[SOLVED] How long is a PSU Expected to last - MTBF vs. Years Warranty?

[avg9956]

So my current power supply I think is over 11 years old now. Its a Thermaltake XT 875W (I'm surprised when I read the PSU Recommendations: https://forums.tomshardware.com/thr...pply-discussion-thread-toms-hardware.3212332/) that Thermaltake is a brand in general to avoid albeit I wonder how the brand in general (as of late 2020) is currently doing because I rarely swap power supplies due to the one I have right now for so long...

I have already ordered a Seasonic Prime Platinum PSU 1000W SSR-1000PD (This PSU has a MBTF of 150,000 hours according to this website)
Due to covid pandemic, its getting hard to get quality PSUs too on my area so supply is limited. I have never tried Seasonic before and this will be my first time, but from what I have consistently read across all the board it is a good brand. I would have gone for Corsair too, but was swayed away when they used Chinese CapXon caps on one of their high end models in some Johnny Guru forums post (I can link later if you guys are curious).

The Thermaltake PSU has 5 Years warranty that is labeled in its box
In the manual, the Thermaltake PSU has a MBTF of > 120,000 hours - (13.69 years) - I don't think all PSU manufacturers indicate this, but fortunately when I read the manual it was there.

So - how long is a PSU expected to last based on the years warranty indicated on the box and MBTF indicated on the manual?
At what point should you replace the power supply between 5 years and 13.69 years? or can you really go over 13.69 years?
(There maybe no definitive answer, but I would like to have ideas about these two metrics).

Now here below are some of my speculations, of course nothing can really be set in stone because there are a lot of factors that can affect the life expectancy of your PSU. But I want to go to the purchasing stage scenario where you make a Purchase decision for PSU based on their lifetime expectancy - only visible information such as the number of years warranty indicated in the PSU box and the MBTF (if you can get it off from the manufacturer's website or the manual for the PSU hosted online). All other factors such as customer reviews of the model, reading info on websites that actually test the PSU and open its insides (such as tweaktown), etc.

Generally if I see a PSU that has a high year warranty on the box, it must be good (at least 5 years or more). Maybe I should up my standards a bit to 7 years or more...

If the year warranty on the box is close to the value of the MBTF (example: 7 years warranty, MBTF ~7 years) , then it kind of hints to the consumer that you should replace the PSU after that much years has passed?

Likewise, if there is a larger difference between the year warranty indicated on the box compared to the MBTF (ex: 5 year warranty on box vs. 14 years MBTF), then it would mean that the PSU's life expectancy isn't consistent and the expected window of failure is between 5-14 years?

Also, I calculated the difference between the MBTF and the number of years passed that I have been using my power supply (13-11) = 2 years. Would it be ok to keep using it for a year or two or I made the right call to replace the PSU right now?

I cannot find too much info on PSU MBTF vs. Years warranty on box other than this thread from another website, but I would like to expound on it more.

 

[Lutfij]

MTBF is something that the company shows on it's packaging to entice customers. Often times this ifs falsified while some give it a ballpark area. If you do some math, you're going to see that it's impossible for a product to be in testing for maybe 5 years(the suggested warranty period for the unit) instead the testing is sped up or the taxing on the unit is simulated to represent the 5 years(some more or less).

That MTBF will often times correspond with the years of warranty given to the end user.

On that note, of the brands you've mentioned above, only Seasonic is the only brand to be an OEM for other brands. Thermaltake and Corsair source their units from others and slap a sticker on them(or often times put in a design that they will hold unique to the brand).

You also need to factor in that not all system's that are used for testing are the same system you're going to build/own since some system's will draw less or more power depending on parts used for the build, not to mention the use case scenarios for said build + PSU. Miners will have their system's run at high temps for long periods of time, non stop, which is another thing to look at...the temp graph that shows the effective power output from the PSU.

