well my machine killed another power supply, how often do you overclockers go through PSUs?

[Beanie46]

Originally posted by: IBDoomed
I'm having trouble reading westom's posts but it seems like the idea he's trying to impart is the same as the rest of us. A non-faulty power supply built to specs and of high quality components should be a safe bet. I'm not an electrical engineer and I've only been doing corporate IT support for a decade or so but this falls in line with what I've seen. Cheap crap oem power supplies can take out entire systems but good quality ones never have.

Unfortunately, that's not what he's saying.

Consider these statements of his:

Originally posted by: westom

The informed know an overloaded supply is not damaged ... and know why.


A power supply must never harm a computer. And a supply must not be harmed by the load.


Power supplies could never be killed by the load even long before PCs existed. Even Intel specs for power supplies required functions that make such damage impossible.


No minimally acceptable supply can damage a computer.

Then go to my link in one of my above posts and read the current ATX specifications and about the warning Intel puts in there about power supplies that only meet minimum specifications for over current protection. Sort of refutes the "power supply cannot damage a computer" comment.


Then, as to "a power supply could never be killed by the load" comment......follow the link below and watch Corsair put a 75% load on a power supply and watch it explode.....over and over and over.

Corsair power supply test.


Then, follow the links below to see some others die under very light load, such as a 650W ps dying at 183W of load:

JonnyGuru's Bargain Basement ps test.

Death of a Gutless Wonder.

Best Buy's Dynex and Rocketfish ps testing. (Hint: The Dynex failed at 75% load and, as was said, "Test #3 saw the first Dynex 400w blowout on the primary side destroying our Variac." Guess that refutes a power supply cannot damage anything connected to it.)

Sort of refutes that argument, too. All those power supplies met ATX spec, btw.

I do agree that cutting corners to save pennies on a power supply is ridiculous, but it is a fact that just meeting ATX 2.2 or 2.3 specs does NOT ensure a quality power supply, just as a power supply being able to get 80plus Bronze or Silver certification for efficiency does NOT ensure quality.

 

[IBDoomed]

Originally posted by: Beanie46

Originally posted by: IBDoomed
I'm having trouble reading westom's posts but it seems like the idea he's trying to impart is the same as the rest of us. A non-faulty power supply built to specs and of high quality components should be a safe bet. I'm not an electrical engineer and I've only been doing corporate IT support for a decade or so but this falls in line with what I've seen. Cheap crap oem power supplies can take out entire systems but good quality ones never have.

Unfortunately, that's not what he's saying.

Consider these statements of his:

Originally posted by: westom

The informed know an overloaded supply is not damaged ... and know why.


A power supply must never harm a computer. And a supply must not be harmed by the load.


Power supplies could never be killed by the load even long before PCs existed. Even Intel specs for power supplies required functions that make such damage impossible.


No minimally acceptable supply can damage a computer.

Then go to my link in one of my above posts and read the current ATX specifications and about the warning Intel puts in there about power supplies that only meet minimum specifications for over current protection. Sort of refutes the "power supply cannot damage a computer" comment.


Then, as to "a power supply could never be killed by the load" comment......follow the link below and watch Corsair put a 75% load on a power supply and watch it explode.....over and over and over.

Corsair power supply test.


Then, follow the links below to see some others die under very light load, such as a 650W ps dying at 183W of load:

JonnyGuru's Bargain Basement ps test.

Death of a Gutless Wonder.

Best Buy's Dynex and Rocketfish ps testing. (Hint: The Dynex failed at 75% load and, as was said, "Test #3 saw the first Dynex 400w blowout on the primary side destroying our Variac." Guess that refutes a power supply cannot damage anything connected to it.)

Sort of refutes that argument, too. All those power supplies met ATX spec, btw.

I do agree that cutting corners to save pennies on a power supply is ridiculous, but it is a fact that just meeting ATX 2.2 or 2.3 specs does NOT ensure a quality power supply, just as a power supply being able to get 80plus Bronze or Silver certification for efficiency does NOT ensure quality.

Ah good. Extracting important quotes was very helpful in understanding westom. I kept reading his posts and my eyes would gloss over trying to get past the language issues.

