I've never seen actual empirical tests/reports/articles showing how an unused PSU ages over time.
 
I actually trust old power supplies (like 5 years old, even used) as long as they pass a series of tests.

With every single computer I have ever repaired for anyone, I always BEGIN with opening up the PSU and checking for bulging caps, signs of overheating (dark/brown spots on the PCB), dust buildup.

I check for insulation damage, the works.
Bulging output cap? I assume most caps from the vendor in the device are candidates - toss.
Dark spots around the main transistors? Garbage.
Non-heatsinked bridge rectifier? PCB darker around points where wires go in? Toss.
Matte-looking solder points? with brown stuff precipitating? Nope.

But - sometimes I am pleasantly surprised to open up a 7 year old PSU and it looks great. If it looks great I'll stress the crap out of it on a known good system.

If everything is 555 I often leave the old PSU in and it's never came back to bite me (yet).

I don't bother with multimeter checks, I just stress it.
After stressing the PSU (and discharging any left over power in the caps) it gets opened up again and I check for heavily overheated components (a sign of possible failing).

IMHO it's not black-white and even capacitor aging might go unnoticed as long as the manufacturer left some headroom spec-wise.
 
micharlz post above receives the jojo69 stamp of approval

But the Fuhjyyu caps in the OPs particular Antec supplies are kind of notorious, I would reiterate that those are not really suitable for clients machines without replacement.
 
Thanks jojo69 :)

Recapping is a topic in itself - in the ROHS world I usually use a cheapo solder pump to remove the PB-free solder, clean the wound and then solder with my stuff.

I don't have the iron or expertise to do it the 'best way' for ROHS. Is it a horrible sin to use 60%Sn-40%Pb solder in ROHS PSUs? Will I bump into cracking later on?
 
I have never had a problem sucking out and reworking with real solder.

the lead free stuff is a disaster, lead sticks to everything IMO
 
But the Fuhjyyu caps in the OPs particular Antec supplies are kind of notorious, I would reiterate that those are not really suitable for clients machines without replacement.

And I will +1 that for sure. Those units are from the height of CWT's Fuhjyyu obsession, and as such cannot be trusted. These caps were terrible new, let alone ten years down the line. It was not uncommon to find those capacitors failed after only a year in use.

Recapped, those PSUs will be decent. Until then, I really wouldn't use them.
 
The capacitor "plague" is largely a thing of the past at this point.

I really wish this was the case, but unfortunately it isn't. Manufacturers are still using shoddy caps to this day in new products.

Antec is a good example. Their PSUs starting in about 2003-2004 used those godawful Fuhjyyu caps that would explode or leak when they failed. Then around 2006 they switched to a variety of other capacitor brands like Teapo, United Chemi-Con and Ost which still fail, but in a more discrete way; The electrolyte would dry up. The rubber seals on these caps are defective and the electrolyte will slowly dry out and crystallize. In some cases they'd also vent electrolyte as well.

Today they still use those bad caps, and I see them on other things too like televisions, monitors and even on the cheaper motherboards that don't use the more expensive polymer capacitors in the brushed aluminum cans.

Unfortunately, Antec made the decision to have the PSU fan only ramp at really high temps in an effort to keep the PSU quiet. However, just medium to high temps were enough to screw up the caps so really high temps were murderous. Over time, this degraded the quality and performance of those PSU series.

I know of this problem all too well. Probably 5-6 years ago I was getting tons of computers with Antec PSUs that were overheating. I caught most of them before they cooked completely and developed a mod to disable the fan control unit and run the fans at either half or full speed depending on how much noise the customer could tolerate.

There were a few of them where I replaced the fans with thermistor controlled ones, which would properly respond to the internal heat of the PSU and were less noisy.
 
I have never had a problem sucking out and reworking with real solder.

the lead free stuff is a disaster, lead sticks to everything IMO

I have not had any problems with lead free solder. You just have to tin stuff and use solder paste to get it to flow good.

Lead based solder on the other hand tends to crack after so many heating/cooling cycles.

I have had to repair multiple electronic things that used lead based solder, the worst of which was an old TV that shot flames out the back because the solder cracked, the connection heated up too much and it burnt some leads and even burnt part of the board. That was a fun project with multiple jumper wires.

A few more examples are:
Main relay in a Honda Civic - this part was widely known to have problems with the solder joints cracking

Stove/Oven controller. I must have resoldered 20+ connections on this thing after it flaked out. Saved around $300 in parts just by doing that though.

Roland SC 55 - multiple cracked solder joints when I got it.

Netgear router - the thing fried because the solder joint holding the post that held the heatsink in place cracked and the heatsink moved out of place. Stupidest design ever.

Basically if I get any older electronics that has any problems whatsoever, the first thing I look for is cracked solder connections. Lead based solder just doesn't seem to hold up from what I have seen.
 