Lastly, the more stress you throw on the PSU, the closer you're reaching to that MTBF...think of it like a car in an accident. A minor ding doesn't affect the car. Get into a T-bone and then you're looking at scrapping the entire chassis or even getting it re-aligned.

 

[alceryes]

I ran an 800W BFG PSU for over 10 years. It didn't even die on me. When I upgraded to my Vega 64 GPU it started having some coil whine from the extra power draw so I decided to get my current RM 750x PSU. I still have it in storage as a backup.
If it's a good name brand PSU, use it until you can't anymore (it becomes insufficient, develops coil whine, dies, etc.) or it becomes suspect due to extended age. There's no set rule per se.

 

[avg9956]

Interesting
As per my current computer setup, I don't overclock. I use a UPS and I have the computer turned on around 14 hrs a day on average. I also have a spare UPS that I can use just in case my current UPS dies out. I clean my pc once every 2-3 months (I'm fan cooled, not liquid cooled) and follow all the safety precautions to avoid ESD.

I was looking for a good brand PSU that has the same amount of watts as my current PSU, but with no luck (I've tried for weeks already) and I'm forced to get a 1000W unit instead. With more load allowance from the newer PSU I think this will translate to less heat produced = longer lifespan. Sellers tried to coax me with brands such as Cooler Master or Gigabyte but I knew my brand research better. Its better than nothing though, because we all don't know how long till this covid pandemic will end. The only thing that kinda pushed me to buy is because it may become out of stock as well just as Nvidia cards did in the midst of the pandemic and my PSU is well over a decade old already.

 

[nostall]

For an idea of how well your Thermaltake 875 is running/performing, download and run HWInfo64. If you have a second monitor you can watch your 12v, 5v, & 3.3v performance while using the computer; or click on "record" and let it create a record for your review. I just replaced an older Corsair AX850(Gold) that was 9 years old because of serious voltage swings (5-11%) on the rails - it was top-of-the-line when I got it, but even it wore out.

 

[USAFRet]

how long is a PSU expected to last

MTBF and warranty is based on thousands of devices, and what the manufacturer expects. The vast majority will last beyond one or both of those.

Your individual device, however....that may die tomorrow.

 

[avg9956]

I tried running HWinfo64 to find out about my PSU performance but it seems I can't find any info about it
Experiencing same issue just as this guy

 

[jonnyguru]

MTBF vs. Years Warranty

Neither.

MTBF is mean time BETWEEN failures.

Warranty is marketing. They factor in how much money they lose supporting the product and amortize dead units into their profit margin. If they make a high profit margin (because they're the manufacturer, for example) and sell a lot of units, then they can afford to extend the warranty to 10 or 12 years, scrap more units and take the hit against their profit.

DMTBF (DEMONSTRATED mean time betwen failure) is a more accurate determination, but consumers don't get that data because, unless EVERYONE shared that data, it may actually scare off customers. Some lower tier manufacturers don't even bother with DMTBF because you end up scrapping at least 40 units per wattage and spending about $4000 (again, per wattage) in electricity during the process.

 

[jonnyguru]

I tried running HWinfo64 to find out about my PSU performance but it seems I can't find any info about it
Experiencing same issue just as this guy

That's because HWInfo doesn't get any data from your power supply unless your power supply has a PMBUS, which it definitely DOES NOT.

 

[refillable]

I think Jon explained it well. MTBF and warranty is not the same. What I'd like to add is that It's hard to quantify things just by brand alone. You can't say brand X is better than Y because of reasons, because brands often has both crap PSUs and great PSUs. Seasonic has their own problems. What matters most is the internals of the PSU, and of course, its performance.

 

[DSzymborski]

Thermaltake is a really mixed bag across-the-board. The Toughpowers tend to be fine, everything else in their PSUs tends to be mediocre or worse. It's a theme that goes across their products, really. They have some really cool, imaginative cases, but they also have some metal-and-glass chunks that have awful thermals.

Steve over at GamersNexus really took their engineering/marketing to task a little while ago.