 

[westom]

Originally posted by: Beanie46
Wow. westom came back and again, as predicted, made circular arguments that ended up meaning absolutely nothing, esp. considering had Intel's documentation in the ATX design guide spec. about how over current can damage a computer.

Beanie46 does not have basic electrical knowledge. Even foolishly thought OCP is for motherboard protection. Did not know that OVP is the motherboard protection.

Beanie46 is so technically ignorant and such a demeaning person as to not read the specs - Beanie46 must post insults to cover up so many obvious examples of complete technical ignorance. Attack the messenger when exposed as dumb. Beanie46 does not even know the most basic electrical concepts.

Facts provided and industry standards that have exists long before PCs existed both state that a power supply cannot harm the load (ie motherboard). And that the load cannot harm a supply. Personal insults are posted by one whose ego is too big to be honest.

Insults are posted because Beanie46 cannot dispute the science - does not even understand what something as simple as OCP does. OCP is not for protecting the motherboard. Beanie46 should have known that. Intel even states that a defective motherboard cannot even be protected by OCP. But other power supply functions mean a power supply cannot damage a motherboard. How can an American could be so technically ignorant? But then America is now importing massive numbers of engineers because so many like Beanie46 do not even know how electricity works. But then Beanie46 demonstrates why some people will even blame the car for running out of gasoline. And so we define Beanie46 dumber than a duck - with an attitude that is even uglier.

Amazing how an engineer who did this stuff for generations is told he is wrong. Long before PCs even existed, it was routine for engineers (ie this author) to design power supplies that would not harm the load. And that could not be damaged by the load. Only the most technically ignorant would dispute it. Which explains why Beanie46 posts no technical facts and plenty of insults.

Stop slinging insults and personally attacking members. Back up your claims with FACT or i'll show you our over current protection and zap you out of here myself
Anandtech Moderator
Gillbot

 

[westom]

Originally posted by: Beanie46
Then go to my link in one of my above posts and read the current ATX specifications and about the warning Intel puts in there about power supplies that only meet minimum specifications for over current protection. Sort of refutes the "power supply cannot damage a computer" comment.

I you had any decency or knowledge, you would have grasp what OCP does. It does not protect a motherboard. It is not suppose to protect a motherboard. A motherboard that is not damaged never draws excessive current. So why would OCP even exist to protect that motherboard? With any electrical knowledge, Beanie46 would know that.

Spec says even OCP will not limit current enough to keep a failed motherboard from drawing excessive current. Spec also says OCP is why the load - a defective motherboard or even a short circuit - cannot damage a power supply. Just another industry standard that has existed long before PCs did.

One who understands basics also read that OVP is why a power supply cannot damage the motherboard. Beanie46 intentionally ignored that part. Motherboard (to draw excessive current) first must be damaged by excessive voltage. The spec then says a power supply cannot output excessive voltage ... and says why.

Obviously, any minimally acceptable supply will meet those specs. But Chinese have discovered Beanie46 and a market for supplies that are missing essential functions. Sold at less price AND for higher profits. Supplies sold to computer assemblers - people without electrical knowledge - are routinely dumped into the US market. Even recommended by people such as Beanie46. Those supplies violate ATX standards AND the standards we had to design to even 30 years ago.

Who actually did this stuff? The person who is posting technical facts and numbers - or the person who is replying with propaganda, misquotes, and no technical numbers?

Beanie46 cites power supplies that are missing essential functions such as OCP - the Brand X power supply. Why are these power supplies so often found in America? Notice Beanie46 does not even understand: OCP is missing because that counterfeit ATX supply is marketed to people just like Beanie46. Required functions are missing in those exploding power supply videos. Those exploding supply videos would not exist if Beanie46, et al had minimal electrical knowledge.

To post lies, Beanie46 even intentionally misquotes what I never said:
> Guess that refutes a power supply cannot damage anything connected to it.)

Minimal knowledge says a variac is not the load on a power supply. AND that other functions required in a power supply would have made that variac damage impossible. But many supplies are marketed and purchased by people with Beanie46's minimal knowledge. Even Beanie46 videos and tech reviews define power supplies that fail because they are sold to Beanie46 as ATX supplies - and are not.

