PSU tier list 2.0

[Darkbreeze]

Same reason subwoofers and other speakers have continuous and peak specifications. Because they can handle brief peaks at a certain frequency, or current draw, but cannot sustain that due to not being designed to do so. That's part of the reason most recommendation lists are well above what is actually needed. Because people do not or cannot understand the difference.

 

[turkey3_scratch]

Yeah but I'm looking more for why peak power exists. If the components are rate to handle the peak current, why is it peak? Does it have to do with heat buildup when operating at peak?

 

[Darkbreeze]

It peaks when you demand more than it was designed to handle. "IT" doesn't peak. Draw or frequency, depending on device, peaks. If you ask more of it than it was designed to handle, then in some cases it WILL deliver it, for a short time, but that isn't an inherent feature of the device itself. It simply means that for a short time, it will tolerate X spec if you cause it to. For example, if you have a very good 300w PSU, but the system actually causes a 350w draw, then a very good unit MIGHT handle that peak draw for a short amount of time, but it certainly isn't going to SUSTAIN that draw except on a very few units that were probably underrated on their labeled capacity to begin with, for the purpose of playing it safe.

Or in cases where they intend it to be X amount of capacity but provide for components that IF more is demanded, can temporarily withstand that peak draw without burning out, blowing up or catching on fire. Peak power doesn't exist as an intentional factor, it simply IS.

 

[Karadjgne]

Older AMD gpus are a prime example of need for peak power. Take the 290x for example, most models operated in a 200w-250w range, however, there were occasions when some models like the VaporX would spike draw upwards of 325w. Recommended wattage for a 290x is 650w, which would account for a healthy OC heavy gaming session pulling 500w or so sustained, but able to absorb a 300w+ spike draw without overloading the psu.

 

[turkey3_scratch]

I've seen overclocked GTX 970 models also spike into the 300s. Is overcurrent protection usually quick enough to detect these brief spikes if it passes the trigger point?

 

[Someone Somewhere]

Heat is a major part. Not just the heatsinked stuff like the transistors and diodes, but also the windings on the transformers & inductors. More current also usually means that there's more ripple, so that will heat up the caps more.

I think the really short spikes would also just come out of the caps without the electronics having time to ramp up output.

I'm guessing that there's software filtering on overcurrent, but I don't know.

 

[Karadjgne]

Mostly depends on the psu and the quality/settings of the OCP circuitry. On low quality psus these settings are set so outrageously high, those triggers won't pop for anything less than a direct 120/240v main short, whereas higher grade psus triggers are much closer to accurate rail capabilities. Whether they actually trip if over drawn on such a short duration spike will be upto the designer of that particular circuit and just how tight the settings are. But I wouldn't doubt most high grade psu's will trip if the draw is sufficient and of sufficient duration, sorta like a slow burn fuse vrs an instant burn fuse.

 

[turkey3_scratch]

That'd be something I'd like reviewers to test, it would be interesting to see.

I see my 850 G2 handles surges and AC loss very well. I was just hit with 3 surges (lights slowly dimmed out, were black for a brief moment, then slowly came back to normal brightness; I think that'd be a surge) and there was definitely AC loss in the time period, and all was well. No UPS (though I am hooked up to just a Walmart surge suppressor). My Xbox shut off from it but not my computer. Glad to see that holdup time on the 850 is better than the 550 G2's 12ms.

Edit: Just happened again!

 

[turkey3_scratch]

This is really strange. As of late, my computer has been taking a long time (about 15 seconds) to post, but after the above AC losses happened, my post screen is like instant and my computer boots super quickly. Make sense to anybody?

 

[Someone Somewhere]

Note that the hold-up time only applies at full load - at lower loads, the energy stored in the primary cap lasts a lot longer. One of the few cases where an oversized PSU provides a substantial performance increase.