View: https://www.youtube.com/watch?v=vhkYcO1VxOk

 

[avg9956]

The Toughpowers tend to be fine, everything else in their PSUs tends to be mediocre or worse.

I agree. Actually I bought that toughpower supply model a decade ago and I even had absolutely NO idea of a PSU Tier list or what Thermaltake PSU model was good or bad. I didn't even know what brands were good, or what Japanese capacitors were, etc.

I remembered the salesman offering me that PSU model. Its through just sheer luck that I made the right choice buying that PSU. However its time to try another brand. The Seasonic Prime Platinum PSU 1000W SSR-1000PD had just arrived and I will be installing it in the near future (still deciding on CPU and motherboard). Its about time I do an upgrade. Its been 5 years now since my PC is running without a reformat.

I guess they don't make them like they used to - as the saying goes. It's kind of sad for me as a Thermaltake customer to see that after a decade, the brand wasn't able to build more high enough quality PSUs for consumers and become par with Corsair or Seasonic, but at least I could say I'm equally blessed to not have my PSU fail for a decade an opportunity that is seldom achieved.

 

[DSzymborski]

I agree. Actually I bought that toughpower supply model a decade ago and I even had absolutely NO idea of a PSU Tier list or what Thermaltake PSU model was good or bad. I didn't even know what brands were good, or what Japanese capacitors were, etc.

I remembered the salesman offering me that PSU model. Its through just sheer luck that I made the right choice buying that PSU. However its time to try another brand. The Seasonic Prime Platinum PSU 1000W SSR-1000PD had just arrived and I will be installing it in the near future (still deciding on CPU and motherboard). Its about time I do an upgrade. Its been 5 years now since my PC is running without a reformat.

I guess they don't make them like they used to - as the saying goes. It's kind of sad for me as a Thermaltake customer to see that after a decade, the brand wasn't able to build more high enough quality PSUs for consumers and become par with Corsair or Seasonic, but at least I could say I'm equally blessed to not have my PSU fail for a decade an opportunity that is seldom achieved.

It was their budget side. Lots of awful TR2s and Litepowers made to cut every corner possible. I would probably guess yours would be a Tier B on our tier list today; it's not a particularly modern platform, but at least it's not group-regulated. I don't think this one is even one as it wasn't a big seller and this particular model hasn't been made in more than a decade. Even the most elite PSU of this age, I'd be easing into retirement now anyway.

 

[OrlyP]

What a timely thread. I was about to ask a similar question.

Anyway, the background is that I have an Antec EarthWatts EA-450 Platinum from 2014. It's been through two motherboard upgrades and currently it's supporting the following :

Xeon E3-1285v4
Gigabyte Z97-D3H
Palit GTX1050 Ti
32GB DDR3 (4x 8GB)
IBM ServeRAID m1015 SAS/SATA controller
10x SATA HDDs
1x SATA SSD

I should add that this has been running 24x7 in the last 7 years. Overall, this "server" has been stable in any type of workload I dish out.

As with the tip above, I ran HWInfo and at idle, the 12V rail sits at 12.024V. When the system is at 100% utilization, running Folding@Home, it dips down to 11.592V.

Is the voltage drop still acceptable? And other than a catastrophic failure of the PSU, what other tell-tale signs would signify that it's time to replace it?

 

[DSzymborski]

I wouldn't worry, simply because desktop software is really bad at properly reading voltages. Just right now, software is insisting one of my file servers has its +12V rail at 9.5. There's no way it's true.

 

[OrlyP]

I wouldn't worry, simply because desktop software is really bad at properly reading voltages. Just right now, software is insisting one of my file servers has its +12V rail at 9.5. There's no way it's true.

That is true. You gave me the idea to use a multimeter this time... I'm reading 12.1V idle and 11.9-12.0V loaded. I don't know what the ripple voltage is but I guess this is still a pass, voltage-wise.