How does a power supply with 70% efficiency even get into this country? At 70%, it violates a long list of well proven and required standards - including Energy Star and even Chinese standards. That supply would even be illegal even in China. But in America with so many Beanie46s, that supply becomes profitable.

No supply can damage a computer. And no computer (the load) can damage a power supply. Why are so many supplies now sold in America that violate that standard? Beanie46's insufficient knowledge demonstrates why. So Beanie46 must attack me - to mask massive technical naivety. Yes - I am replying to Beanie46's disrespect for the many who obviously bothered to first learn before posting.

No supply should harm a computer. And no computer should harm the supply. A standard that existed 20 years before Beanie46 was born.

IBdoomed accurately saw what I was saying:

A non-faulty power supply built to specs and of high quality components should be a safe bet.

To dispute that, Beanie46 has even twisted what was posted. IBdoomed, you had me quoted correctly the first time. However, do you believe one who does this stuff, or a computer assembler who does not even know how electricity works - even foolishly thought OCP might protect a motherboard. Beanie46 even ignored a power supply functions that means the supply cannot damage hardware - OVP. Why? OVP has been standard on power supplies long before PCs even existed. OVP is a required ATX functions often missing in supplies marketed to Beanie46 and others without electrical knowledge.

 

[westom]

Gilbot - I was clearly personally attaacked by Beanie46. I have posted fact after fact. Beanie46 has even intentionaly misquoted me to attack me. Show me one fact that Beanie46 is posting. I am being personally attacked because Beanie46 cannot confront the technology. And I have only replied in the same tone and demenaor found first in numerous Beanie46's posts.

Again, many power supplies are even missing essential functions because so many do not have basic electical knowledge. That is not an attack on anyone. That is a technical reality too common among computer assemblers. Example: OCP is not to protect any hardware. A technical fact. Only one without any basic power supply knowledge would not know that. Is that an attack on a poster? No. That is a technical fact about why so many power supplies sold in America are counterfeit. Why so many supplies do damage motherboards and can be damaged by the load. But to say that also says that Beanie46 clearly has no electrical knowledge. That OCP post clearly demonstrates a lack even of basic electrical principles of voltage and current.

No power supply could damage a computer even before PCs existed. And no motherboard can damage an ATX power supply. Why do the videos shows damage? Because many have no electrical knowledge and buy supplies 'rumored' to be ATX that are missing ATX functions. For having said that, Beanie46 even calls me a 14 year old - and you think that was acceptable?

 

[HOOfan 1]

"Cannot, cannot, cannot, cannot"

You keep saying that.

The phrase you are looking for is SHOULD NOT.

A faulty power supply ABSOLTUTELY CAN damage your components. Intel wrote the ATX specifications so that a power supply SHOULD NOT damage components.

ATX is a specification (e.g. A GUIDLINE), not a law, nor a code. There are plenty of parts of the ATX specification that even high quality power supplies do not adhere to. There is nothing that mandates anyone follow the ATX specifications, as there are plenty of power supplies out there that, for various reasons do not, or cannot. The Underwriters Labratory tests them to make sure they are not a danger to the users nor a danger of causing major damage to dwelling property, they do not test them to make sure they will not damage computer equipment.

I highly doubt that the engineers designing even the faulty power supplies "don't understand electricity". A crappy power supply is more often a case of cost and corner cutting, than it is of ignorance from the designer.

 

[daw123]

Westom, if I were to take sides I would have to go with Beanie46, since she (I assume she is a she from her icon) has at least included links in her comments / opinions to reputable sites and documents, which support her posts.

The comments / opinions that you have posted, you have stated as being fact. You may or may not be correct in your statements, but without links or documentary evidence to back up what you are saying, I (and possibly others) see these statements as being tenuously factual.

Just by saying something is a fact, does not make it so.

On the other hand, if you can provide supporting documentation, then it would add credence to your statements and others may then deem what you have said to be fact.

With regard to any perceived personal attacks; what I have said above is probably what Beanie46 has been attempting to convey in her earlier posts, although I've stated it in a more tactful way.

 

[Beanie46]

daw123, yes, I am a lady.....a rather crusty, old one at that, turning 55 this year.

