Announcement

Collapse
No announcement yet.

PSU Load Testers - Designs?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    #21
    Re: PSU Load Testers - Designs?

    Especially when the load Goes quickly form a relatively small load to nearly max of the PSU's capacity. Its hard to properly test a PSU.
    regarding quoted sentence no.1:
    we need to say if we'll be testing conditions of a relatively small sample of hardcore gamers and simillar(IF THAT, do we have any numbers about increase in psu load between idle and full load of cpu and gpu at all? also i would say with psus of todays wattage this is less of an issue...) or we'll be measuring load of most users, which is mostly idle cpu and gpu time(surfing internet, doing docs and spreadsheets etc.).

    sentence no.2: indeed, only real test is putting psu in a pc and waiting(for years) to establish reliability and durability. everything else is not a real-life test and you might as well discard it.
    for example i was reading johhnyguru test of 'basement bargain' psus last night: nothing in that test suggests all of those psus won't work quite satisfactory for years for most people. full load they can't do for any time at all, but they won't be pushed that hard at all.
    reasoning is simple: if that test was real-life test those mfrs would be non-existant by today(who would buy psu that can't last few minutes?).
    but they do last because load is usually waaay lower than full so these mfrs of crap just survive.

    in the end this puts testing into grim perspective:
    first of all full-load tests show nothing about real world experience(just like any other way of accelerated testing).
    2nd, as pcbonez said 2-3years tests equally make lil sense(i mentioned people who "repair pcs": soon even such people won't exist as pcs will be so cheap you'll be just swapping them as soon as they break...).

    so as a conclusion you just have no good way to thoroughly test a psu.
    all you can do is inspect design and quality of components and hope for the best, or just buy laptop (only to find a smps in laptop's power adaptor???..hehe).

    Comment


      #22
      Re: PSU Load Testers - Designs?

      I agree the proof of the pudding is in the eating

      but I think you can use a crystal ball to gauge how well a psu will stand up to the test of time
      by looking at design and type of components used plus some testing to see how it reacts to certain conditions that in may encounter in the field.
      Also under more Ideal conditions for use
      that is, as you say still not totally conclusive

      The big unknown factor is the "abuse" it maybe subjected to in its working life like being run close to ratings poor ventilation etc whatever.

      there is crap out there like the wun hung low $10 500W special
      (that would be lucky to see 250W on a good day)

      So to some extent you get what you pay for.
      (or maybe over pay for)

      As has been pointed out before some, what are considered crap caps can last for a fair while depending on conditions.

      I have an old Enemax that still ok after 5 or so years but they have been known to turn their toes up in 2 or 3 years
      last time I check ESR of caps they were ok (not totally conclusive I know)

      Even well built supplies will fail in no time if extremely abused.

      thats just my opinion on it, others I know here are far more in touch with this

      Cheers
      Last edited by starfury1; 03-12-2008, 06:39 PM.
      You step into the Road, and if you don't keep your feet, there is no knowing where you may be swept off to." Bilbo Baggins ...

      Comment


        #23
        Re: PSU Load Testers - Designs?

        mpilchfamily, Thank you for the links. Has given me some new ideas.

        ~~~

        Not to anyone in particular:

        Personally I don't think testing a PSU with an end use in mind is an appropriate way to test a PSU.
        -
        You are assuming that you know what the person that reads the review needs a PSU for and in fact you don't. For all you know the person is looking for a good replacement for a 4 or 5 year old system. Assuming everyone that reads your review is building a brand new gaming system is rather stupid. You personally may be into that sort of thing but have limited yourself to a very small audience.

        A PSU should be tested to the spec that PSU was designed for.
        - That's the only way to give it a fair test.
        If it's ATX 1.3 then test it to ATX 1.3.
        If it's ATX 2.0 then test it to ATX 2.0.
        If it's ATX 2.2 then test it to ATX 2.2.

