PSU voltage rail discussion

[HOOfan_1]

I am bringing this over from anandtech, because it is obvious not much discussion on this topic is going to go on over there


That's because practically all "anyone" failed to recognize that 99.99999% of PSUs on the market are "12V single-rail" PSUs.

Since they don't actually know about "12V single-rail" PSUs except aping the words, they failed to test for the inherent weakness of single-rail PSU design architecture assuming they even knew what that was.

"12V single-rail" PSU designs are easily recognizable by reading a literate PSU label like THIS ONE

The linked literate label indicated:

3.3V and 5V on one rail.
all 12V on one rail.
-12 and 5Vsb on one rail.

There is a standard test for single-rail architecture weakness and it is called a cross-regulation test.

For example: 3.3V and 5V on one rail meant 3.3V can cause cross-regulation failure in 5V and vice-versa.

In multi-rail, each one of the rails are independent from the others thus no cross-regulation problem.

He is of course claiming that any PSU with a single transformer for the 12v rail and a single transformer for the 3.33v and 5v rails will fail in crossload testing. However, I have seen multiple crossload tests here on jonnyguru.com that disprove that.

He even admits that 99% of PSUs are designed this way, but basically says that the Corsair HX620 and HX520 are faulty because of this and that a person should not buy them. So I suppose he expects them to go look for those 1% of PSUs out there that are true multi-rail 12V and maybe the .05% that have true rails for the 3.3v and 5v

[Spectre]

Was there some sort of prior context?

[HOOfan_1]

A poster asked if his HX520 would be sufficient for a GTX280 and a heavy overclock.

beray linked to a thread in which an author with an HX620 was having problems with constant reboots after installing an HD4870, although he tried another PSU and that did not fix the problem. Everyone else in the thread thought it might point to a motherboard problem.

Beray took this thread and began bashing Corsair PSUs as insufficient and told the poster he should get another PSU, without directing him to the specific problem with the Corsairs.

I stated that with his heavy overclock the HX520 might not be sufficient, but it probably still was fine...and that all the sites that I have read who have load tested the Corsair PSUs show they are excellent.

He states that the people testing them apparently don't know about the inherent flaws in single rail systems where the 12V rail is a single rail and the 3.33 and 5 V share the same transformer (regardless of whether they have their own rectifiers or not)

[Spectre]

A poster asked if his HX520 would be sufficient for a GTX280 and a heavy overclock.

beray linked to a thread in which an author with an HX620 was having problems with constant reboots after installing an HD4870, although he tried another PSU and that did not fix the problem. Everyone else in the thread thought it might point to a motherboard problem.

Beray took this thread and began bashing Corsair PSUs as insufficient and told the poster he should get another PSU, without directing him to the specific problem with the Corsairs.

I stated that with his heavy overclock the HX520 might not be sufficient, but it probably still was fine...and that all the sites that I have read who have load tested the Corsair PSUs show they are excellent.

He states that the people testing them apparently don't know about the inherent flaws in single rail systems where the 12V rail is a single rail and the 3.33 and 5 V share the same transformer (regardless of whether they have their own rectifiers or not)

Interesting. Well they do crossload fine. I don't specifically corssload units but at 100% with those kinds of units it typically has a heavy crossload given the high 12v capacity and they work fine. I think he may not quite understand the issue himself and the rest of his post is a headache.

[HOOfan_1]

well here is some insight into his background. I am barely even a layman when it comes to PSUs so I really haven't the knowledge to debate the subject.

@HOOfan 1
Find a real native English speaking PSU expert for your discussion...

1 - I'm not a PSU expert, my PSU skill is JOE AVERAGE, my specialty is realtime graphic hardware. I do PSU when needed in my hardware requirement though graphic hardware do have a bad habit of extreme power requirement, I don't do PSU for a living.

2 - I achieved my Engrish skill with a $30 3000-word dictionary, you can't really do much with 1-cent-per-word Engrish skill. But I'm doing OK with what little I'd invested, I did have more use for my 1-cent-per-word Engrish than some English natives with Master degree in English.

3 - There're know nothing idiot experts wanting everything "factually corrected" to below gradeschool ignorance and stupidity.

[rip97000]

@HOOfan 1
Find a real native English speaking PSU expert for your discussion...

1 - I'm not a PSU expert, my PSU skill is JOE AVERAGE, my specialty is realtime graphic hardware. I do PSU when needed in my hardware requirement though graphic hardware do have a bad habit of extreme power requirement, I don't do PSU for a living.

2 - I achieved my Engrish skill with a $30 3000-word dictionary, you can't really do much with 1-cent-per-word Engrish skill. But I'm doing OK with what little I'd invested, I did have more use for my 1-cent-per-word Engrish than some English natives with Master degree in English.

3 - There're know nothing idiot experts wanting everything "factually corrected" to below gradeschool ignorance and stupidity.

1 - i THOUGHT i was JOE AVERAGE when it came to PSUs also, until i started reading here

2 - Engrish? :wall::wall::wall::wall: maybe he should learn how to write the word first and then bash us later!