And I'm not exactly new to electrical engineering, either. Got well versed in it from my father who earned his BS in Electrical Engineering from Anapolis, Class of '52, and his Masters from GT in '64. (He was serving out his required term in the Navy and beginning a family between the bachelor's and master's degrees.....)

So, not exactly unversed in the ways of the electron. 🙂

And I'll retract my 14 year old comment........hate slamming all the other pre-teens like that.

 

[F1shF4t]

Originally posted by: westom
Gilbot - I was clearly personally attaacked by Beanie46. I have posted fact after fact. Beanie46 has even intentionaly misquoted me to attack me. Show me one fact that Beanie46 is posting. I am being personally attacked because Beanie46 cannot confront the technology. And I have only replied in the same tone and demenaor found first in numerous Beanie46's posts.

Again, many power supplies are even missing essential functions because so many do not have basic electical knowledge. That is not an attack on anyone. That is a technical reality too common among computer assemblers. Example: OCP is not to protect any hardware. A technical fact. Only one without any basic power supply knowledge would not know that. Is that an attack on a poster? No. That is a technical fact about why so many power supplies sold in America are counterfeit. Why so many supplies do damage motherboards and can be damaged by the load. But to say that also says that Beanie46 clearly has no electrical knowledge. That OCP post clearly demonstrates a lack even of basic electrical principles of voltage and current.

No power supply could damage a computer even before PCs existed. And no motherboard can damage an ATX power supply. Why do the videos shows damage? Because many have no electrical knowledge and buy supplies 'rumored' to be ATX that are missing ATX functions. For having said that, Beanie46 even calls me a 14 year old - and you think that was acceptable?

No one was saying OCP is to protect the hardware. Everyone was saying (Beanie46 included) that because so many units don't have this protection set up properly, the consequent damage to the psu CAN and WILL damage the hardware.

The fact that these crap psus are on the market is a fact. Just because you think they should not be doesn't change that fact. We're all saying which ones to avoid, while your saying why they should not be like that, apples and oranges dude but same fruit salad 😕

 

[westom]

Originally posted by: daw123
Westom, if I were to take sides I would have to go with Beanie46, since she (I assume she is a she from her icon) has at least included links in her comments / opinions to reputable sites and documents, which support her posts.

Using her links, adding additional facts (simple electrical concepts), and those links show what I have said.

Now, remember. I am coming with generations of design experience. Beanie46 is posting popular ?computer assembler? myths. Each of those videos and reviews says the power supply violates numerous standards and violates numerous ATX requirements. Some even violate UL safety requirements. If you need more, then, well, I am still waiting for someone to ask logical questions; not post hysterical accusations.

So yes, you tone is one asking to learn whereas Beanie46 is accusations based - very obviously - on insufficient electrical knowledge.

For example, the world's largest power supply can be connected a motherboard and it still must not consume excessive or destructive currents. And, if the defective motherboard draws destructive currents, even the OCP function will not protect it. This is obvious to anyone with electrical knowledge.

But since OCP does not protect the motherboard (is not even supposed to), that proves that power supplies can destroy motherboards? That is what Beanie46 also posted. That is Beanie46 completely ignoring the other function that means no power supply can ever damage a motherboard.

Beanie46 did what Rush Limbaugh does. Take sentences completely out of context and hype it into a half truth that *proves the world is flat*.

Why do so many power supplies exist that self destruct? No power supply - even before PCs exists - even when I was designing and building them - could do that. However there you have it. Chinese can dump power supplies into America, lie (call then ATX supplies), and get Beanie46 to conclude such destruction is normal.

More numbers if I recall the video. Listed is a 450 watt supply. That means its rated output is maybe 315 watts (long details provided only if sufficiently interested in learning). Well, that supply self destructed on 330 watts? That can only happen when the supply does not have OCP. Even Beanie46 admits that OCP must be part of an ATX power supply. So why did this supply self destruct? Its target market is massive numbers of computer assemblers who knew learned basic voltage and current concepts. That supply spit sparks and smoke because it was missing essential functions that were standard even long before the PC existed.

