PSU life expectancy / performance degradation

[ZenMonkey]

I'm just curious because I don't see this topic discussed. Barring specific hardware malfunction, does PSU performance naturally degrade over time (i.e. become less efficient or less stable in power output)? If so, what's the average reliable performance life expectancy for a PSU in a normal use case?

[STRMfrmXMN]

10 minutes ago, ZenMonkey said:

I'm just curious because I don't see this topic discussed. Barring specific hardware malfunction, does PSU performance naturally degrade over time (i.e. become less efficient or less stable in power output)? If so, what's the average reliable performance life expectancy for a PSU in a normal use case?

Yes, but a decent quality unit will be outputting its rated wattage for several years before it's a problem. You could be running a Seasonic PRIME Titanium 15 years from now and as long as standards don't change so drastically that the unit is considered crap by 2032 you'd be fine.

[Cyracus]

partially depends on its operating environment (heat, dust, exposure to anything that could harm it) and partially depends on build quality. Even the best PSU today could be toast in a few years if it gets choked with dust and operated at high temperatures

[Stefan Payne]

1 hour ago, ZenMonkey said:

I'm just curious because I don't see this topic discussed.

Because there is no actual evidence for it.

Nothing.
We are in the area of faith/believe...

 

1 hour ago, ZenMonkey said:

Barring specific hardware malfunction, does PSU performance naturally degrade over time (i.e. become less efficient or less stable in power output)? If so, what's the average reliable performance life expectancy for a PSU in a normal use case?

As stated above, there is no evidence for the 'PSU degrading' Myth. Nothing.

Only some people talking about it, not one actual fact.

 

In fact this (GERMAN!!!) Article disproves this statement:

https://www.computerbase.de/2013-11/alte-netzteile-test/

 

So I don't know where this 'PSU degrading' Myth comes from. And I could not find one piece of evicence that proves it in any way.

Only two articles that disproves it...

The other one is the Silverstone Decathlon or Olympia with CapXon Caps at HardOCP.

Ripple/Noise got a bit worse (obviously) but there was absolutely no evicence for this degrading shit...

[ZenMonkey]

@Cyracus - That's why I said normal use. It's just sitting in the computer room running around 4-6 hours per day somewhere in the 70s Fahrenheit using around half of the wattage the PSU is able to output.

 

@Stefan Payne - I appreciate your desire to dispel any misconceptions, but you seem particularly incensed by this topic and come off a little angry.

[Cyracus]

6 minutes ago, ZenMonkey said:

That's why I said normal use. It's just sitting in the computer room running around 4-6 hours per day somewhere in the 70s Fahrenheit using around half of the wattage the PSU is able to output.

For some people it is normal... Capacitors can and do fail, even high quality ones are typically rated for something like 100,000 hours, sure that's like 11.5 years of 24/7 operation, but I know some people who practically never turn off their computer... lower end PSUs use less good components and are more likely to fail earlier. solder can crack as well...

[Stefan Payne]

47 minutes ago, Cyracus said:

Capacitors can and do fail, even high quality ones are typically rated for something like 100,000 hours, 

Capacitors will fail, the questiuon is just when (and how) 

 

And you seem to have a zero too many.

It's more like 10.000 hours lifetime at max Ripple + Temperature for normal wet aluminium electrolytic capacitors...

 

 

@ZenMonkey

All I have seen is that many rather uninformed people  talked abouit this 'issue' but never ever have I seen any prove that supports this theory.

It is probably based on Capacitor datasheet and the statement of the endurance specifications. That's all I could find about that.

And it's only a thing in the english speaking area, not in for example German area.

 

So all we have is some capacitor datasheets and someone somewhere concluded that you that means that the PSU itself degrades within time and does not output the maximum capacitance. Wich is just totally wrong.

The capacitors don't really have something to do with the max. capacitance - look at for example be quiet Pure Power 10. ALL have the same secondary capacitors.

The 400W has the same mixture of 2200 and 3300µF ones as the 600W one has. 

