Running a 1000W PSU for 13 years now. Should I replace it or wait for it to fail?

[tulx]

Hello everyone,

 

I´ve been running a Zalman ZM1000-HP as the only constant component in my various builds since 2008 and, apart from some slightly annoying coil whine, it´s been working perfectly. I´m beginning to wonder now if I should proactively replace it at some point. I think the PSU is rated at 100.000 hours and I could be at about 75.000 hours accoring to a very generous estimation. I´d have no issue with running it until it fails in another 10 or so years if had the certainty that it would not take my other components with it when it eventually does. Are there any experiences with similarly old components or a general best practice for such cases?

Thanks a lot in advance for your time and have a pleasant week!

 

Kind regards,

tulx 

[jaslion]

Yes do replace it. BUT it depends on your hardware.

 

The main issue here is that it is just extremely out of date. It lacks all the new regulations and protections common in current good quality psu's that new hardware expects you to have and you lacking these can cause damage or when this psu fails a total fried system.

 

The best practice for psu use is to always get a GOOD QUALITY UNIT and basically replace it when:

 

1. It's not enough for the build

2. You are doing a totally new system and it has gone beyond 5 years of use (this depends on if it has current expected protections or not of course)

3. You have a bad quality psu

 

You had a good psu for the time but by todays standards it would be seen as pretty bad and not recommended to be used at all.

 

[Skiiwee29]

PSU technology has advanced a ton since then with protections and such. I would replace it sooner rather than later. General rule of thumb for PSUs is to replace them when warranty time is up or close to expiring. 

[mariushm]

Eh, I'd have to disagree. 

 

Unless you ran the power supply 24/7 at close to 100 watts for those 13 years, there's really not much reason to replace it or panic about it. 

If your system averaged maybe 400-600 watts, then that power supply most likely has at least 5-10 years more life. 

 

The 100.000 is a MTBF value, not the number of hours the power supply is rated for or warrantied for.

Components like diodes and mosfets can last for 30 years+ of continuous use, the only components that can break are moving ones (fans, ex losing lubrication, plastic becoming brittle from heat/cold cycles, things falling down onto the blades etc)  and components like electrolytic capacitors which use semi-liquid electrolyte which can be chemically damaged over time or the rubber seal on the capacitors may break down after around 20-25 years. 

 

So the parts that can fail are the fan and electrolytic capacitors. Fan can be easily replaced even by yourself, just have to be careful to get one with similar amount of air flow (you have the part number printed on the fan sticker, so you just look in the datasheet and then get a fan with similar specifications)

 

Quality high end capacitors for power supplies are not expensive and can be bought at distributors of electronic components so you know they're not ebay fakes... would probably cost you around 10-20$ to buy the capacitors that are most likely to be "degraded" due to heat inside the power supply. 

You can take the capacitors and the power supply to any place that services laptops or phones and they'd be able to replace the capacitors for you, if you don't feel like doing it. 

 

I wouldn't rush to replace it. if it's not causing you problems now (ex random resets or stuff like that), then continue using it. 

 

I'd open it and visually inspect capacitors to see if they're fine (tops not swelling - that's sign the electrolyte produced gas and swollen the capacitor case, sign of capacitors going bad) and maybe I'd consider replacing the fan if it's making noises.. But that's just me, I'm biased a bit, as  I'm comfortable opening stuff because I know how to repair things and have the equipment to solder components. 

I can understand if someone else is afraid of opening a power supply or doesn't trust replacing components in a power supply.

[Somerandomtechyboi]

29 minutes ago, mariushm said:

I wouldn't rush to replace it. if it's not causing you problems now (ex random resets or stuff like that), then continue using it. 

Yep have to agree on this

 

Recently i bought this "dazumba" 450w and after checking the warranty seal thingy this thing seems to be ~11-12 years old, idk what the hell it was used for before it came into my possesion but it still works fine, lets me actually overclock on my p5q without worrying about stupid volts going haywire

 

And judging by the 12v rail it seems to be of decent quality (30a on 12v rail so 360w) not to mention how heavy it is

 

Capacity still seems abit low tho so i think ill replace it with a rexcool 700w once i upgrade to a better platform like 1156 or 1366

 

 

Tbh if it isnt causing any problems then why bother replacing it? 1000w is still plenty even for todays hardware so just keep using it till it dies

[Guest]

9 hours ago, mariushm said:

Eh, I'd have to disagree. 

