Discussion PSU recommendations and power supply discussion thread - Tom's hardware

Page 23 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Apr 4, 2019
59
18
35
If a PSU is rated for 450w, it ought to be able to sustain 450w with a gaming load.

If the PSU cannot do that, there is a problem.

And should be able to run with hardware that consumes less power than the rated wattage while not having its protections kick in.
 
Apr 4, 2019
59
18
35
even the old green label CX,

Not all greenlabels. 750w+ have the same internals as Cxm greylabem.

It's only the higher end, enthusiast class, pro editor, content creator type builds running high end gpus, massive cpus that need the appropriate psu, both in wattage and quality.

Unless you are running HEDT hardware or multiGPUs you rarely if ever need to exceed a good 550w.

And if you are gonna exceed 750, you should aim for multirail units.
 
  • Like
Reactions: NightHawkRMX

4745454b

Titan
Moderator
And if you are gonna exceed 750, you should aim for multirail units.

And why is that? I know of only one reason to use a multi rail unit, and that doesn't happen until around 1500W.

As for the green label, you went well past it's warranty. I don't remember if they were warrantied for 2 or 3 years, but you got four out of it. I would have stopped using it then. Before the fire I was starting to look at new PSUs because my Leadex based PSU was at the end of its warranty. Anything longer and you are playing with fire. Possibly literally.
 
Apr 4, 2019
59
18
35
And why is that? I know of only one reason to use a multi rail unit, and that doesn't happen until around 1500W.


reaction time/sensetivity during certain load scenarios to protect the PSU and the system its powering. and when you need more than a 750w PSU, you are within the reach of such scenarios.

As for the green label, you went well past it's warranty. I don't remember if they were warrantied for 2 or 3 years, but you got four out of it. I would have stopped using it then. Before the fire I was starting to look at new PSUs because my Leadex based PSU was at the end of its warranty. Anything longer and you are playing with fire. Possibly literally.

that is very paranoid to stopp using a PSU when its warranty ends....... as if a PSU has lasted that long, its likely to last for a long long time.

edit: also fires afaik only really happen in dissasterunits
edit2: if one is gonna be that paranoid regarding warranty, id especially be on the more conservative side regarding high wattage single-rail PSUs
 
Last edited:
It is not paranoid, it is preemptive, and it's not a bad thing. For the comparatively cheap cost of a decent new power supply compared to the replacement of a good motherboard, CPU or graphics card, or a large drive, it makes a lot of sense to replace a power supply that has reached it's warranty age if you still value the hardware it is powering by that point and then potentially move that still working PSU into a system that is either low value and not a big concern OR shelf it for a backup in the even that your replacement unit must be RMA'd. Not everybody's unit, after being shelved, blows up their motherboard. Although that's certainly a possibility and you might alternatively wish to simply throw it away.

I don't recommend continuing to use power supplies with hardware you care about much beyond the warranty UNLESS it is a unit that the manufacturer actually offered extended warranty coverage for beyond it's standard warranty. They don't usually do that unless they believe the platform and component selection is good enough to warrant it.
 

TJ Hooker

Titan
Ambassador
that is very paranoid to stopp using a PSU when its warranty ends....... as if a PSU has lasted that long, its likely to last for a long long time.
What are you basing this on? I would assume that PSUs follow the typical 'bathtub' curve for failures. It is in the manufacturer's best interest to set the warranty somewhere just before the knee point of the curve, i.e. where failure rate accelerates.

For instance, electrolytic caps are rated for a finite lifespan in operating hours.
 
Apr 4, 2019
59
18
35
For instance, electrolytic caps are rated for a finite lifespan in operating hours.

they are rated at a certain lifespan at a given temperature. usually a fairly high temperature you dont see in regular PSUs. meaning they will last longer than their rated lifecycle. could be a culprit for the warranty, but i doubt it.

What are you basing this on? I would assume that PSUs follow the typical 'bathtub' curve for failures. It is in the manufacturer's best interest to set the warranty somewhere just before the knee point of the curve, i.e. where failure rate accelerates.

if there was an issue with the PSU it would most likely have issues in the beginning or when it reaches a certain load scenario.

the issue with it following i failure rate curve is that the warranty is put there since day 1. its most likely a extent of time that given manufacturer want to give warranty for.