 

[Kernel-Fergus]

quick question:

after my cooler master ucp 900w decided to blow something (looks like the relay, but the question is what blew the relay) im looking at a replacement.

my question is about the evga supernova GQ. i found this:

https://www.scan.co.uk/products/750w-evga-gq-series-80-plus-gold-hybrid-modular-sli-crossfire-single-rail-624a-plus12v-1x135mm-fan-a

and im wondering if this is the Supernova GQ

 

[kwa-e]

It's definitely a GQ unit, though I'd prefer a G2 unit over this (if it isn't much more, of course) one due to the so-so ripple and voltage reg present within the GQ unit.

 

[Karadjgne]

Yeah turkey, it makes sense. Usually due to a Windows update fixing a prior update of an update that didn't need fixing in the first place.

 

[Kernel-Fergus]

It's definitely a GQ unit, though I'd prefer a G2 unit over this (if it isn't much more, of course) one due to the so-so ripple and voltage reg present within the GQ unit.

the g2 for the same wattage is coming in at nearly £25 more (a 1/3 increase) than the gq unit.

even the SuperNOVA B2 costs slightly more, but would it be preferable?

https://www.scan.co.uk/products/750w-evga-supernova-750-b2-80-plus-bronze-hybrid-modular-sli-crossfire-single-rail-62a-plus12v-1x140


since about £80 is my limit (i just bought 16GB ram to go with my 8GB,so budget is limited), what do you think is my best choice? to me,as long as it will be reliable, within my price range and get me up and running with some headroom for the future (major graphics cards updates) i have no brand loyalty whatsoever. to me the EVGA GQ looks like my best bet

ebuyer, scan and amazon are my preferred stores since ive had good experiences buying from them in the past.

all help is greatly apprecited by the way

 

[Darkbreeze]

B2 would be better than GQ IMO.

 

[turkey3_scratch]

Note that the hold-up time only applies at full load - at lower loads, the energy stored in the primary cap lasts a lot longer. One of the few cases where an oversized PSU provides a substantial performance increase.

True, totally didn't think of that. Yeah, under the minor load my machine was on, the bulk cap easily could have handled 1/3 of a second of AC loss.

I don't like the 750 GQ because on the Tomshardware review it does not even meet ATX specification for 3.3V ripple and nosie.

 

[turkey3_scratch]

Yeah turkey, it makes sense. Usually due to a Windows update fixing a prior update of an update that didn't need fixing in the first place.

What? I didn't say anything about a Windows update.

 

[Someone Somewhere]

Given Windows, it has a habit of installing Windows Updates at random times without you noticing.

 

[turkey3_scratch]

What's this stuff I'm hearing about "holding the power button" to "discharge capacitors"? Is this total bologna or is this, like, actually somewhat of a thing? Doesn't make sense to me, since all the power switch does is sends an electronic signal to the motherboard.

 

[Someone Somewhere]

PSUs have loads of stored energy, as do motherboards. And the standby current drawn by the ICs that are always on is really really low, so those caps could stay fairly full for minutes to hours.

Pushing the power button makes it try to turn on, which uses enough power to empty those caps.

 

[turkey3_scratch]

Which caps? The bulk ones, all of them?

 

[Someone Somewhere]

I think it's mainly the 5VSB rail that matters in this case. So every cap from the wall to the super IO chip.

 

[turkey3_scratch]

So what would be the purpose of doing this anyway?

 

[Someone Somewhere]

Clear the volatile memory/current state of the firmware chip, forcing it to restart from scratch. It might have gotten confused, or you changed hardware and it didn't notice, or you got it into an infinite loop.

 

[tiny voices]

I have done this many times to get to the BIOS after a failed overclock. Sometimes when I boot after a failed overclock it would only give me a black screen. If I flip the PSU switch and press the case power button, the PC will try to boot (fans twitch, lights flicker) and then die. After, I would flip the PSU switch back on and it will boot right to the BIOS no issue at all. That act of pressing the power button with the PSU switched off is draining the caps.