I can write pages exposing Beanie46?s myths and half truths. Which ones do you want to discuss? Appreciate that no soundbyte can explain reality. Reasons will be provided with lots of facts and numbers (even the simplest confused IBdoomed). Beanie46 can only reply with subjective and superficial reasoning encased in accusations. IBdoomed was far closer to the truth by saying

I've only been doing corporate IT support for a decade or so but this falls in line with what I've seen. Cheap crap oem power supplies can take out entire systems but good quality ones never have.

BTW, price is not determining factor. Some most expensive supplies are also missing essential ATX functions.

There is no way around reality. A power supply must never damage the computer. And a motherboard must never damage a power supply. But when so many 'experts' do not even know what OCP does, then those experts have made a very lucrative market for Chinese to dump inferior supplies, sell at less money, for even higher profits. Why are so many if not most engineers in the Silicon Valley now from India or China? Because too many Americans would attack the engineer with generations of experience rather than learn how things work. Your opportunity. Ask questions and learn - with numbers. Or encourage the naive to attack the messenger with hyperbolic accusations and outright technical lies.

Those power supplies demonstrated in Beanie46's video do not even meet UL and FCC standards. Obviously. Just another fact 'forgotten' due to significant technical naivety and 'forgotten' to justify personal attacks.

This is the post that says ask for additional facts. Facts not posted because posts that define reality are already too long for most to read. Even Beanie46's citations say 180 degrees different (agree with what I have posted) once we include some basic electrical concepts. Your choice. See those facts or just believe hyperbolic and technically naive accusations.

 

[westom]

Originally posted by: Beanie46
And I'm not exactly new to electrical engineering, either. Got well versed in it from my father who earned his BS in Electrical Engineering from Anapolis, ...

My next door neighbor was a nuclear physicist. So I am also knowledgeable on quantum physics? Nonsense. While you were growing up, I was already building electronic circuits with OP amps (that sometimes got so hot as to burn skin). Little hint. I bought my first transistor in a shop that would later be torn down to build the World Trade Center.

Your posts demonstrate insufficient electrical knowledge AND are typically of self proclaimed 'experts' who never designed anything. Notice, I never said you never designed anything - so don't even tried to misrepresent what I posted. However, your every citation demonstrates technical arrogance associated with significant technical naivety.

But again - because this is the fundamental point: A power supply cannot harm a computer - a very old industry standard. And a computer cannot cause power supply failure - as even OCP says. Beanie46 has even posted videos that demonstrate how lucrative the market is selling woefully defective supplies to the so many technically naive computer assemblers. Even OCP says those many 450 watt supplies must never fail catastrophically. In so many cases, a supply was obviously missing OCP functions. A common problem now that so many know only because a relative once knew. The one who actually did this stuff by literally designing and then soldering things wire by wire - that person even got accused as a 14 year old by one who clearly does not have even basic electrical knowledge.

Anyone with power supply knowledge even 20 years ago knows that supplies could only fail that way because basic and required functions were missing. A common problem in America now that so many just know because they one knew an engineer. I am an engineer, designed this stuff, and have known these concepts for how many decades? Arrogance combined with insufficient electrical knowledge is the only reason for those personal attacks.

 

[daw123]

I openly admit that I'm noob when it comes to power supplies.

The very simplistic (and possibly ill informed) criteria I used when choosing my PSU was:
Cost (I had and spent the budget of £150)
Manufacturer, although this is a misnomer since I only recently learned that a lot of PSUs are re-badged. e.g. a PSU sold by company X is a re-badged PSU manufactured by company Y and company Z also sells company Y's re-badged PSUs.
Reviews on the 'Net
Power output
Current output on each rail
Had to be modular (for cable tidyness inside the case)
Had to be capable of powering 2x HD4870 X2s in Quadfire amongst the other usual components in a PC (I checked the ATI website for cetified PSUs - not that this means much)
ATX standard

I figured that if I spent a reasonable amount of money then in general the PSU would likely have a reasonable lifespan and would not kill components if it died. The latter factor could have been a misconception on my part but I figured spending £140 on a PSU in a rig, which cost well in excess of £2K was money well spent (and it will be £3K+ once I've finished).

That was pretty much it and I bought the Antec TruePower Quattro 1000W. I've no idea if it is actually a 'good' PSU when it comes to how it was built and what inherent safety / protection measures it has.