3 - i get my facts from a reliable source - like here!

man what an ass

[HOOfan_1]

I think he was spelling it as Engrish to indicate that he is not a fluent speaker.

[rip97000]

still he comes across as stupid and an ass

[HOOfan_1]

here is some more of what he has had to say

Yes, there are experts be stupid enough to test cross-regulation between 3.3V and 5V rail against 12V rail. Even though they are independent separate rails, not susceptible to cross-regulation failure from each other.

The real experts would do 3.3V against 5V and vice-versa as they are both on one single rail and would have cross-regulation problem with each other.

The real experts would do 12V against other 12V as they are all on one single rail and would have cross-regulation problem with each other.

single-rails are susceptible to cross-regulation, multi-rails don't as they are "independent rails/supplies".

Is there any reason to crossload the 3.33V rail against the 5V rail???

[Spectre]

here is some more of what he has had to say



Is there any reason to crossload the 3.33V rail against the 5V rail???

No especially since the 3.3v hardly ever gets loaded which measn it is almost always 5v high crossloaded and they work just fine. BTW indy regulation means the 3.3v and 5v are independatly regulated rails. Also units like the CWT PCU and others generate their 3.3v or 5v from a strait 12v DC-DC conversion and in the case of the PUC from seperate 12v sources.

[Travis]

There's one thing called "Mag-Amp" which makes the output "independent". Also we have DC-DC VRMs. A unit using Mag-Amp or VRM circuits on +5V/+3.3V output doesn't necessarily need an independent transformer for an independent output "rail", and it doesn't suffer from cross-loads.

[Spectre]

There's one thing called "Mag-Amp" which makes the output "independent". Also we have DC-DC VRMs. A unit using Mag-Amp or VRM circuits on +5V/+3.3V output doesn't necessarily need an independent transformer for an independent output "rail", and it doesn't suffer from cross-loads.

True like I said. The point of differentiation I was making though was the DC-DC is a more specific smaller subset of the total methods of independant regulation.

[HOOfan_1]

with the PUC the 3.33V is generated from one of the 12V transformers and the 5V is generated from the other 12V transformer, correct?

Anyway the guy is over there completely confusing people with misinformation now, which is sad, because someone might actually think he knows what he is talking about.

[MrWicked1968]

is beray related to gr33ngecko of the sapphire forums? It sounds as though they might be kindred spirits.

[HOOfan_1]

I thought so as well, but then he started ragging on Enermax PSUs also

[Quiksilver]

A poster asked if his HX520 would be sufficient for a GTX280 and a heavy overclock.

beray linked to a thread in which an author with an HX620 was having problems with constant reboots after installing an HD4870, although he tried another PSU and that did not fix the problem. Everyone else in the thread thought it might point to a motherboard problem.

Beray took this thread and began bashing Corsair PSUs as insufficient and told the poster he should get another PSU, without directing him to the specific problem with the Corsairs.

I stated that with his heavy overclock the HX520 might not be sufficient, but it probably still was fine...and that all the sites that I have read who have load tested the Corsair PSUs show they are excellent.

This "beray" guy is truly a misinformed asshole and comes off as a troll to be blunt.

In that thread (which had to be cleaned up by an AT mod) after me and him were arguing about his facts and the non-sense he was posting.

Like that 4870 thread (which he later admits he didn't even read at all) he was using to say Corsair was a bad PSU because of that thread; the thing is had he bothered to read it; he would have found out the problem for that user was a defective 4870 and had nothing to do with the PSU.

In that same thread, he continues babbling on about how Corsair PSU have "problems" and continued to back up what he had to say with random PSU threads. One of those threads was a "Should I buy an iStar PSU or Rosewill PSU" thread and the second was "My Enermax Exploded" thread. Neither of which had nothing to do with Corsair at all.

It was just silly, also because of this whole theory on PSU's he claims the VX series of Corsair PSU's are superior to the HX series. I tried to show him the VX550 VS HX520 review posted on HardOCP, but he ignored it and continued to post the same nonsense.

[Travis]

It was just silly, also because of this whole theory on PSU's he claims the VX series of Corsair PSU's are superior to the HX series. I tried to show him the VX550 VS HX520 review posted on HardOCP, but he ignored it and continued to post the same nonsense.

Well, I also found there's really some guy who's totally ignorant but unfortunately a troll. All he teach others in forums are nonsense. After I clarify everything he ever mentioned, all he's doing is blame me being "ignorant and misleading people".
I've no interest in educating him the right knowledge. This kind of guy is neither a good teacher nor a good student.

[Oklahoma Wolf]

which had to be cleaned up by an AT mod

I'm keeping a close eye on this thread myself.

[signmeuptoo]

Sorry if I did something wrong.

[Quiksilver]

Sorry if I did something wrong.

? Are you saying your 'beray'?

---

Also it now seems Paul is now mixing it up with the troll.

[OingoBoingo]

So the plot thickens