Sign in with
Sign up | Sign in
Your question
Solved

Upgrade: Replace old high quality PSU?

Tags:
  • Power Supplies
  • Components
Last response: in Components
Share
September 2, 2014 5:04:29 AM

Hello All!

I am wondering what the conventional wisdom is about how often a PSU should be replaced.

I currently have a PC (i7 960, HD5870) running a Corsair HX750W power supply that is getting onto 5 years old (along with the rest of the computer). Peak system power draw is <500W (I read 500W off my UPS when running furmark + prime 95, but my monitor, speakers, modem, WD NAS are also connected to UPS). I have also not been bitcoin mining/folding/etc. so the PSU has not been significantly stressed near to its rated limit.

I am intending to make some upgrades to the PC in the nearish future; is it safe to keep the same PSU for another few years? Or is it likely that even high quality PSUs will die after ~5-6 years and take the system with it?

Note: the upgrades I plan to do will not significantly increase the system power consumption, so the PSU size is not an issue.

Thanks!

More about : upgrade replace high quality psu

a b ) Power supply
September 2, 2014 5:14:19 AM

yeah your psu is quite a good one so it will last you even in your new build i'd say, noly when you get pwoer cuts from your pc do you need to upgrade, but as long as you haven't physically damaged it it should be fine
m
0
l
a c 2488 ) Power supply
September 2, 2014 6:28:43 AM

Corsair HX750W will be fine to use in your next build.
m
0
l
Related resources
September 2, 2014 6:14:43 PM

I know that the corsair HX750W is/was a good quality power supply, that isn't really the question.

My main question is:

Is even a good quality power supply likely to blow other components when it fails, and at what age should I start worrying about this happening?
m
0
l

Best solution

a c 2488 ) Power supply
September 2, 2014 6:18:22 PM

A great quality power supply like yours is less likely to take out other components.

You can monitor the health of your power supply by checking the voltages with Hwmonitor.

http://pcsupport.about.com/od/insidethepc/a/power-suppl...

Here's the ranges , they can also be checked in the BIOS for better accuracy.
Share
September 2, 2014 9:31:11 PM

Thank you for the information SR-71!

I tried looking at my voltages both with HWmonitor and BIOS, and they gave significantly different results (See these pictures http://imgur.com/kqqQyPC http://imgur.com/0ZSxZmq). BIOS showed 12V rail at 12.025V, while HWMonitor showed it at 11.821V while the PC was idling on desktop.

However, I did notice that the 12V rail dropped by 0.1V in HWMonitor when I did stress testing (Furmark burn-in + prime95 8 threads). This lead me to believe that perhaps the difference between BIOS and HWMonitor was from a change in the power draw between BIOS and desktop.

So... I decided to forgo software solutions altogether, and took out an analog voltmeter to directly measure the 12V and 5V rails off of a molex connector while my PC was running. Probably not the safest idea, but I didn't blow anything up this time (Yay?). The result was that, as far as i could tell, the 12V rail was exactly 12V, and did not change when stress testing was turned on. The 5V rail also seemed to read exactly 5V, but I didn't check it again with stress testing on. My reading uncertainty off that voltmeter is probably no better then +/- 0.1V however, so its hard to tell whether there really was a 0.1V drop or not when Furmark/prime95 was engaged. Unfortunately I don't have a fancy digital voltmeter handy to measure these more accurately at the moment.

In any case, based on your link indicating +/-5% is safe, all readings indicate my power supply is healthy.

I was wondering if you could elaborate on the significance on these readings however: If there is a drop, still within the 11.4-12.6V range, does this indicate my power supply is getting closer to dieing? Or will it likely just all of a sudden go out of range and this is a last-ditch warning sign to replace before sparks fly and plastic burns?
m
0
l
a c 2488 ) Power supply
September 2, 2014 9:34:55 PM

Go by the BIOS , software tends to be wrong.

From the readings in the BIOS it looks pretty healthy.
m
0
l
September 3, 2014 3:14:48 PM

Well, I borrowed a high quality digital voltmeter and checked the results again. The 5V and 12V readings were taken off of a 4pin molex connector that is on a cord hooked up to 3 HDDs, two optical disc drives, and four fans. Here is what I found:

In BIOS:
BIOS Reading DVM Reading
12V Rail 12.025V 12.11V
5V Rail 5.094V 5.106V

While booting windows:
DVM Reading
12V Rail 12.11V
5V Rail 5.109V

On windows Desktop (Mostly idle)
HWMonitor DVM Reading
12V Rail 11.821V 12.10V
5V Rail 4.896V 5.111V

During stress test (Furmark 100% GPU use, prime95 100% CPU use)
HWMonitor DVM Reading
12V Rail 11.710V 12.10V
5V Rail 4.848V 5.111V


A couple of things stand out.

First, SR-71, you are correct that the BIOS readings are more accurate than software, but they still are a bit off from the DVM on the 12V rail. The 5V rail reading I would say is within error.

Second, the HWmonitor readings are inaccurate. Not only are they a good 2% off the DVM readings in normal times, the 1% drop on both the 12V and 5V rail that HWMonitor reports does not seem to exist.
m
0
l
!