        Taking a ATX 1.3 and testing it to ATX 2.2 is inherently stupid.
        Have you ever heard: "Use the right tool for the job."?
        If you are going to test a tool it should be tested for the job it was designed for, not tested for how well it does some other job.
        -
        While johhnyguru does a better job than most, he tests PSU's to specs that are N/A to the design of the PSU quite frequently. He even jokes about it. This is a huge beef I have with the way he tests PSU.
        -
        It's like, when the label on the side of the thing says whatever ATX spec and 15 amps max on some rail and then you test it to some other ATX spec and crank that rail up to 18 amps in test 2 or 3 and it blows out...
        - WELL, NO SHIT SHERLOCK!
        - Do you use a screwdriver for a jack handle too?

        .
        Mann-Made Global Warming.
        - We should be more concerned about the Intellectual Climate.

        -
        Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.

        - Dr Seuss
        -
        You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
        -

        Comment


          #24
          Re: PSU Load Testers - Designs?

          I agree thought too
          with most review sites it doesn't give you a great indication of expected life and anyone reviewing really needs to know their "nuts and bolts" in this respect to give a truly informative opinion on a PSU

          Its one thing to try and confirm given specs as a lot of review sites tend to do (nothing wrong with that)
          its totally another to offer a true opinion of build quality and design and how much abuse a supply is liable to with stand.

          I know there are people who frequent this forum that could literally tell at a glance if they are dealing with a decent well made unit or a heap of crap.

          I unfortunately (at this point anyway) aren't one of them.

          Cheers

          we posted at near the same time Pcbonez
          You step into the Road, and if you don't keep your feet, there is no knowing where you may be swept off to." Bilbo Baggins ...

          Comment


            #25
            Re: PSU Load Testers - Designs?

            starfury, I think you know more than you think you know.

            ~~~~

            One of the big HUGE things a reviewer of PSU's should give in relation to reliability is what make and what series of caps are installed.

            All a load test tells you is it that it can manage to start up your system on the first day and it might last a few hours or weeks.

            A PSU with caps that are total CRAP can get through a load test just fine, it will start the system just fine, but it might well DIE in a very short time and it could take your motherboard out with it.
            -
            My favorite example: Antec TP2 (and other) PSU's passed all sorts of load tests and got glowing reviews from everyone that tested them. - But they shit themselves faster that $15 PSUs and DO kill motherboards on the way out.
            (Somewhere I have photos of Antec TP2 550w carnage including a hole blown in a MOSFET and a melted chipset.)

            If you review a PSU and you don't report the specifics of the caps then you aren't doing a complete review. - There is no exception.

            In not doing so, you may even be screwing over some of your readers.
            - Don't count on them taking your advice again!
            - Don't count on them being return visitors to your site!
            - Don't count on them telling their buddies: "hey go read here".

            .
            Mann-Made Global Warming.
            - We should be more concerned about the Intellectual Climate.

            -
            Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.

            - Dr Seuss
            -
            You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
            -

            Comment


              #26
              Re: PSU Load Testers - Designs?

              starfury, indeed you're too humble.

              and then you test it to some other ATX spec and crank that rail up to 18 amps in test 2 or 3 and it blows out...
              i would be interested to know what that blow did to output, but that most important thing is something they all skip.
              you have info about blown switchers, but about output nothing.
              benefit of loading it with laods that can take almost anything.
              if load was a mobo that fried they would be monitoring the event more closely.

              to that extent: where are the elements that would limit the primary current in such violent events that blow the switchers. is anybody putting those? (fuses they put are obviously not cutting it)
              meh....
              situation is not good if one must rely on caps at all: one must have extra-layer of protection. both on primary AND secondary side.
              such design would be what i would call "nice". and worth paying for. 2, 3, 4 times the "usual" price, doesn't matter.

              sometimes i think linear supply the size and the weight of a tank would be good thing!
              hehe...

              btw. what caps does Antec TP2 use?
              Last edited by i4004; 03-12-2008, 08:30 PM.

              Comment


                #27
                Re: PSU Load Testers - Designs?

                famous fuh, it seems
                http://www.phoronix.net/image.php?id..._tp2_open2_lrg
                ( http://www.phoronix.com/scan.php?pag...item=171&num=2 )

                Comment


                  #28
                  Re: PSU Load Testers - Designs?