[Cyracus]

10 minutes ago, Stefan Payne said:

And you seem to have a zero too many.

It's more like 10.000 hours lifetime at max Ripple + Temperature for normal wet aluminium electrolytic capacitors...

idk, pretty sure I've heard about 100k hours, maybe something like this...  other bits of this also explain what can affect capacitor life http://www.tomshardware.com/reviews/power-supplies-101,4193-4.html

[IKnight]

Welp. I'm using a 10-year-old PSU and it's still going strong. Even the fan has managed to stay quiet. 

[loculus]

8 minutes ago, Cyracus said:

idk, pretty sure I've heard about 100k hours, maybe something like this...  other bits of this also explain what can affect capacitor life http://www.tomshardware.com/reviews/power-supplies-101,4193-4.html

You're misunderstanding the point. The rating is usually something more like thousands of hours (not 100k), but this is at specified temps, etc. If they are run in less stressful conditions, like lower temps, then they should be fine for potentially quite a bit longer than that rating. And that is how they will be run in a typical power supply.

[Cyracus]

2 minutes ago, loculus said:

You're misunderstanding the point. The rating is usually something more like thousands of hours (not 100k), but this is at specified temps, etc. If they are run in less stressful conditions, like lower temps, then they should be fine for potentially quite a bit longer than that rating. And that is how they will be run in a typical power supply.

yeah... that's basically the dumbed down version of what they say in that link 

[Dabombinable]

45 minutes ago, Stefan Payne said:

Capacitors will fail, the questiuon is just when (and how) 

 

And you seem to have a zero too many.

It's more like 10.000 hours lifetime at max Ripple + Temperature for normal wet aluminium electrolytic capacitors...

 

 

@ZenMonkey

All I have seen is that many rather uninformed people  talked abouit this 'issue' but never ever have I seen any prove that supports this theory.

It is probably based on Capacitor datasheet and the statement of the endurance specifications. That's all I could find about that.

And it's only a thing in the english speaking area, not in for example German area.

 

So all we have is some capacitor datasheets and someone somewhere concluded that you that means that the PSU itself degrades within time and does not output the maximum capacitance. Wich is just totally wrong.

The capacitors don't really have something to do with the max. capacitance - look at for example be quiet Pure Power 10. ALL have the same secondary capacitors.

The 400W has the same mixture of 2200 and 3300µF ones as the 600W one has. 

TBH with capacitors, I'd only worry about them if they got to around 20+ years without failing due to the rubber seals. Considering I've got PSU and motherboards from the time of the capacitor plague that are still fully functional after only having 1-2 caps fail (swollen tops, burst seal). And in the case of 1 PSU it was because they used capacitors of the incorrect dimensions which meant that they didn't dissipate heat well enough (I luckily had some new caps that were of the correct specifications).

[Stefan Payne]

4 hours ago, Cyracus said:

idk, pretty sure I've heard about 100k hours, maybe something like this...  other bits of this also explain what can affect capacitor life http://www.tomshardware.com/reviews/power-supplies-101,4193-4.html

no, look at a datasheet.

Most Capacitors in PSU (even the Wapanese ones) are rated for something between 2000 and 4000h.

Like Teapo SC, SY, SJ or Nippon Chemicon KZE.

 

BUT:
That's specified at rated temperature with rated current.

So with just 55°C instead of 105 the lifetime is extended enormously, to be precise it's said to duble every 10°C so we are talking about around 128k-256k in theory.

 

BUT capacitors also age because of chemical reactions, without doing anything. So that time might be irrelevant and the capacitor might die anyways...

 

Either way, it's in no way something that proves the PSU derating.

If the capacitors fais softly (like Capacitance and/or ESR slightly out of spec), nothing happesn, just a bit more ripple/noise than when the PSU was new...

 

But if they really fail, you have to replace them (or the PSU). So in the end, it doesn't really matter anyway....

[Guest]

6 hours ago, STRMfrmXMN said:

You could be running a Seasonic PRIME Titanium 15 years from now and as long as standards don't change so drastically that the unit is considered crap by 2032 you'd be fine.