They don't usually do that unless they believe the platform and component selection is good enough to warrant it.

from what ive noticed, a lot of companies dont do extended warranty regardless of how good the plattform may be. and warranty really isnt the expected lifespan of a unit, its more how much money a company is willing to spend on maintaining a warranty service. and longer warranty is something you can do with a higher margin product.
 
It is ALSO something you would NOT do with a product you know you are going to have to watch come back in your door after a certain period of time, so again, using it beyond the warranty is poorly advised if you value the hardware. Sure, I have seen plenty of power supplies last years beyond the warranty. I've also seen plenty of them either not make it to the warranty's conclusion or die/start having problems not long after. Manufacturers know "about" how long the hardware is going to last based on testing and selected component expectations. If a company REALLY believes a unit will last 7-8 years, then they will set the warranty to 7 years, not 5. It is a selling point. If they think the unit can last ten, they are not going to set it to 7 when a competitor has a similarly priced unit with a 10 year warranty.

I have a feeling we've met you before, by a different name.
 
  • Like
Reactions: TJ Hooker

TJ Hooker

Titan
Ambassador
they are rated at a certain lifespan at a given temperature. usually a fairly high temperature you dont see in regular PSUs. meaning they will last longer than their rated lifecycle. could be a culprit for the warranty, but i doubt it.
When I referred to lifespan I was meant the rated lifespan after it had been appropriately derated (or whatever is the opposite of derating I guess) based on expected operating temps.

if there was an issue with the PSU it would most likely have issues in the beginning or when it reaches a certain load scenario.

the issue with it following i failure rate curve is that the warranty is put there since day 1. its most likely a extent of time that given manufacturer want to give warranty for.
Are you suggesting that manufacturers just pick a number for warranty with no idea how long the PSUs will actually last in practice? I would assume they do some sort of accelerated life testing, and/or take failure data from previous comparable units (or use industry resources for failure rates for different components), to come up with an estimate for how long a new unit can be expected to last.
 
Last edited:
In my opinion using an out of warranty isn't really an issue in many cases.

If you are bareily loading a quality PSU for many years, it should be fine out of warranty as it has not been "worn out".

As they do age i wouldn't load an old PSU with high end hardware tho.
 
Obviously, we are conversing about the use of systems with heavy demands. If somebody is using a five year warranty unit that is seven years old and has a 550w capacity, but is only running a basic system using integrated graphics or a slot powered graphics card, then it's probably not a major concern. Even so, once you leave the warranty period the chances that SOMETHING is going to go wrong with the unit jumps exponentially so it STILL might be a good idea to replace it. Capacitors really don't care that you are using only 250w out of the 550w it was designed to accommodate. They really only know that they are old and weepy.
 
Apr 4, 2019
59
18
35
It is a selling point

its also a cost, a cost you wouldnt put on lower margin products. can you afford to put a 7 year warranty on the 80$ PSU? you might not be able. so you drop the warranty period. under no fault of the PSU.

If they think the unit can last ten, they are not going to set it to 7 when a competitor has a similarly priced unit with a 10 year warranty.

they might do so if it gives them slightly more profit margin.

Are you suggesting that manufacturers just pick a number for warranty with no idea how long the PSUs will actually last in practice?

they might do so, but i more expect them to pick a number they can appropriatly give within their margins. they probably has some groundwork based on components lifespan and failure rates from manufacture.

As they do age i wouldn't load an old PSU with high end hardware tho.

just so you know, degredation of PSUs was debunked years ago. when it comes to ATX spec requirements of newer hardware, thats a different matter.

Im pretty sure i know who you are talking about
I have a feeling we've met you before, by a different name

id love to hear who, because im fairly certain im not the person you think i am.
 
Apr 4, 2019
59
18
35
Capacitors really don't care that you are using only 250w out of the 550w it was designed to accommodate. They really only know that they are old and weepy.

are you arguing capacitor degredation? because that was also debunked years ago. there is a liferating at a certain temperature sure, but thats at a temp a PSU usually doesnt run at.