                  Antec True Power 2 was examined assiduously!!! No Cap mention just visual. Guess that is all we can expect.
                  Last edited by davmax; 03-13-2008, 12:44 AM.
                  Gigabyte EP45-DS3L Ultra Reliable (Power saver)
                  Intel E8400 (3000Mhz) Bios temps. 4096Mb 800Mhz DDR2 Corsair XMS2 4-4-4-12
                  160Gb WD SATAII Server grade
                  Nvidia 8500GT 256Mb
                  160Gb WD eSATAII Server grade for backup.
                  Samsung 18x DVD writer
                  Pioneer 16x DVD writer + 6x Dual layer
                  33 way card reader
                  Windows XP Pro SP3
                  Thermaltake Matrix case with 430W Silent Power
                  17" Benq FP737s LCD monitor
                  HP Officejet Pro K5300 with refillable tanks

                  Comment


                    #29
                    Re: PSU Load Testers - Designs?

                    There is general agreement that to test a PSU for reliability is very difficult indeed. The aircraft and construction industries can test to destruction and we all benefit. Difficult to see how PSU testing to destruction is going to help.

                    Few of us would see it worthwhile creating banks of resistors and switches just to prove manufacturers power claims.

                    There is some merit in a simple PSU tester to make sure all voltages are correct before risking a system.

                    Most users would probably wish to go through a set of steps something like this:

                    1. Purchase a PSU that will have more than enough power. Deciding what power level needs some research.
                    2. Make sure quality capacitors are used in the PSU. Other components rarely fail. (Possible warranty risk).
                    3. If possible verify all voltages of PSU are correct before risking system.
                    4. Measure the PSU loading for system idle and system 100% loading using an AC ammeter in the AC line. To achieve 100% run software such as BurnIn Test which loads all components and run for at least 10 minutes (preferably watch temperatures with monitoring software).

                    Test 4 will not only check the whole system functional reliability it will enable an assessment of how stressed the PSU is under 100% load and thereby give a measure of how reliable you can expect it to be. (running most things at 100% long term is going to reduce life eg flog your car at max rpm continuously and it will die pretty quickly).

                    Example.

                    I tested my system (my signature).
                    At Idle AC current = 0.3Amps @ 230VAC = 70VA (70watts with PF correction).
                    At 100% AC current = 0.43 = 100watts. (Input power)

                    ( With 100% on both CPU the temp rise from idle is 10C)

                    The PSU is rated at 430watts so running rarely at 100% there is low PSU stress. I feel comfortable that a long life can be expected.
                    Last edited by davmax; 03-13-2008, 01:37 AM.
                    Gigabyte EP45-DS3L Ultra Reliable (Power saver)
                    Intel E8400 (3000Mhz) Bios temps. 4096Mb 800Mhz DDR2 Corsair XMS2 4-4-4-12
                    160Gb WD SATAII Server grade
                    Nvidia 8500GT 256Mb
                    160Gb WD eSATAII Server grade for backup.
                    Samsung 18x DVD writer
                    Pioneer 16x DVD writer + 6x Dual layer
                    33 way card reader
                    Windows XP Pro SP3
                    Thermaltake Matrix case with 430W Silent Power
                    17" Benq FP737s LCD monitor
                    HP Officejet Pro K5300 with refillable tanks

                    Comment


                      #30
                      Re: PSU Load Testers - Designs?

                      Originally posted by davmax
                      Antec True Power 2 was examined assiduously!!! No Cap mention just visual. Guess that is all we can expect.
                      i see a bunch of fuhkkyu's
                      nuff said.

                      Comment


                        #31
                        Re: PSU Load Testers - Designs?

                        Gosh thanks guys

                        I guess how far you want to take a Dummy load setup is really up to the individual.

                        But I think a useful item to have if repairing psu's


                        The supply for this build is over the top really but thats the reason I went that way, so it would be ample therefore not under much stress and hopefully last without issues...thats the game plan.... time will tell thought

                        decent protection is an important point
                        bad enough when your supply decides to do a 4th of July but taking your system with it is just not good.
                        I don't think enough attention is paid to the "What IF Factor" in reviews

                        Yes it seems a lot of the issues are caps in supplies
                        (except maybe in the case of Liberties)

                        Anyway

                        Cheers
                        You step into the Road, and if you don't keep your feet, there is no knowing where you may be swept off to." Bilbo Baggins ...