 

Unless you ran the power supply 24/7 at close to 100 watts for those 13 years, there's really not much reason to replace it or panic about it. 

If your system averaged maybe 400-600 watts, then that power supply most likely has at least 5-10 years more life. 

 

The 100.000 is a MTBF value, not the number of hours the power supply is rated for or warrantied for.

Components like diodes and mosfets can last for 30 years+ of continuous use, the only components that can break are moving ones (fans, ex losing lubrication, plastic becoming brittle from heat/cold cycles, things falling down onto the blades etc)  and components like electrolytic capacitors which use semi-liquid electrolyte which can be chemically damaged over time or the rubber seal on the capacitors may break down after around 20-25 years. 

 

So the parts that can fail are the fan and electrolytic capacitors. Fan can be easily replaced even by yourself, just have to be careful to get one with similar amount of air flow (you have the part number printed on the fan sticker, so you just look in the datasheet and then get a fan with similar specifications)

 

Quality high end capacitors for power supplies are not expensive and can be bought at distributors of electronic components so you know they're not ebay fakes... would probably cost you around 10-20$ to buy the capacitors that are most likely to be "degraded" due to heat inside the power supply. 

You can take the capacitors and the power supply to any place that services laptops or phones and they'd be able to replace the capacitors for you, if you don't feel like doing it. 

 

I wouldn't rush to replace it. if it's not causing you problems now (ex random resets or stuff like that), then continue using it. 

 

I'd open it and visually inspect capacitors to see if they're fine (tops not swelling - that's sign the electrolyte produced gas and swollen the capacitor case, sign of capacitors going bad) and maybe I'd consider replacing the fan if it's making noises.. But that's just me, I'm biased a bit, as  I'm comfortable opening stuff because I know how to repair things and have the equipment to solder components. 

I can understand if someone else is afraid of opening a power supply or doesn't trust replacing components in a power supply.

Power supplies are more than a box of capacitors.  How do you suggest he checks all of the TIM under the MOSFETs to make sure they're not dried up after 10 years of use?

 

So you'd suggest waiting for the ripple to be so high or an SR MOSFET to blow up and take out some of his hardware instead of replacing it because "it works" now?

 

Wow.

[Dedayog]

8 minutes ago, jonnyGURU said:

Power supplies are more than a box of capacitors.  How do you suggest he checks all of the TIM under the MOSFETs to make sure they're not dried up after 10 years of use?

 

So you'd suggest waiting for the ripple to be so high or an SR MOSFET to blow up and take out some of his hardware instead of replacing it because "it works" now?

 

Wow.

I heartily agree with you, but playing devil's advocate... Why not design a PSU to fail or stop once the ripple gets bad?  Do they HAVE to blow up?

 

Give them a stated lifespan?  A turn off date, rather than we all just wait for them to fail or waste money replacing them while they still work well?  Invent a TIM that doesn't dry up in 10 years?

 

I guess since I deal with cars so much, I get tired of the "it's just a matter of time" mindset.  Make better shit, or figure out how to end-of-life stuff without catastrophe.

 

Granted I know shit all about PSU's, so I can dig it if there isn't a way to do any of that yet.  Just curious.

 

 

[Mister Woof]

17 minutes ago, Dedayog said:

I heartily agree with you, but playing devil's advocate... Why not design a PSU to fail or stop once the ripple gets bad?  Do they HAVE to blow up?

 

Give them a stated lifespan?  A turn off date, rather than we all just wait for them to fail or waste money replacing them while they still work well?  Invent a TIM that doesn't dry up in 10 years?

 

I guess since I deal with cars so much, I get tired of the "it's just a matter of time" mindset.  Make better shit, or figure out how to end-of-life stuff without catastrophe.

 

Granted I know shit all about PSU's, so I can dig it if there isn't a way to do any of that yet.  Just curious.

 

 

You mean making a mechanical part embedded under mountains of clips, panels, and wires with a piece of plastic instead of metal to save $.05?

 

Like that?

[seon123]

19 minutes ago, Dedayog said:

Why not design a PSU to fail or stop once the ripple gets bad? 

How would a PSU know? Are you proposing putting an oscilloscope into every single PSU...?

20 minutes ago, Dedayog said:

Give them a stated lifespan?  A turn off date, rather than we all just wait for them to fail or waste money replacing them while they still work well? 