Westom, what would you look for if you were to buy a PSU. What features do see as essential and what design would you go for (in terms of how it transforms the AC to the 3.3V, 5V and 12V rails, how it regulates current, 'smooths' the voltages, etc.)

Specifics for this and links are welcome (provided they don't become too complicated and full of technobabble - think layman's terms 😀) because I would like to learn.

 

[westom]

Originally posted by: daw123
I openly admit that I'm noob when it comes to power supplies.

The very simplistic (and possibly ill informed) criteria I used when choosing my PSU was:
Cost (I had and spent the budget of £150)

I even designed these things; even repaired them not to fix them; to learn why failures happen.

For example, incoming filter electrolytic sometimes failed repeatedly. Well, two electrolytics with the same capacitance, voltage, and temperature had a massively different failure rate. The electrolytic with more 'volume' also almost never failed. We never got a good engineering reason for why. But we who actually did this stuff would sometimes notice the better supply also had electrically equivalent capacitors ... that were larger.

If a supply does not sell for at least $60 retail, then it is probably missing essential functions. That does not say for one minute that the $100 power supply is better. But it is a lower limit to quickly define defective supplies selling for higher profit by missing required functions.

Does it come with a full sheet of numeric specs? Missing written numeric specs are necessary to sell a supply missing those essential functions. Don't make the mistake of using binary logic. A supply without written specs could be sufficient. However, to sell supplies that are missing essential functions, they must *forget* to include a long sheet of numbers that (at best) only 1% may understand. To keep the 1% silent (and other legal reasons), they must *forget* to include written specs. Note: ternary logic.

Even watts says nothing. As noted previously, the 450 watt supply is also rated by more responsible computer manufacturers as a 315 watt supply. And nobody lied. Welcome to a game of specmanship that is popular when the market has so many technical naive customers.

Total wattage also says little. Far more important is the wattage (amperes) for each voltage. But too many only buy supplies only on dollars and watts. A symptom that also explains why so much of the Silicon Valley now needs immigrants.

And finally, anyone who knows power supplies also knows that a defective power supply can still power and run a computer. How to find the defective supply before it caused computer crashes months or years later? Multitask that machine so that it accesses every peripheral simultaneously. Plays complex video graphics while downloading from the internet, while playing loud audio, while searching the hard drive, etc. Then measure each VDC voltage. Those numbers must be higher than limits in the ATX specs (for other electrical reasons). If any number is too low (and yet still above the limits in ATX spec), then the supply is insufficient on that voltage. Defective and yet still boots the computer.

Some simplest guidelines so that any layman can select a power supply.

Did that power supply review use oscilloscopes or reactive loads to truly verify supply numeric specs? Most reviews are based in silly things such as appearance or supply weight. Above are how layman can start to select a supply. Reviewers should provide articles chock full of numbers. Most reviews don?t ? make me sick with so much writing and almost no useful facts.

 

[daw123]

This was one of the reviews I looked at. I can't remember the other review sites.
http://www.overclock3d.net/rev...uattro_1000w_atx_psu/4

And obviously the Antec website (although it clearly is for marketing purposes and selling PSUs)
http://www.antec.com/Believe_it/product.php?id=NTk=

 

[HOOfan 1]

All of this vitriol and bickering over something that should have never been brought up in this thread in the first place.

Standard practices, best practices and PSU manufacturing guidlines WERE NEVER at question within this thread until page 4, so I do not understand what bringing them into this thread has accomplished, other than to foment an argument and further misunderstandings.

cannot -something that is not able to happen. In this case a physical impossibility. If a Power Supply cannot damage components, that means it is physically impossible due to the properties of electricity beyond human control.

may not -something that can possibly happen, but is not allowed or permitted to happen. In this case a Power Supply is physically able to damage components but best practices and design guidelines dictate that it is not permitted to damage components.

Here is a legendary conditioning exercise that American English teachers use to teach their students the difference between "can" and "may"

Student- Teacher can I go to the bathroom?
Teacher- I don't know. Are you able to go to the bathroom? If you want to use the restroom, you should ask: "May I go to the bathroom"

The entire point of this thread was to teach what happens when a crappy power supply is used, that doesn't conform to ATX specifications, so what was the point of dragging best practices into this thread?