                        Comment


                          #32
                          Re: PSU Load Testers - Designs?

                          Yes,
                          There are a lot of different reasons for a PSU tester and many different kinds.

                          All I personally want to do is to load all the rails to some % of their rating (probably 20% and 80%) and then measure the ripple with an O'scope.
                          - I would be looking at the effectiveness of mods to the output filters.

                          Don't need an particularly advanced tester for that but more features might come in handy for something else.

                          Like I said, is a back burner project.
                          Just looking for interesting ideas at this point.

                          .
                          Mann-Made Global Warming.
                          - We should be more concerned about the Intellectual Climate.

                          -
                          Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.

                          - Dr Seuss
                          -
                          You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
                          -

                          Comment


                            #33
                            Re: PSU Load Testers - Designs?

                            The issue of reliability beyond testing has been raised and testing a PSU in a 100% loaded system.

                            What is the consensus of how far a power supply should be pushed for long term reliability?

                            I'll put my hat in the ring and suggest:

                            For light user push 50 -70% of PSU rating when 100% loaded.

                            For heavy user with long periods close to 100%, 30 -50% of PSU rating when @ 100%

                            What do you think?
                            Gigabyte EP45-DS3L Ultra Reliable (Power saver)
                            Intel E8400 (3000Mhz) Bios temps. 4096Mb 800Mhz DDR2 Corsair XMS2 4-4-4-12
                            160Gb WD SATAII Server grade
                            Nvidia 8500GT 256Mb
                            160Gb WD eSATAII Server grade for backup.
                            Samsung 18x DVD writer
                            Pioneer 16x DVD writer + 6x Dual layer
                            33 way card reader
                            Windows XP Pro SP3
                            Thermaltake Matrix case with 430W Silent Power
                            17" Benq FP737s LCD monitor
                            HP Officejet Pro K5300 with refillable tanks

                            Comment


                              #34
                              Re: PSU Load Testers - Designs?

                              Interesting question.

                              Here is what I do:
                              I look up an Intel board that has the same chipset, CPU's, number of slots, and similar chips (LAN, sound, so forth).
                              The Intel documentation gives the max amps for each rail based on the current capacity of the circuitry on the board. [It either assumes all slots are filled or gives the max amps for the slots which can be added in. Same with memory.]
                              I then add the max amps for the expected loads that do not pass through the motherboard. (primarily drives). I assume and calculate for the max # of drives I will ever use in it.
                              -
                              With that info I calculate the amps for each rail, add 10%, and then fail-up.
                              That is the minimum amps for the PSU I'd want and more is better.

                              Couple examples:

                              An i845 uATX board I did for 2 optical and 3 HDD came in at 310w + 10% = 341 watts. (This would be the bare minimum and I check suitability of a PSU by each rail not total watts.) By fail-up I mean I would be looking for at least a 350 watt for that one. Next standard size up from my numbers. [Got a 400w]

                              An E7501 EATX dual Xeon board for 6 HDD came in at 578w + 10% = 636 watts. Would fail-up at least to 650 watts. [Got a 700w]

                              Now, all that estimating is based on full/peak load with all the memory and add-in slots filled and it's not going to be peaked often once it's built. But I know I have enough margin to add-in more cards or memory later without re-evaluating the power supply and I know that even after I do any upgrades that I'm not going to exceed the PSU ratings during a cold start-up.

                              My guess is that puts me at 70-85% when I'm running the system hard and probably around 50% normally. I run them all the time, 12-16 hours a day, but not hard very often.


                              .
                              And you AMD guys, sorry, I dunno.
                              Intel doesn't make AMD boards for,, um,, some reason. ???

                              .

                              .
                              Mann-Made Global Warming.
                              - We should be more concerned about the Intellectual Climate.

                              -
                              Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.

                              - Dr Seuss
                              -
                              You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
                              -

                              Comment

                              Working...
                              X