Ah, great idea, a product that refuses to work after a certain amount of time. No one could possibly see any issue with that.

20 minutes ago, Dedayog said:

Make better shit, or figure out how to end-of-life stuff without catastrophe.

PSUs do have protections, and you'll typically see PSUs that just stop working.

[Dedayog]

1 minute ago, seon123 said:

How would a PSU know? Are you proposing putting an oscilloscope into every single PSU...?

Ah, great idea, a product that refuses to work after a certain amount of time. No one could possibly see any issue with that.

PSUs do have protections, and you'll typically see PSUs that just stop working.

Well the ripple would have a trip switch, like a.... protection?

 

I'm guessing, but you can't play both sides.  You can't say we can't protect against ripple and then tout protections.

 

Again, I don't know what to do but if it's not feasible then okay.  If the best answer is to replace a PSU when the warranty period is up (since those are calculated), then cool.

 

Yeah, I guess integrating a failure part isn't going to stop catastrophic failures beforehand.  But I would prefer a known date of end of life rather than knowingly throw out still working stuff.  Waste, ya know.  

 

Thanks.

[Elisis]

23 minutes ago, Dedayog said:

You can't say we can't protect against ripple and then tout protections.

Except... they can? Current protections work on very specific, (mostly) static thresholds. Ripple is much harder to catch and would likely increase cost considerably if you were to try and add a safeguard.

[Guest]

I suppose you could make a digital PSU that can somehow measure the output ripple and commit hara-kiri once it sees it's close to out of spec.  But I gather that would be quite expensive.

 

Still doesn't solve the problem of TIM drying up.  With CPUs and GPUs, assuming we keep them as long as we do PSUs, we can just reapply TIM.  That's not so easy with MOSFETs and bridge diodes.     That's why I don't like the full package FETs, but as some of you know, there's some peers out there that disagree with me in that there's new problems introduced like the metal of the FET corroding or a bunch of dust clogging up around the thermal pad.

 

 

[IIIIIIIIII]

1 hour ago, jonnyGURU said:

SR MOSFET to blow up and take out some of his hardware instead of replacing it because "it works" now?

15 minutes ago, jonnyGURU said:

FET corroding or a bunch of dust clogging up around the thermal pad.

Not the SR MOSFET... but what about this case? Could it be a similar cause and effect?

Edited November 16, 2021 by IIIIIIIIII

[Guest]

48 minutes ago, IIIIIIIIII said:

Not the SR MOSFET... but what about this case? Could it be a similar cause and effect?

That PSU doesn't work at all, though.  Right?

 

[dilpickle]

Jesus is this thread some viral marketing for hardware manufacturers? Who the hell "proactively" replaces parts while they are still working?

 

There are millions of people out there using hardware far older than this. Most of the parts in my PC are over 10 years old now including the PSU. It does what I need it to do and I don't see the need to replace anything.

[Elisis]

1 minute ago, dilpickle said:

Jesus is this thread some viral marketing for hardware manufacturers? Who the hell "proactively" replaces parts while they are still working?

 

People who don't want their PSUs to kill their hardware as a result of a secondary FET overheating and fusing closed, just a thought.

[Guest]

21 minutes ago, dilpickle said:

Jesus is this thread some viral marketing for hardware manufacturers? Who the hell "proactively" replaces parts while they are still working?

 

There are millions of people out there using hardware far older than this. Most of the parts in my PC are over 10 years old now including the PSU. It does what I need it to do and I don't see the need to replace anything.

You have no idea what you're talking about... but thank you for your contributions.

 

Things "work correctly" until they don't.  We're not talking about a car with a radiator leak here.

 

[Mister Woof]

1 hour ago, dilpickle said:

Jesus is this thread some viral marketing for hardware manufacturers? Who the hell "proactively" replaces parts while they are still working?

 

There are millions of people out there using hardware far older than this. Most of the parts in my PC are over 10 years old now including the PSU. It does what I need it to do and I don't see the need to replace anything.

I replace tires probably more often than is absolutely needed.

 

Don't mind one bit.

[dilpickle]

2 hours ago, jonnyGURU said:

 

 

Things "work correctly" until they don't. 

 

Oh really Sherlock? Try telling the businesses and schools with 10,000 PCs that they need to change all their power supplies every 5 years and watch them laugh in your face.

