PSU

Jhunxkie · December 18, 2019

Hi! It's me again hihi and with noob question

What is the good in this two?

Evga supernova g1+ 650 and Corsair Rm650i?

Thanks

More power!

Mantabase · December 18, 2019

I used to test PSU's - and review them. But it has been a while.

These two look evenly matched. In my experience, I would go for the Corsair - based on the quality I have seen inside them. However - that could just be my bias from back in the day since it has been a bit. I would look at proper reviews that actually open the PSU, look around, and also load test it - in an actual build or with a tester.

You can pretty much stop here. That is my answer.

As an aside - though I never published it, EVGA (and other 80+ supplies from certain brands) used to give off enough heat that I would question the 80+ rating. If LTT wants to set me up with a load tester (which they don't) I could tell you for more certain. What you care about (and what a proper review covers) is this:

Can it take full load for 15 minutes or more - requires test.

Can it take 10% overload for 14 seconds or more - requires test

Under heavy load does the heat output indicate it is 80%+ (or whatever rating) - requires test

What is the load heat? Because that will impact case heat.

Are the cables long enough? If documented you can measure - easy enough.

Is it modular? Modular cables may shorten the PSU's lifetime and can be harder on components and produce more heat. But in reality, for most applications the advantages outweigh the setbacks.

Is it power dirty - requires test

What is the main cooling fan rating (MTBE, Flow, db, and power)?

Does it have incoming power housekeeping (does it clean up the wall power).

Are the components quality - requires opening.

Is the assembly quality - requires opening.

Unlike RAM, the MB, the CPU, the Vid card - if you invest in a top notch PSU, it will last multiple build generations

LukeSavenije · December 18, 2019

4 hours ago, Mantabase said:

These two look evenly matched

as a former reviewer i want you to look a bit closer

acrf don't react well to transients, generally speaking

https://www.tomshardware.com/reviews/evga-supernova-650-g1-plus-psu,5661-8.html

compared to rmi

https://www.tomshardware.com/reviews/corsair-rm750i-power-supply,4223-7.html

with that the rmi was monitoring functions, can be configured on the rails and is less likely to whine

does that make g1+ a horrible psu? no

does this make the rmi better? yes

so at the same price, I'm going to say rmi

Mezoxin · December 18, 2019

with the rmi you get all possible protections available to a PSU at this time , also you get monitoring options and silent operation and fan curve control , i agree its the better option

Mantabase · December 19, 2019

7 hours ago, LukeSavenije said:

as a former reviewer i want you to look a bit closer

Well I did say look at a proper review (being sure to point out I didn't offer one) - but point taken. Should have waited I guess. Glad you pointed out the EVGA is not a bad choice - just not the best at equal price. And I'll agree with that. I don't frequent Tom's as much as I used to - and in fact I didn't realize folks still cared about PSUs. It was hard to get folks to ever care about one of the most important parts in their system for me.

EDIT - you don't need to read the rest unless you want to (lol....not to imply you have to read anything).

You seem to be up on the subject, so I apologize for the following rambling, but how much does the 1ms deviation matter on the 3.3V rail? Maybe you can tell me. Back in the day (early/mid 2000's) folks bashed modular units. They have higher resistance and they create more heat and lower the efficiency. There was also a huge amount of attention given to how clean the power out of the PSU was. We tested PSU's on isolated circuits to make sure the output was -super- clean - using super clean input power. Yet, The added resistance has almost no impact on efficiency or added heat and power quality coming out of the wall in most homes is so crappy it can't be totally fixed with simply housekeeping at the PSU - so the test meas little IMO.

Point is - what really matters - because some of these tests I don't think do - is reality. I am not sure how PSU's are tested for review is really realistic. I would prefer a static base load put on a PSU with a real reference (computer) system then used to push it. And why is undervolting treated the same as overvolting? They are spec'ed that way - but should not be graded that way IMO - they have different consequences. And I wonder (since I also used to race cars), high spec does not always mean best long term performance. Some of the best PSU's I tested actually didn't last long in use. High precision can sometimes mean more influence from the weakest link. SO, a real world test would test a PSU using a real system (in part). But no tests I see check for longevity in spec and that is what is important. H[OCP] used to test to failure - and it was useless - it's not real world.

Anyway - The rig I am using right now is @ 24.9 C and using an Antec 1100. Just pointing out a good PSU choice can last for many builds. It's a sloppy PSU compared to other of the time - but it is running like 15 years later. Thanks for the reply - Peace.

LukeSavenije · December 19, 2019

2 hours ago, Mantabase said:

You seem to be up on the subject, so I apologize for the following rambling, but how much does the 1ms deviation matter on the 3.3V rail?

it's actually something I'm looking for myself. i know what transient responses could do in bigger numbers, but what they do in smaller ones, not sure. the only thing i know for sure is that a worse transient response will make it harder for the motherboard, but this difference might as well be caught by the board.

2 hours ago, Mantabase said:

I would prefer a static base load put on a PSU with a real reference (computer) system then used to push it.

some actually do still test this way, but that's generally when they can't afford a good loadtester, and use a pc to do this. there are still tests done in house at the companies themselves with even the heaviest GPUs like the 295x2 (able to push around 500w on it's own).

currently it's just done with actual tools by the better reviewers, to see what numbers come out of it. I'm personally someone who focusses most on protections, so for me the way they test those is very realistic. some other parts are debatable, but i think the more consistent way a load tester pulls power matters for the testing of the PSU

Mantabase · December 20, 2019

19 hours ago, LukeSavenije said:

some other parts are debatable, but i think the more consistent way a load tester pulls power matters for the testing of the PSU

Hmm....must not have hit submit.

Agree mostly. Consistency is key always, and using a PSU tester is important for QA/QC and can be very useful for assessing a PSU for a user as well. I just think some types of tests can be misleading - or failing them can. If I were to do it again, I would change many things - including how I treated the tester and the PSU. I've learned a lot about quantitative analysis since then. I would be doing cali checks, drift testing and environmental monitoring during every test now. But I am not sure it would translate into anything more than splitting hairs. I think the 1ms test "fail" is about as telling as the old modular cable "fail" - theoretical but not practical. Though, if the price is the same....

Think about this - have you lost a MB to a PSU? A CPU to one - or GPU? With some debatable exceptions I have not. Granted, I use quality PSU's, but I have no doubt a few of them would fail the 1ms or even 200ms test. I've been under PC hoods since 1982 when I first installed an AST-6Pak - so I have been through my share of PSUs and I don't like replacing hardware until it is dead or useless (I have a Verge S3 card on my desk). Maybe I have been lucky. But when I looked at loads during modelling and gaming - they are not crazy like some of the test folks run. You do not see dynamic loads change every tenth of a second. I have only looked at the 12 and 5 - but you just don't see it (or I don't). Maybe my meter setup can't catch it - but it seems to me that numerous large and very fast changes in rail load are rare.

Anyways - it's been a good chat and I ramble I know. Thanks for the interesting convo. Happy Winter Season.