Darkbreeze :
I don't see an in depth review for any of those three models anywhere, do you? So how do we assume they are worthwhile units?
Vengeance models have been reviewed and CXM models are just Vengeance models with a single rail design instead of multi rail and not all Japanese caps (the Vengeance are all Japanese).
With the reviews that Tomshardware puts out these days, I don't think Jonnyguru cuts it with the testing done anymore. Here is the Silverstone ST80F-Ti review on Tomshardware
http://www.tomshardware.com/reviews/silverstone-strider-titanium-st80f-ti-psu,4600.html
And the one on Jonnyguru:
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=478
Jonnyguru has a limited budget, so it is understandable that it only covers:
1) Progesssive load regulation
2) two crossload scenarios
3) turn-on transient tests
4) ripple
5) internal analysis
Tomshardware/TPU covers:
1) Transient response testing
2) PWR_OK and holdup time testing
3) literally hundreds of automatically run crossloading scenarios and how the corresponding voltages, ripple, and efficiency react to those hundreds (might even be more than a thousand) load scenarios.
4) Infared shots at the internals
5) In general, an even deeper internal analysis
6) Voltage regulation
7) ripple
8) Standby power
9) Turn-on transient tests
10) Protection evaluation
The last one is the best test that has ever been added (just last review it was), since most 800W power supplies are more like 1000W power supplies in reality. For the Silverstone unit, overpower protection kicked in at 1038W. So he load tested it all the way up to that, which resulted in out-of-spec ripple on the Silverstone unit. This brings into question: whose fault is it? Is it the end user's fault for subjecting his computer to loads above the specifications of the power supply's rated amperage? Or is it the power supply's fault for not being able to shut off safely before that stuff goes out of spec?
Well, it's so difficult because OPP and OCP limits are always set so high to take into account those transient spikes. Like how a GTX 970 can spike well into the 350W territory. If we set the limits of these things to the label of the power supply, we would really have to start buying overkill PSUs to prevent them from shutting down. I guess the solution would be expensive, but would revolve around monitoring the average power over a short duration of time and using that instead to determine if the unit should shut down or not.
The current protection ICs operate in a cycle-based manner in that they only check the outputs every so often, like every .75 seconds for instance (spec sheets tell how often, lower is better). Since it checks instantaneous power, those transient spikes could indeed be a problem and shut down the unit if a spike is in progress when the chip is checking everything. That's why the limits are so high. In addition, the shunts used to read this stuff are not always 100% accurate, so the limits are set high in order to be safe.
Maybe whatever surpasses the AXi will have some form of new protection circuitry since the AXi already has purely digital regulation which accounts for its unbelievable voltage stability. These days, the most important testing for a power supply has to be transient response. Take a look at any graphics card power graph and it is constant spikes. It doesn't matter if progressive load regulation is top-notch because when gaming there will never be a constant load, it is always changing. Spikes up and down usually within 5-10A on the 12V rail. Transient response testing is certainly one of the single most important things to test, and hopefully at some point in time JG will upgrade their equipment.
Although I do have to take into question what exact load is being subjected when OPP is tested on the Tomshardware reviews. OPP is measured from the wall, so in a less efficient scenario it will take less power on the rails to trigger OPP< and in a more efficient scenario the computer can be under heavier load before OPP kicks in.
Facebook
Twitter
Instagram