Probably not. 

 

And I'm not being biased.  I would say the same thing about a Corsair AX or AXi.

[Guest]

5 hours ago, ZenMonkey said:

 

@Stefan Payne - I appreciate your desire to dispel any misconceptions, but you seem particularly incensed by this topic and come off a little angry.

That's Stefan.  But I've come to accept that a lot of it comes from the fact that he's German.

 

Fact of the matter is, real life testing isn't available.  i.e.  Same unit tested 10 years ago is tested again today.

 

There's MTBF, but that's done at maximum load/temperature/etc.

 

There's DMTBF, but that's even done at maximum usage.

 

Some things we do know: 

 

1.  Fans fail first.

 

2.  Capacitors fail second.

 

3.  Most PSUs *should* last 5 years at 100% load, max temperature rating if used 24/7.

 

4.  Based on experience, typical use of a PSU (typically 50% or less of PSU's capability) should last about 10 to 15 years before the fans die or caps start to leak.

[Cyracus]

11 minutes ago, Stefan Payne said:

no, look at a datasheet.

Most Capacitors in PSU (even the Wapanese ones) are rated for something between 2000 and 4000h.

Like Teapo SC, SY, SJ or Nippon Chemicon KZE.

 

BUT:
That's specified at rated temperature with rated current.

So with just 55°C instead of 105 the lifetime is extended enormously, to be precise it's said to duble every 10°C so we are talking about around 128k-256k in theory.

 

Yeah, except they kinda explain how to calculate the actual life expectancy 

Spoiler

Electrolytic Cap Life Calculation

 

Having a way to easily calculate a cap's lifetime is very important. Here is the basic equation that we will use.

 

        (5)                 L₂ = L₁ x (V_r/V_o) x 2^x   

 

In the above equation where:    x = (T_m-(T_a+ΔT))/10

 

    T_a  represents the ambient temperature.
    T_m  is the max rated temperature of the capacitor. 
    ΔT  is the temperature rise of the capacitor due to ripple current.
    V_r  is the maximum voltage rating of the capacitor.
    V_o  is the operating voltage of the capacitor.
    L₁  indicates the load life rating of the capacitor (provided by its manufacturer).
    L₂  is the calculated lifetime of the capacitor under the current operating conditions.

 

What is APFC?

APFC stands for active power correction factor and an APFC converter shapes the current waveform to make it match the voltage waveform. It also makes the PSU compatible with a wide range of voltage inputs.

Now, let's say that we have an electrolytic bulk cap with a 2000-hour load life rating, 420V max voltage rating and 105 °C max temperature rating. If we use this capacitor inside an APFC converter where the DC bus is around 380VDC, assuming that its internal temperature will be kept at 50 °C while the temperature rise of the cap due to ripple current will be 10 °C, its estimated lifetime will be calculated by the following equation:

 

        (6)            L_{2 =} 2000 x (420/380) x 2^4.5 ≈ 50,019 hours     

 

If we assume that the ambient temperature is 40 °C, then we have the following:

 

        (7)            L_{2 =} 2000 x (420/380) x 2^5.5 ≈ 100,037 hours      

 

The above example clearly shows that 10 °C difference in the operating temperature of an electrolytic capacitor can double or cut in half its estimated lifetime. In addition, caps with 105 °C rating have four times the life compared with similar specification caps that have an 85 °C temperature rating.

 

12 minutes ago, Stefan Payne said:

BUT capacitors also age because of chemical reactions, without doing anything. So that time might be irrelevant and the capacitor might die anyways...

 

Either way, it's in no way something that proves the PSU derating.

Yeah, caps can degrade or have flaws, something that tends to be considered a manufacturing flaw if it happens when the caps fail well before their expected lifetime, it makes sense that good PSUs tend to have a 5 year warrranty and really good PSUs have 10 year warranty, companies wouldn't be making money on PSUs if they kept having failures within those time spans. And nobody was talking about PSU derating