Even so, once you leave the warranty period the chances that SOMETHING is going to go wrong with the unit jumps exponentially so it STILL might be a good idea to replace it.

if you like being under the blanket of a warranty, its not a terrible idea of replacing it. people do like comfort of not thinking about the PSU. not that a lot of people do, but there is allways some.

is it a major driving factor of swapping a PSU? not really.
 
just so you know, degredation of PSUs was debunked years ago. when it comes to ATX spec requirements of newer hardware, thats a different matter.
Source?

Capacitors do not age well. I have seen capacitors bulge and leak with age, especially in PC/monitor power supplies or even motherboard vrms. Solid capacitors help, but still wear.

Heck, capacitor leakage over time is a real problem. There were some apple computers back in the 80s which could have the motherboards completely eaten away by leaking caps.

PSU fans fail commonly. Especially low end sleeve or rifle bearing fans.

PSUs build with dust and can run hot after years.
 
  • Like
Reactions: TJ Hooker
Apr 4, 2019
59
18
35
PSU fans fail commonly. Especially low end sleeve or rifle bearing fans

its why OTP is more important that some people claim it is.

PSUs build with dust and can run hot after years.

even better reason why OTP is so important. but it also runs under the assumption that the PSU runs in a borderline state of shutdown for shorter than expected overtime of the rated lifespan of a cap.


i believe i got that claim from JonnyGuru, and/or other member of the PSU tierlist at LTT Forum.

Capacitors do not age well. I have seen capacitors bulge and leak with age, especially in PC/monitor power supplies or even motherboard vrms. Solid capacitors help, but still wear.

¨bloatet caps usually come from wrong input voltage if ive understood correctly. and that voltage would come from the PSU. such voltage issues either from Group reg (without UVP or OVP) or otherwise is from what ive understood one of the most common reason why caps bloat and die.

and unless you electrical grid is actually horrid, it doesnt really apply to PSUs as much as it applies to motherboards and GPUs.
 

TJ Hooker

Titan
Ambassador
are you arguing capacitor degredation? because that was also debunked years ago. there is a liferating at a certain temperature sure, but thats at a temp a PSU usually doesnt run at.
Ok, now I'm starting to think you're just a troll. Here's an electrolytic cap data sheet, take a look at the Endurance section under Specifications.
https://www.mouser.com/datasheet/2/315/ABA0000C1181-947564.pdf
If capacitor degradation wasn't a thing, why is the tolerance for capacitance greater after 2000 hours of operation greater compared to initial tolerance? Why is tan delta (dissipation factor) allowed to increase by up to double compared to initial value?
 
Last edited:
Apr 4, 2019
59
18
35
If capacitor degradation wasn't a thing, why is the tolerance for capacitance greater after 2000 hours of operation greater compared to initial tolerance? Why is tan delta (dissipation factor) allowed to increase by up to double compared to initial value?

probably because thats how the manufacturer expect the unit to behave over time. more insight i dont personally have over how this increase in tolerance effect the performance of cap.
 
Apr 4, 2019
59
18
35
So, they expect it to behave worse over time.

i dont have the insight in capacitor performance to know if that is what actually entails with the greater tolerances.

You just basically agreed that capacitors do degrade.

agreed that that is how the manufacturer expects the unit to behave, nothing extra.

and from what i can gather from someone i just doublechecked with. its values increasing doesnt entail the degredation of the cap, but more the death of the cap.

edit: and upon more info i got. the increase in tolerance isnt a big issue due ripple&noise being set low
 
Last edited:

TJ Hooker

Titan
Ambassador
agreed that that is how the manufacturer expects the unit to behave, nothing extra.

and from what i can gather from someone i just doublechecked with. its values increasing doesnt entail the degredation of the cap, but more the death of the cap.

edit: and upon more info i got. the increase in tolerance isnt a big issue due ripple&noise being set low
Please define exactly what constitutes "degradation". Because if decreasing capacitance and increasing ESR aren't degradation, what is?

How are ripple and noise "set low"?

By all means, let us know who your secret source is, or encourage them to chime in directly.