I really wish this was the case, but unfortunately it isn't. Manufacturers are still using shoddy caps to this day in new products.

Indeed. Cheap capacitors aren't cheap because a company likes you the manufacturer of the product. They are cheap because the product, QC, supply chain, distribution, etc. are lacking in one or many ways.
 
I am certainly seeing plenty of cheap, substandard caps, and plenty of capacitor related failures, certainly not arguing that. But this is nothing compared to the actual plague from the turn of the century till 06 or so when the Taiwanese factories were churning out units with an incomplete electrolyte formula they had obtained in a failed act of industrial espionage.

There were good opportunities to be had back then if you knew what you were doing, I picked up a pallet load of linear 12V supplies for a song, once my digikey order came those fed me for a good while.
 
I am certainly seeing plenty of cheap, substandard caps, and plenty of capacitor related failures, certainly not arguing that. But this is nothing compared to the actual plague from the turn of the century till 06 or so when the Taiwanese factories were churning out units with an incomplete electrolyte formula they had obtained in a failed act of industrial espionage.

There were good opportunities to be had back then if you knew what you were doing, I picked up a pallet load of linear 12V supplies for a song, once my digikey order came those fed me for a good while.


That's what I remember. I did not matter what brand of caps were on the boards or PS. Every cap manufacturer was suffering from the counterfeit electrolyte that everyone purchased dirt cheap in 2003. I had a Antec 500 watt Earthwatts supply fail within a year and take out my whole entire system along with the peripherals back in 2005.
 
...The horror....

34e4xhi.jpg


...... the horror.....​
 
That's what I remember. I did not matter what brand of caps were on the boards or PS. Every cap manufacturer was suffering from the counterfeit electrolyte that everyone purchased dirt cheap in 2003. I had a Antec 500 watt Earthwatts supply fail within a year and take out my whole entire system along with the peripherals back in 2005.

Unfortunately, you remember incorrectly. It's a long story I have posted about before but the short version is the P50 formula from Rubycon was "acquired" by an individual who then made their way with it to LTEC (Luminous Town Electric Company) where a version of that formula was produced. When people left LTEC and began working at various Taiwanese and Chinese manufacturers they took that formula but the production was not correct due to transcribing and QC.

Electrolytic formula is not the only reason a capacitor can be bad as the aluminum, seals, etc are all critical. Lately, because people have short memories, people have been getting all wet on low quality Chinese capacitors as being quality products simply because some Japanese companies are manufacturing some of their product on mainland China. Just because Japanese companies are making products in their factories, with their QC, their formula, their stock sourcing, and their training does not mean that the local production by indigenous firms is anywhere near the same quality.
 
...The horror....

34e4xhi.jpg


...... the horror.....​

Those look like either Teapo or UCC capacitors. Teapo used to use an ugly olive green plastic wrapper for their caps but changed to a brown plastic later on. UCC has pretty much always used brown plastic.

If those are UCC caps, they'll have KZG on the side (which is the capacitor line.) For whatever reason, UCC never puts their name on their caps.
 
Yes, IIRC they had KZG printed on them but that was a good while ago. Changed that whole cluster to Panasonics (I think). It's important to replace all units of a given type if one fails.
 
Lead based solder on the other hand tends to crack after so many heating/cooling cycles.

Isn't it the other way around? I thought the switch to lead free solder was the cause of the entire nvidia fiasco and even the xbox 360 rrod.
 
I remember seeing a paper that, to my surprise as well, proved that ROHS stuff actually did win over PbSn in things like Ball Grid Arrays, but needed higher temperatures (thus expensive). But that was long ago. I've aged since.
 
Unfortunately, you remember incorrectly. It's a long story I have posted about before but the short version is the P50 formula from Rubycon was "acquired" by an individual who then made their way with it to LTEC (Luminous Town Electric Company) where a version of that formula was produced. When people left LTEC and began working at various Taiwanese and Chinese manufacturers they took that formula but the production was not correct due to transcribing and QC.

Electrolytic formula is not the only reason a capacitor can be bad as the aluminum, seals, etc are all critical. Lately, because people have short memories, people have been getting all wet on low quality Chinese capacitors as being quality products simply because some Japanese companies are manufacturing some of their product on mainland China. Just because Japanese companies are making products in their factories, with their QC, their formula, their stock sourcing, and their training does not mean that the local production by indigenous firms is anywhere near the same quality.


Rather than rehash it all over again, any links to more details? I remember this when it went down, but its been so long I can't recall that well.
 
I have some Antec BasiQ 500W PSUs built by Fortron-Source, and their CapXon capacitors bloated even just sitting in storage. CapXon is considered one of the worst brands, along with Fuhjyyu, found in the original Antec SmartPower and TruePower models. It's been recommended to fire up the Farnsworth every 6 months for an hour, to reduce the odds of the capacitors going bad.
 
Back
Top