Moral of this post:

Best practices and the ATX specifications state a power supply MAY NOT damage components.
Since not all PSUs comply with the ATX specifications or best practices, a power supply CAN damage components, but if it does damage components, then it cannot be considered ATX compliant.
The ATX specification is a man-made guideline stating what is permissible in power supply design. ATX specifications say what a power supply MAY or MAY NOT do.
The natural physics of electricity determine what a power supply CAN or CANNOT do.

 

[westom]

Originally posted by: HOOfan 1
Since not all PSUs comply with the ATX specifications or best practices, a power supply CAN damage components, but if it does damage components, then it cannot be considered ATX compliant.

Only reason a power supply would not be ATX compliant: a technically naive human who created the failure. That one conclusion alone is widely disputed by computer repairmen who would blame anything else rather than technical naivety. A most common blamed source is surges. Blame what nobody understands and therefore will not dispute.

Provided were suggestions so that even layman are not foolishly buying defective supplies. And how to detect and prevent future and confusing failures directly traceable to that same mistake.

And finally, if a power supply causes motherboard damage, well, he now has a principle that identifies himself as the source of that problem. Too routinely observed are people instead blame anything else; avoid dealing with the actual mistake and reason for the additional damage. For example, deny that power supply that should never have been purchased in the first place.

How often do you always demand written numeric specifications always be provided with a supply? Why not? Is failure optional?

Routinely seen are posts that say, "My computer destroyed my power supply". Nonsense. That reason for failure is wild speculation based on feelings and no technical knowledge; or a human who is responsible for the entire mess again due to technical ignorance. Such failures do not happen even when using minimally acceptable components.

Ironically, that poster would never realize he is only implicating himself.

 

[HOOfan 1]

Originally posted by: westom

Only reason a power supply would not be ATX compliant: a technically naive human who created the failure.

If you consider companies that cut corners, to cut cost naive...I suppose that may fit.

However, you can also argue that the companies are not naive, but instead rely on the naivete of potential customers, who fail to realize that buying cheap means buying poor quality.

You have a point that often problems are misdiagnosed, and the power supply is often wrongly blamed. At the same time, however you are not there to examine their equipment, and thus you are in no position to say otherwise.

Simply stating that there are best practices and specification in place that dictate the prevention of damage by the PSU to components and vice versa, certainly is not a scientific method. The only way to diagnose a problem, is to examine the hardware and try troubleshooting.

Do you think a fire investigator would summarily dismiss an electrical problem as the cause of a fire, simply because the electrical code provides for the prevention of fires due to electrical problems? No, they go in and do some actual investigations.

 

[Golgatha]

I've never had a power supply die. I also use an APC brand UPS unit, which gives my system nice clean power. The APC also prevents damage due to brownouts, which aren't frequent where I live, but do happen from time to time.

I follow some simple criteria that have served me well when choosing a power supply.

1) I allow 150w for the CPU, and 75w (PCIe slot power) + 75w times however many additional power plugs are needed (e.g. Radeon 4870 1GB is 75w + 150w or two plugs = 225w).

2) I allow for the PSU to be up to 80% loaded on the 12V rail and tend to gravitate towards single rail PSUs. I only consider the 12V rail when looking at wattage, because the 3.3V and 5V rails will never get overloaded.

3) Warranty needs to be 3 years or better. I prefer 5 years, as this gives me an indication of how the manufacturer backs up the quality of the PSU. I also purchase expensive PSUs, so having the warranty lets me know I've got at least 3-5 years before I have to buy another PSU.

4) Previous experience with a brand and price.

An example would be my current system. 150w for the Core i7 920 and 550w for the dual 4870 graphics cards or 700w total. I have a Corsair 850w PSU with a single rail and 70A on the 12V rail is (70A x 12V) 840w usable wattage on the 12V rail. If you do the math, that's 83.8% loaded on the 12V rail, which for me is close enough to 80%, as I'm sure my estimates are a bit high given that the 4870 barely needs the 2nd 6pin plug to function.

 

[Gillbot]

Ok, this thread has run it's course. With all the bickering I see, this thread has degraded to below uselessness.
Anandtech Moderator
Gillbot