Page 1 of 2 12 LastLast
Results 1 to 10 of 11

Thread: Longevity of power supplies

  1. #1
    Join Date
    Mar 2013
    Posts
    13
    Thanks Thanks Given 
    1
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default Longevity of power supplies

    Have an XFX 750w Pro for 5 years now and wondering if its reasonable to expect it good for another 5-6 years. What factors most impact the longevity of power supplies?

    My no.1 assumed factor (aside from quality of parts) is heat. And caps are (again, assuming) most likely of components to fail. This is why I tend to go for more powerful units than I need. ie, if a 600-700w unit draws only half its power (my usage 300-350w) it should be cooler, less stressed, therefore longer lasting. Could be a faulty assumption, but would love to hear your views. Thanks.

  2. #2
    quest for silence's Avatar
    quest for silence is offline Silencer & heathen relic of a human (™ by OW)
    Join Date
    Sep 2013
    Location
    Italy
    Posts
    3,353
    Thanks Thanks Given 
    1,037
    Thanks Thanks Received 
    1
    Thanked in
    1 Post

    Default

    Heat is mainly due to conversion losses, so it mostly depends of relevant efficiency at actual power draw of the specific unit. With reference to that, just overkill is then pointless, IMHO.

    OTOH there's an article on [H] about the re-test of a Silverstone unit, which lost about 1/5 of its capacity over about 7 years, so some extra capacity might be even welcome in the long run. Might.
    Best, Luca

  3. #3
    Join Date
    Nov 2014
    Posts
    224
    Thanks Thanks Given 
    35
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    Quote Originally Posted by taco View Post
    ... I tend to go for more powerful units than I need. ie, if a 600-700w unit draws only half its power (my usage 300-350w) it should be cooler, less stressed, therefore longer lasting ...
    I concur.

    Looking at the HardOCP test of the Silverstone Olympia 1000 after ~7 years, at full 1000W load it was unable to complete the full load test @ 45C, and ripple (which was borderline on +12V & +5V to negin with) had gone out of spec.


    However at half its rated wattage (~500W) it still behaved fairly well:



    ^ Slight drops in +5VSB and efficiency, but much the same overall.





    ^ Ripple still within spec @ 500W after 7 years (even with Crapxons populating the secondary).

  4. #4
    Join Date
    Aug 2010
    Posts
    2,700
    Thanks Thanks Given 
    116
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    Note though that OP1000 is using inferior capacitors to XFX, so results for OP1000 may not carry over 1:1 to XFX. I'd expect XFX to fare better after same time passed, especially since baseline performance was more solid IIRC.

  5. #5
    Join Date
    Mar 2013
    Posts
    13
    Thanks Thanks Given 
    1
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    Quote Originally Posted by quest for silence View Post
    Heat is mainly due to conversion losses, so it mostly depends of relevant efficiency at actual power draw of the specific unit. With reference to that, just overkill is then pointless, IMHO.
    I was basically referring to the impact of heat on caps. Again caps, being the most likely of components to fail and most susceptible to heat. Its why we see caps with heat ratings, ie, 85c, 105c, etc. 'Overkill' PSUs are more likely to keep heat well below that if drawing only half their rated power at peak usage. This is my basic assumption.

  6. #6
    Join Date
    Mar 2013
    Posts
    13
    Thanks Thanks Given 
    1
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    Quote Originally Posted by TELVM View Post
    ^ Ripple still within spec @ 500W after 7 years (even with Crapxons populating the secondary).
    Ah, so ripple can affect caps as well. I knew it was bad for hardware in general, ie, motherboards, GPUs, etc. So it can impact a PSUs longevity as well?

  7. #7
    Join Date
    Aug 2010
    Posts
    2,700
    Thanks Thanks Given 
    116
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    The other way round: deteriorating caps affect ripple voltage as their ESR increases over time.

  8. #8
    Join Date
    Oct 2016
    Location
    Brazil
    Posts
    206
    Thanks Thanks Given 
    177
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    To be fair, the caps deteriorate faster when you have a not so-good design. And I believe this is the case of the mentioned SilverStone.

    The secondary caps get almost no airflow there because of the PCB's layout and location of the fan.

    And the primary part of this PSU doesn't seem to be very well designed, you can easily notice this looking to the difference of the performance on ripple between the 120 VAC and 100 VAC input tests.

    Also: http://www.jonnyguru.com/forums/showthread.php?t=13985

    So, I don't conccur with the OP's idea. Getting an well-designed PSU, with higher temperature ratings (50° C) and having a dedicated intake for it makes much more sense. Also, in the real life, you're using less than 100 W. You're only usinng 300, 400 W and so on only with heavily stressing applications. And those applications must use 100% of the CPU and GPU at the same time. Not a common scenario.

  9. #9
    Join Date
    Nov 2015
    Location
    U.S.A.
    Posts
    1,264
    Thanks Thanks Given 
    327
    Thanks Thanks Received 
    0
    Thanked in
    0 Posts

    Default

    Quote Originally Posted by rafal_iB_PL View Post
    The other way round: deteriorating caps affect ripple voltage as their ESR increases over time.
    The other way around applies, too. If you add more capacitors to a series it lowers ripple and each capacitor theoretically should then last longer.

  10. #10
    Join Date
    Jul 2009
    Location
    Germany
    Posts
    3,641
    Thanks Thanks Given 
    6
    Thanks Thanks Received 
    66
    Thanked in
    60 Posts

    Default

    And don't forget the load from the components that can also stress the caps of a PSU.
    SO it's not easy at all to say that a PSU is OK or not, after a longer period of time.

    Even if they use those "Japanese capacitors", they still might fail and/or be insufficient for various reasons...

    Especially in the last couple of years, the power consumption of graphics cards changed drastically and now is very ripplelastic...

Similar Threads

  1. New XFX XTR power supplies
    By SmokingCrop in forum PC Power Supply Discussion
    Replies: 5
    Last Post: 12-21-2013, 05:14 PM
  2. Power supplies degrade in power output over time?
    By dangman4ever in forum PC Power Supply Discussion
    Replies: 20
    Last Post: 11-19-2010, 10:26 AM
  3. K-Mex Power Supplies
    By Abu Som3a in forum PC Power Supply Discussion
    Replies: 24
    Last Post: 01-10-2009, 04:06 PM
  4. Balancing power noise, longevity
    By Grond in forum PC Power Supply Discussion
    Replies: 8
    Last Post: 11-19-2008, 02:40 AM
  5. PSU Longevity Ratings
    By Merc in forum PC Power Supply Discussion
    Replies: 6
    Last Post: 02-01-2007, 01:59 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •