PSU Tier List [OLD]

[Energycore]

9 hours ago, RadiatingLight said:

Wait...

How is the EVGA B2 better than the B3.

Seems slightly counterintuitive.

If only EVGA had put a good fan into their B3 PSUs.

[LienusLateTips]

Why hasen't my FSP Aurum Pro oddity been added? It looks to be around Tier 2 according to reviews but I recommended it be changed before the B3 was added and now the B3 was added but my recommendation hasen't?

[comicsansms]

24 minutes ago, JDE said:

Why hasen't my FSP Aurum Pro oddity been added? It looks to be around Tier 2 according to reviews but I recommended it be changed before the B3 was added and now the B3 was added but my recommendation hasen't?

Aurum Pro is in tier 4, the ripple isn't very good, (850W unit) 53mV on the 12V rails.

www.hardwaresecrets.com/fsp-aurum-pro-850-w-power-supply-review/8/

 

The Corsair HX850 is an example of a tier 2 PSU, it has a 12V rail ripple of 26mV

www.hardwaresecrets.com/corsair-hx850-gold-power-supply-review/8/

 

[quan289]

1 hour ago, Comic_Sans_MS said:

Aurum Pro is in tier 4, the ripple isn't very good, (850W unit) 53mV on the 12V rails.

www.hardwaresecrets.com/fsp-aurum-pro-850-w-power-supply-review/8/

 

The Corsair HX850 is an example of a tier 2 PSU, it has a 12V rail ripple of 26mV

www.hardwaresecrets.com/corsair-hx850-gold-power-supply-review/8/

Quite honestly, power supplies nowadays are so well suppress, that it alone shouldn't have a drastic effect on what tier it should be place. Being under half of the specification, 53mV is still quite good. Otherwise, units like the EVGA Supernova 550/650GS should be drop there as well, as the 650w version had hit 57mV: https://www.techpowerup.com/reviews/EVGA/SuperNOVA_GS_650/9.html

 

[Energycore]

37 minutes ago, quan289 said:

Quite honestly, power supplies nowadays are so well suppress, that it alone shouldn't have a drastic effect on what tier it should be place. Being under half of the specification, 53mV is still quite good. Otherwise, units like the EVGA Supernova 550/650GS should be drop there as well, as the 650w version had hit 57mV: https://www.techpowerup.com/reviews/EVGA/SuperNOVA_GS_650/9.html

 

Yeah I think T4 is generally reserved for a unit that's decent but has one pretty bad thing whether it be ripple over 80mV in one rail, or terrible crossloading regulation like the NEX. It's really not that hard to make a safe PSU these days.

[LienusLateTips]

2 hours ago, Comic_Sans_MS said:

Aurum Pro is in tier 4, the ripple isn't very good, (850W unit) 53mV on the 12V rails.

www.hardwaresecrets.com/fsp-aurum-pro-850-w-power-supply-review/8/

 

The Corsair HX850 is an example of a tier 2 PSU, it has a 12V rail ripple of 26mV

www.hardwaresecrets.com/corsair-hx850-gold-power-supply-review/8/

 

Other reviews were showing it to be Tier 2. That's weird.

 

[LienusLateTips]

For some reason, according to Jonnyguru, the SeaSonic PRIME Titanium has a sleeve bearing fan. @STRMfrmXMN RIP

[quan289]

7 minutes ago, JDE said:

For some reason, according to Jonnyguru, the SeaSonic PRIME Titanium has a sleeve bearing fan. @STRMfrmXMN RIP

Seasonic Prime Titanium uses Fluid Dynamic Bearing (FDB) fan - not the traditional sleeve bearing that he doesn't like.

[Guest]

2 hours ago, JDE said:

Other reviews were showing it to be Tier 2. That's weird.

Do reviewers make up their own TIER list?

[Herman Mcpootis]

5 minutes ago, deXxterlab97 said:

Do reviewers make up their own TIER list?

IIRC johnnyguru has made one before, with around 14 or so tiers(roughly)

[LienusLateTips]

16 minutes ago, deXxterlab97 said:

Do reviewers make up their own TIER list?

AROUND Tier 2.

About 8-9 in performance category, according to jonnyguru

[Clyne]

I got a Antec Truepower 550W(Green Label), but lacking 1 more PCIE plug. Should be fine to use molex to 6 pin adapter right? Just powering a Q6600/GTX 680(2x6Pin), if molex is sketchy i'll just get some cables off of ebay. 

This PSU model

[MEC-777]

3 minutes ago, Bubblewhale said:

I got a Antec Truepower 550W(Green Label), but lacking 1 more PCIE plug. Should be fine to use molex to 6 pin adapter right? Just powering a Q6600/GTX 680(2x6Pin), if molex is sketchy i'll just get some cables off of ebay. 

If you have nothing else powered by molex on that PSU, you *should* be ok to use molex to single 6-pin. Otherwise, I'd be hesitant. Reason being is the label shows this unit has 4 +12v rails @ 20A each, which is kind of odd. I've never seen that before. 

[Clyne]

Just now, MEC-777 said:

If you have nothing else powered by molex on that PSU, you *should* be ok to use molex to single 6-pin. Otherwise, I'd be hesitant. Reason being is the label shows this unit has 4 +12v rails @ 20A each, which is kind of odd. I've never seen that before. 

EVGA G1 PSU has 4 +12v rails, but I may get a PCIE cable just in case molex might be sketchy. Anyway, thanks. 

[MEC-777]

2 minutes ago, Bubblewhale said:

EVGA G1 PSU has 4 +12v rails, but I may get a PCIE cable just in case molex might be sketchy. Anyway, thanks. 

Yeah, you're right, it does. But I don't understand why they do this. Because, if you look at the output chart, it shows the max draw on all 4 rails on both the G1 and yours (the Truepower) is about half that of all the 4 rails combined. Unfortunately, I don't know enough about these multi-rail units to explain how his works. 

 

Maybe someone more knowledgeable could chime in?

[quan289]

On 6/11/2017 at 0:07 AM, MEC-777 said:

Yeah, you're right, it does. But I don't understand why they do this. Because, if you look at the output chart, it shows the max draw on all 4 rails on both the G1 and yours (the Truepower) is about half that of all the 4 rails combined. Unfortunately, I don't know enough about these multi-rail units to explain how his works. 

 

Maybe someone more knowledgeable could chime in?

It's to add an extra layer of protection in case something goes wrong. In a single rail design, there isn't any sort of over current protection that prevent a certain group of wires from drawing too much power, while a multi-rail unit will limit the amount of current that can passed through them.

 

In an event of component failure that has a parasitic short, the PSU may continue to deliver more and more power to it, and depending on how large the unit it is, it will take longer for the Over Power Protection to detect and cut power. By the time the OPP finally cut power, wires / connectors may have been melted and welded together. There's even a possibility of it catching fire.

 

Here's an example of can happen: http://www.overclock.net/t/943852/computer-caught-fire

Here's a test of three PSU being put in a simulated short (the video is in German, but you can understand what's going on by looking at the video). Two single rails (AX750 and AX1200) and one multi-rail ( LC Power Legion X2 1000W) http://www.pcgameshardware.de/Panorama-Thema-233992/Videos/PCGH-in-Gefahr-Rauchexplosion-bei-Netzteilkurzschluss-1057938/

[MEC-777]

9 hours ago, quan289 said:

It's to add an extra layer of protection in case something goes wrong. In a single rail design, there isn't any sort of over current protection that prevent a certain group of wires from drawing too much power, while a multi-rail unit will limit the amount of current that can passed through them.

 

In an event of component failure that has a parasitic short, the PSU may continue to deliver more and more power to it, and depending on how large the unit it is, it will take longer for the Over Power Protection to detect and cut power. By the time the OPP finally cut power, wires / connectors may have been melted and welded together. There's even a possibility of it catching fire.

 

Here's an example of can happen: http://www.overclock.net/t/943852/computer-caught-fire

Here's a test of three PSU being put in a simulated short (the video is in German, but you can understand what's going on by looking at the video). Two single rails (AX750 and AX1200) and one multi-rail ( LC Power Legion X2 1000W)

A lot of modern high-end units (probably most) are all single rail. I would assume they now have ways of detecting overdraw on specific cables (PCIe or SATA power, for example) to prevent something like that from happening and shut the unit down. I think they are now able to have them switch from single to multi, depending on the load as well. I think I remember watching a video about one of the Corsair units (1000 or 1200w) and the guy said it can switch between single and multi. 

 

The problem with fixed multi-rail is each rail has a much lower limitation, so you have to be mindful of what components are connected to which rail. If you want to run a power-hungry GPU it has the potential to overload one of those rails. 

[quan289]

7 hours ago, MEC-777 said:

A lot of modern high-end units (probably most) are all single rail. I would assume they now have ways of detecting overdraw on specific cables (PCIe or SATA power, for example) to prevent something like that from happening and shut the unit down. I think they are now able to have them switch from single to multi, depending on the load as well. I think I remember watching a video about one of the Corsair units (1000 or 1200w) and the guy said it can switch between single and multi. 

 

The problem with fixed multi-rail is each rail has a much lower limitation, so you have to be mindful of what components are connected to which rail. If you want to run a power-hungry GPU it has the potential to overload one of those rails. 

Yes, a lot of modern PSU are indeed single rail units, due to it being easier and less expensive to make. This means more profit for their business, especially if there are also have been an idea that single rail are better than multi-rail units among the consumers. The cause of this was at first many companies had an issue with their power supplies prematurely shut off because they did not understand that if you put all of 6+2 Pin PEG off of one rail, it will easily exceed the 240VA or 20A @12V limit requirement the time. So rather than trying to fix their screw up and properly distribute the PEG cable across multiple of different rails, they decided make all of their PSUs single rail and marketed why they are superior by spreading a lot of misconception and FUD. You can still see this going on today with companies marketing single rail using fancy terms.

 

However, engineers have gotten a lot smarter and realized that the original limits that is based on server standards are not proper for modern day 12V-heavy, consumer rigs, so they (those who are still making MR units in the consumer market) have increase that limit, as well as have a more well planned cable distribution among the rails. A well-design multi-rail can allow the end-user simply plug in their components into it without having to worry about setting off the OCP.

 

Anyways, single rails units does not have any sort of OCP monitoring specific cables. If they did, then it's a falsely advertised unit like the original Seasonic Platinum which had four 12V rails despite being marketed as single rail. And while it is true that there are high-end units with the ability to switch between single rail and multi-rail (Corsair RMi / HXi / AXi / HX Platinum), the majority of them are still single rail.

 

Since you have brought up Corsair, recently there have been an issue with certain x99 motherboards due to a faulty soldering joint for the VRM [LINK]. The solder joint had failed and and went up in smoke. The power supply that had been used was the AX860i which the earlier samples was set in single rail by default. While the motherboard would have failed no matter what, the fact that it was in single rail mode did exacerbated the issue. In fact, before this had happen at Legit Reviews, Corsair had already set it to multi-rail by default to address the issues that is reported by their customers, which the OCP point is set at 30A (360w) per cable.

 

Another example would be the lack of an OCP on not just the 12V rail but the minor rails as well, where two German veterans at Jonnyguru had blew up their EVGA 550/650GS at the 5V rail.

[MEC-777]

21 hours ago, quan289 said:

-snipity-

So I guess my next question is; how do we know which modern multi-rail units have proper cable distribution across the rails? We'd have to dig into the schematics which could be difficult to get a hold of. 

 

And also, what you're saying is basically all modern single-rail units do not have proper OCP by nature of the single-rail design? Wouldn't this be known by now as a rather significant issue? 

[turkey3_scratch]

1 hour ago, MEC-777 said:

So I guess my next question is; how do we know which modern multi-rail units have proper cable distribution across the rails? We'd have to dig into the schematics which could be difficult to get a hold of. 

