DC Output Quality

For those of you that are curious as to some of the reasoning and equipment behind our PSU testing program here at TheFPSReview, we have put together an introduction for you that shares a lot of the behind the scenes of the program. This program is based on what the author developed at [H]ardOCP and utilizes the equipment bequeathed to the author by Kyle Bennett. The testing we are conducting today is exactly as described in that document and will continue with our examination of the DC Output Quality.

Control Test Graphing

SilverStone NJ700 700W Fanless Power Supply DC Output Quality Control

This image is the blank background control test on an unused connector from our SM-8800 during the testing of the SilverStone NJ700. This lets us determine what the background noise looks like during testing.

120v Input

100v Input

Test #1 is equal to approximately 25% of the rated capacity of the SilverStone NJ700 at 45c. This makes Test #1 equal to 185W by loading the 12v rail to 13a, the 5v rail to 2a, the 3.3v rail to 1a, the +5vsb to 2a, and the -12v to 0.3a. The SilverStone NJ700 is starting off in excellent shape. All of the rails we monitor are peaking at ~10mV of ripple/noise.

120v Input

100v Input

Test #2 is equal to approximately 50% of the rated capacity of the SilverStone NJ700 at 45c. This makes Test #2 equal to 327W by loading the 12v rail to 24a, the 5v rail to 3a, the 3.3v rail to 2a, the +5vsb to 2a, and the -12v to 0.3a. Test #2 sees only minor changes as the 12v rail is coming in at ~20mV of ripple/noise while the minor rails are coming in at ~10mV of ripple/noise.

120v Input

100v Input

Test #3 is equal to approximately 75% of the rated capacity of SilverStone NJ700 at 45c. This makes Test #3 equal to 502W by loading the 12v rail to 37a, the 5v rail to 5a, the 3.3v rail to 4a, the +5vsb to 2a, and the -12v to 0.3a. Test #3 sees the 12v rail peaking at ~20mV of ripple/noise again while the minor rails are coming in at ~10mV of ripple/noise.

120v Input

100v Input

Test #4 is equal to approximately 100% of the rated capacity of the SilverStone NJ700 at 45c. This makes Test #4 equal to 702W by loading the 12v rail to 56a, the 5v rail to 1a, the 3.3v rail to 1a, the +5vsb to 2a, and the -12v to 0.3a. In the final regular test, we see the 12v rail peaking at ~25mV of ripple/noise while the minor rails are peaking at ~10mV of ripple/noise.

Torture Test

The Torture Test is equal to approximately 80% of the rated capacity of the SilverStone NJ700 at 45C. This makes the Torture Test equal 576W by loading the 12v rail to 42a, the 5v rail to 7a, the 3.3v rail to 5a, the +5vsb to 2a, and the -12v to 0.3a. At the end of the Torture Test, we see the 12v rail peaking at ~20mV of ripple/noise while the minor rails are peaking at ~10mV of ripple/noise.

DC Output Quality Summary

The overall DC Output Quality of the SilverStone NJ700 is excellent. The NJ700 ended up posting a peak trace amplitude on the 12v rail of ~25mV followed by ~10mV on the 3.3v rail and ~10mV on the 5v rail. The worst absolute value among these hits less than ~25% of the ATX12v specification limit. In a relative sense, this unit was better than the Antec NEO ECO GOLD ZEN 700W and the Cougar BXM 700. So, it seems that while this unit is quiet in audio sense it is also very quiet in an electrical sense! Let’s move on now and see how all of this wraps up!

Paul Johnson

Paul is a long time PC hobbyist and tech enthusiast having gotten his start when he broke his first C64 quickly followed by breaking his first IBM XT. Most notably however, for 12 years, he served as the...

Join the Conversation

14 Comments

  1. Good review. I think for me, once you mention Seasonic the rest is pretty well known. I hope they keep it up and other OEMs take note.
  2. WHATS OPP ON THIS? :ROFLMAO:


    Batman-Slapping-Robin-Meme-Explained.jpg

    I don’t want to have to get the load tester repaired any more than neccesary.

  3. haha I’m just busting chops.

    I gotcha. But, destructive testing of a power supply results in unexpected events. Each time I have to get the load tester repaired and recalibrated and certified it starts a $2k. So, not something I like to do.

  4. I gotcha. But, destructive testing of a power supply results in unexpected events. Each time I have to get the load tester repaired and recalibrated and certified it starts a $2k. So, not something I like to do.

    Well, in my opinion, OPP testing should not be destructive. Unless the PSU is malfunctioning or you just really wanted it to be and juiced it hard (like 480V 3p).

    Hitting a PSU with some overload, 125-150%, shouldn’t cause it to explode and should gracefully trigger OPP. Most of our industrial tests are to 150% of rated.

  5. Well, in my opinion, OPP testing should not be destructive. Unless the PSU is malfunctioning or you just really wanted it to be and juiced it hard (like 480V 3p).

    Hitting a PSU with some overload, 125-150%, shouldn’t cause it to explode and should gracefully trigger OPP. Most of our industrial tests are to 150% of rated.

    But, like I said before, there is not a spec. And without some sort of definition we have no idea where it is and where a catastrophic load will happen. So, until it is defined somewhere we can’t do a legitimate test for it.

  6. But, like I said before, there is not a spec. And without some sort of definition we have no idea where it is and where a catastrophic load will happen. So, until it is defined somewhere we can’t do a legitimate test for it.

    Ask the manufacturer, if they advertise the feature, it has a setpoint.

    That, or treat it like overclocking and just step it up slowly.

  7. Ask the manufacturer, if they advertise the feature, it has a setpoint.

    That, or treat it like overclocking and just step it up slowly.

    This is on the list of expensive things I’d like to see.

    Just note that they’re expensive, and as @Paul_Johnson notes above, prohibitively so.

    I’ll also say for the sake of perspective – while you want to know how a PSU behaves close to its limit, that’s your ‘reserve’. You’d usually want to leave say a 20% or more bit of ‘on paper’ headroom. Base your estimates on how hard you want to push the system and what components you expect, and add up the power draw, then leave some headroom above that.

    If you’re pushing a PSU close to its limit, you’ve made a.. miscalculation somewhere along the line :).

  8. Ask the manufacturer, if they advertise the feature, it has a setpoint.

    That, or treat it like overclocking and just step it up slowly.

    Well see, that is the problem. Most of the "manufacturers" are not manufacturers and don’t know (though SilverStone is not one of those that does not know). Also, I would have to have not just what they set it to but how they designed the protection. Then write a specific test for each and every unit that caters to their exact design. That means there would be no standardization to be able to compare units to. And without some sort of standardized definition we have no idea where a catastrophic load will happen. So, until it is defined somewhere we can’t do a legitimate test for it.

  9. What? You don’t like ground dwelling birds for your power supplies?

    Birds wouldn’t have been my first choice, but knowing that it is a bird name makes a lot more sense than some sort of jar one uses at night.

    My mind was drawn in the direction of chamber pots!

  10. My mind was drawn in the direction of chamber pots!

    Somehow – I blame Hunger Games – the name pattern sounded like a bird name. But like so many other names that are nominally English coming from countries where it isn’t a first language, I pretty much just accept stuff as it comes. I read plenty a pre-2000 motherboard manual too.

    So ‘chamber pot’ is fair game, IMO!

    On the other hand, with Gigabyte’s fiery ordeal still shaking out and questions on how such occurrences might be prevented, the ‘fanless’ nature of this Silverstone unit just now caught my eye.

Leave a comment