[Guest]

19 minutes ago, dilpickle said:

Oh really Sherlock? Try telling the businesses and schools with 10,000 PCs that they need to change all their power supplies every 5 years and watch them laugh in your face.

Ok. I will.  My company only use Dells for 10 years max.  5 years typically because they're usually too out of date.  My wife is a teacher and they got rid of those multi-colored iMac's 10 years ago.

 

They don't have IT people that repair.  Only replace.  Because if they repair, it's only a matter of time before other things fail.

 

So why are you being so enigmatic.  Who do you work for that uses their PC's for 15+ years?

[dilpickle]

12 minutes ago, jonnyGURU said:

Ok. I will.  My company only use Dells for 10 years max.  5 years typically because they're usually too out of date.  My wife is a teacher and they got rid of those multi-colored iMac's 10 years ago.

 

They don't have IT people that repair.  Only replace.  Because if they repair, it's only a matter of time before other things fail.

 

So why are you being so enigmatic.  Who do you work for that uses their PC's for 15+ years?

Who said anything about 15+?

 

But if you want to pay for some plane tickets I'll show you even older hardware.

[Guest]

35 minutes ago, dilpickle said:

Who said anything about 15+?

 

But if you want to pay for some plane tickets I'll show you even older hardware.

I have a ColecoVision that still works.

 

But times they change.

 

[Juular]

2 hours ago, dilpickle said:

Try telling the businesses and schools with 10,000 PCs that they need to change all their power supplies every 5 years and watch them laugh in your face.

Schools don't care because that's what they got and if it dies they would only be glad to write that shit off and tell the government/board that they need new PCs.

If your PC dies that's your problem. Granted, if the whole PC is exactly as old as that 13yo PSU then i don't really see the reason to get a new PSU either, if it dies then just throw the whole PC away. But if you're going to put some new hardware there, still keeping a 10yo+ PSU wouldn't be very smart. It may work, except it may have ripple to high that it would be slowly killing the rest of PC or it may 'suddenly' fail in a away so it kills something with it. You can put a mini DSO inside to control ripple, you can put thermal probes on every sinked semiconductor to control their operating temperature separately. But first, that's going to be more expensive, possible more expensive in the long run than just buying a good PSU for 10 years and then getting another one. Second, after 10 years chances are that your PSU wouldn't be fit for up-to-date hardware anyway. Third, PSU manufacturers only give about 10 years warranty max not only because of 'planned obsolence', even tho that's definitely one of the reasons. You simply can't ensure your product quality that far unless you'd be ready to also pay for increased support requests down the road, or you add this cost into the price of the product right away like Seasonic does for their Primes with 12y warranty. So you can make an insanely robust PSU with 20 years warranty which would be 4-5 times the cost of the 'regular' one, but who's gonna buy it ?

[blinkzz]

8 hours ago, dilpickle said:

Who said anything about 15+?

 

But if you want to pay for some plane tickets I'll show you even older hardware.

you are kinda arguing with the wrong people here 

[jaslion]

On 11/17/2021 at 5:40 AM, dilpickle said:

Oh really Sherlock? Try telling the businesses and schools with 10,000 PCs that they need to change all their power supplies every 5 years and watch them laugh in your face.

I do this? I literally did this task 2-3 weeks ago for a business once what I said was proven to be true after a psu sparked. I got 3 other contacts from that one task and for 2 businesses I'm swapping out psu's now as a pre-emptive measure because these are 7-8 year old cx green units for one and generic fsp ones for the other. Since these systems are supposed to run for another 4-6 years from now.

 

Old systems that are used often that still work are literally just running on luck and/or quality of the parts from back in the day. I have the ORIGINAL rm1000 psu and man I would not trust that thing for one bit in a modern system but it still works just fine powering any card. But I KNOW it can FAIL AT ANY TIME as it already has shown me that a rtx 3060ti can cause it to shut down.

 

Yes most companies and schools won't believe you till it starts happening that is just how those institutions are spend as less as possible is the motto.

 

This DOES NOT CHANGE THE FACT THAT PSUS AGE. Yeah sure in a office pc that has a rmx 650w I'd be totally fine to let that go a decade plus but if that was a gaming pc that was used on the daily I'd with no doubt swap it out when a upgrade happens around the 5 year mark to not run a risk if I know the psu would be pushed hard.