 

And also, what you're saying is basically all modern single-rail units do not have proper OCP by nature of the single-rail design? Wouldn't this be known by now as a rather significant issue? 

Just wait for the unit to be reviewed, then you'll easily know if cable distribution is good. But it is rare for non-German power supplies like Be Quiet to have multi-rail anyway, and if any normal brand does have multi-rail in 2017 it can sort of be assumed it's not going to be screwed up, I don't see it happen anymore. It never happened much anyway. Very few PSUs had this problem, most multi-rail units were done properly, but between single-rail marketing and people on forums spreading FUD, this idea got out that it's scary to buy a multi-rail unit for fear of it turning off.

 

Single-rail units do have OCP but usually only on the minor rails. They don't have OCP on the 12V rail because OPP is present instead. Having OCP on the 12V rail in a single-rail unit would be sort of pointless since OPP can essentially do the same thing. The only difference is OPP measures total power draw (from all rails) from the wall and shuts down when it reaches a certain point, and OCP actually looks at power output on some designated wires and shuts off when the combined power on those wires reaches a certain point. Seeing as how little power the 3.3V and 5V rails handle these, days, OPP will basically go off still if the 12V rail is overloaded without the necessity of 12V OCP.

[quan289]

1 hour ago, MEC-777 said:

So I guess my next question is; how do we know which modern multi-rail units have proper cable distribution across the rails? We'd have to dig into the schematics which could be difficult to get a hold of. 

 

And also, what you're saying is basically all modern single-rail units do not have proper OCP by nature of the single-rail design? Wouldn't this be known by now as a rather significant issue? 

Well, a proper manufacturer should provide you the necessary datasheet regarding which cable is used by which rail in their user manual. Some of them even list the actual OCP point as well, which is actually higher than what the specification charts states as an overhead (if it's rated for 30A on the specification chart, the actual OCP may be 35A-45A depending how what's hooked up to it). But I do understand that some company may not make this information easily accessible which can be difficult for the end-user to which PSU reviewers can provide such information, or you can contact the company.

 

And yes, modern single rail units does not have OCP implemented on the 12V rail. This is because since modern PSUs have the 12V rail take up the majority the rated capacity (DC-DC regulated PSU will typically have all of it on the 12V), that they would make used of the primary side Over Power Protection instead rather than implementing an OCP that would turn the PSU off at about the same side.

 

[quan289]

8 minutes ago, turkey3_scratch said:

It is rare for non-German power supplies like Be Quiet to have multi-rail anyway

Well, companies like Antec, Enermax / Lepa (mostly ones that's based on their designs), Corsair to some extent, etc. have quite a few multi-rail units. But yeah, there aren't many quality ones in comparison to single rails.

 

It's also not just the fear of having the power supply prematurely turned off that have people preferred single rail either. It's part of it, but there are people who believe that multi-rail provide less stable voltages and are less robust thus increasing the risk of it blowing up / catching fire. I even recall having a "discussion" with someone who was trying to convince me that you should not load each of the 12V rails more than 50% regardless of the overall combined rating...

[MEC-777]

2 hours ago, quan289 said:

Well, companies like Antec, Enermax / Lepa (mostly ones that's based on their designs), Corsair to some extent, etc. have quite a few multi-rail units. But yeah, there aren't many quality ones in comparison to single rails.

 

It's also not just the fear of having the power supply prematurely turned off that have people preferred single rail either. It's part of it, but there are people who believe that multi-rail provide less stable voltages and are less robust thus increasing the risk of it blowing up / catching fire. I even recall having a "discussion" with someone who was trying to convince me that you should not load each of the 12V rails more than 50% regardless of the overall combined rating...

The main reason I have suggested to people to replace their multi-rail units with a better, higher-quality unit is not so much for the voltage regulation stability (which could also be an issue), but for the concern that their system could overload one or more of the individual rails, and simply because the particular unit in question happens to be a very cheap, unknown brand and not even 80+ rated, if that. 