I do hear the PSU fan becoming slightly noisier in recent months. The best I could describe it is a quiet "swishing" sound relative to the fan's RPM. Worn sleeves probably. Gives credence to the fact that it's usually the moving parts that fail first.

 

[avg9956]

PSU fans can be replaced afaik, you can do it yourself if you're confident on the hands on and know what fan to replace it with or ask a technician to do it for you as it involves opening the PSU.

Replacing a PSU fan may not necessarily be as easy as replacing your computer case fans. This depends on the design of the PSU - especially if the fan is soldered. Also note that opening your PSU in general would of course, break off its warranty seal to denote that it has been tampered internally.

Computer Power Supply Fan Replacement

Computer Power Supply Fan Replacement: This Instructable describes how to replace the fan inside a standard PC power supply. You may want to do this because the fan is defective, or to install a different type of fan, for example, an illuminated one. In my case, I decided to replace the…

www.instructables.com

I wouldn't advise opening PSUs especially if you're a beginner, but if you're prepared well for it then possibly give it a try...

 

[OrlyP]

Thanks. My concern was strictly regarding the PSU's overall circuit lifespan... ie. how well it can keep its regulation tight, etc., considering its age. In other words, there's no point in replacing the fan if the other important bits are beginning to go out-of-spec. Might as well replace the whole thing.

Anyway, given that mine seems to be well within spec, I was indeed planning on servicing or replacing the PSU fan. If lubricating it doesn't work, I've got a couple of spare Noctua P12s I could replace it with. I've done it before on a different PSU for a family member.

 

[jonnyguru]

I'm reading 12.1V idle and 11.9-12.0V loaded.

Using a multimeter?

I don't know what the ripple voltage is but I guess this is still a pass, voltage-wise.

So you're ok with frying your components.

Anyway, given that mine seems to be well within spec

Based on what? Did you have a chance to hook it up to an ATE and oscilliscope?

I've got a couple of spare Noctua P12s I could replace it with. I've done it before on a different PSU for a family member.

And that PSU still works? Must not be under much stress.

 

[OrlyP]

Hi Jonnyguru.. love your videos. Learned much from them.

But to your questions... no, I don't have the specialty test equipment that you have. All I had was a multimeter (well, 2 of them giving me the same results).... and I know, it's just practically averaging the voltage output.

I've had my share of PSUs giving up the magic sparkles and smoke but I haven't had a PSU take out PC parts with it.... not once in the 30+ years of fiddling. But I do know that there's always a chance, however slim.... hence, I joined this thread to ask for tell-tale signs of a failing PSU that might need proactive replacing. But at the same time, I don't want to replace one unnecessarily.

On ripple voltage, I'd like to think that if it gets worse enough, stability would take a hit first before PC parts starts to fail. And I've tested this server on full load over several hours recently, and over several weeks with Folding@Home a few months back, and not one hiccup surfaced. On most days however, this (Plex) server is just idling by.

So, barring any access to precision test equipment... it's not like I could just bring it to a corner PSU pit stop to have it tested and recertified... wouldn't the above real-world testing results be correctly construed as a reasonable 'pass'?

 

[jonnyguru]

Which EarthWatts EA450 do you have? Bronze, Gold or Platinum?\

Also, ripple wouldn't impact stability. It just ends up making the caps on the motherboard and graphics card swell until they fail.

 

[OrlyP]

Which EarthWatts EA450 do you have? Bronze, Gold or Platinum?\

Also, ripple wouldn't impact stability. It just ends up making the caps on the motherboard and graphics card swell until they fail.

Thank you. It's Platinum.

 

[jonnyguru]

Thank you. It's Platinum.

Ok. So it's about 9 years old. That's why the fan is dying. The PSU itself might only last another couple years.

 

[OrlyP]

I

Ok. So it's about 9 years old. That's why the fan is dying. The PSU itself might only last another couple years.

7 years from purchase. But I get your point. It is running 24x7.