 

Now, looking at the Antec 550w unit brought up a few posts ago, let's refer to the chart below. It shows 4 +12v rails, each capable of delivering 20A (240w) each. Now, this is where I get confused with multi-rail units. Total combined output should be 4x that, but that's 960w and that can't be right, since it's a 550w rated unit, and it also shows on the label that the total combined max draw for all 4 +12v rails is actually 45A (540w). So how does this work exactly? This doesn't make sense to me. I think this is why some people say you shouldn't load each rail more than 50%. Theoretically, if you have 2 of those 4 rails near max, and you install a power-hungry graphics card on the 3rd rail (or even the 3rd and 4th combined), are you not going to overload this unit? 

[quan289]

3 hours ago, MEC-777 said:

The main reason I have suggested to people to replace their multi-rail units with a better, higher-quality unit is not so much for the voltage regulation stability (which could also be an issue), but for the concern that their system could overload one or more of the individual rails, and simply because the particular unit in question happens to be a very cheap, unknown brand and not even 80+ rated, if that. 

 

Now, looking at the Antec 550w unit brought up a few posts ago, let's refer to the chart below. It shows 4 +12v rails, each capable of delivering 20A (240w) each. Now, this is where I get confused with multi-rail units. Total combined output should be 4x that, but that's 960w and that can't be right, since it's a 550w rated unit, and it also shows on the label that the total combined max draw for all 4 +12v rails is actually 45A (540w). So how does this work exactly? This doesn't make sense to me. I think this is why some people say you shouldn't load each rail more than 50%. Theoretically, if you have 2 of those 4 rails near max, and you install a power-hungry graphics card on the 3rd rail (or even the 3rd and 4th combined), are you not going to overload this unit? 

Well, if you theoretically load two of the 12V rail near max, you would also loading near the 550w rated capacity (40A / 480w). So just like the OPP will shut the PSU off from overloading the PSU in a SR unit, it will also shut the PSU off in a MR unit.

 

With most consumer MR PSUs that's being sold on the market, whether it is a dual rail, quad rail, or a hexa rail PSU, they all come from the same, singular regulated 12V source which determine the rated capability of the 12V (in this cause, 45A / 540w rating) - not the individual 12V ratings (20A). Since modern computers are subject to transient / dynamic loads on the 12V rail with wide range of component configuration, there should be an certain degree of engineering overhead that allows the end-user to easily power their components in any cable configuration, but low enough for the trip point to be relevant in respect to safety to prevent over draw off of a set of cables.

 

To put it simply, the purpose of OPP and OCP in a MR unit are similar in a way that they both limit currents but OPP is more about protecting itself, while OCP is protecting the wires and components that it is monitoring (the OCP on the minor rails are needed to protect itself though). Because of this, if you add all of the individual rails together, it should always be quite a bit higher than the actual combined rating.  If you ever see a MR power supply that doesn't provide you with the combined rating but is additively comparable to to the rated wattage, that power supply is a PoS that is trying to make it look better than it actually is.

 

Anyways, a 550w unit typically have up to two PEG connectors, so it is good for a single GPU setup. On this particular unit, there's four 12V rails that's rated at 20A each, correct? It might be something like this.

12V1 = Peripherals (SATA, Molex, etc) and 24 pin connector

12V2 = 4+4 Pin EPS12V / ATX12V connector

12V3 = One 6+2 Pin PEG connector (rated by specs up to 150w for an 8 pin connector)

12V4 = One 6+2 Pin PEG connector

 

A 1080Ti requires a 6 pin and 8 pin PEG connector, so it will be getting its power from the 12V3 and 12V4, as well as the 12V1 from the 24 pin via PCIe slot (5.5A on the 12V and 3A on the 3.3V by specs).

[awesomegamer919]

Why is the Thermaltake SMART series tier 6? Tweaktown gave it generally favourable reviews...

 

http://www.tweaktown.com/reviews/5613/thermaltake-smart-750-watt-80-plus-bronze-power-supply-review/index.html

 

The SMART M series isn't even on the list despite having a good review by HardOCP...

 

https://www.hardocp.com/article/2014/06/16/thermaltake_smart_sp850m_850w_power_supply_review/