That is classic junk science reasoning. Some supplies are under greatest stress when only outputting half the rated power. At half power, it must restrict more power causing greater stress, generating more heat, and operating at a less efficient operating point. To say more always required spec numbers unique to each supply. It demonstrates how to identify recommendations based in junk science reasoning. Claims made without numbers (subjectively) may be classic junk science.

Another even implied a failing power supply will somehow stop a surge. Total nonsense. First the same surge current is everywhere in a path from cloud, through a building, to distant earthborne charges. Much later, something(s) in that path fails. If a surge is incoming to a PSU, then a same current is outgoing to the motherboard. Then later, something in that path (ie PSU) fails. Another example of junk science reasoning challenged.

Does high power draw cause a PSU to explode? Of course not. Largest power draw is to short all PSU wires together. Intel specs even define how thick that shorting wire must be. Why? Because all power supplies must provide all power levels and not be stressed. Because a standard test for every supply is to short all outputs together and never have damage. Another example of subjective reasoning and wild speculation trying to explain a PSU failure by ignoring ATX standards and numbers.

Most failures are manufacturing defects (ie famous defective electrolytic caps that failed years later). So many use wild speculation to, instead, blame surges. Because advertising (and subjective reasoning) is somehow knowledge. To say more required identification of each damage part. One that exploded. And the many others that have no visual indication. Most failures are manufacturing defects. Most failures have no visual indication. Those recommended protectors do not claim that protection. But that means reading numbers and ignoring every subjective recommendation. Most failures (even those that occur many years later) are only manufacturing defects.

Is your system so hot that you can toast bread? Then it requires a 700 watt supply. Most all computers work just fine with 300 watts. However, we are selling power supplies to consumers who use wild speculation to know things. And who always ignore numbers. So we recommend a 700 watt supply to them so we waste no time on customer support lines. A 700 watt supply was more than twice what was needed. But only the fewer who actually first learn numbers would know that.

This post has been edited by westom: Aug 1 2013, 09:31 PM

OVP is OverVoltage Protection. Shorting all outputs together does not create a higher voltage. Does not create an overvoltage. It creates near zero voltage. OVP does nothing on near zero voltage. OVP is obviously completely irrelevant when all outputs are shorted together - as defined even by Intel ATX standards.

Power supplies, long before PC existed, featured these many functions - standard. Little has changed in PSU functions in over 50 years. What has changed are designs that do the same things even better.

As was true even before PCs existed - any load from near zero to maximum or short circuit will not overstress any properly designed supply. In fact, normal is for a defective or undersized supply to still boot and run a computer normally. And then cause failures months or years later.

If your system is consuming 700 watts, then it can also toast bread. No PC is consuming anywhere near that power. Many myths claim otherwise. But informed consumers instead measure power.

Watts says little about a supply. Informed computer assemblers instead use current for each voltage to select a supply. Others who would not know how to do that - we tell them they need a 700 watt supply for their 300 watt maximum computer. It makes life easier for technical support. And creates many popular myths where recommendations exist without numbers.

A 520 watt supply should be more than sufficient for any upgrades. However nobody can say with certainty without accurate current numbers for each load. Or better answer this by using a multimeter that can see a defect or undersizing months before problems result. To confirm hardware is really what it is supposed to be.

Higher wattage does not reduce stress. That remains a popular urban myth. Once a supply meets a sufficient level, any larger sizing does nothing to increase reliability.

This post has been edited by westom: Aug 2 2013, 09:00 AM

Stress on components was not discussed. Higher wattage does not reduce stress on a supply. That stress remains a popular urban myth. Once a supply meets a sufficient level, any larger sizing does nothing to increase supply reliability.

This post has been edited by westom: Aug 2 2013, 09:17 AM

The PSU calculator is made by people who have little electrical knowledge, and worse, do not have specific numbers for your unique hardware.

Any layman - even a teenager - can measure his own system. For example Kill-A-Watt is a classic and inexpensive tool to have numbers. Since only those who have numbers can answer with honesty.

Or learn from so many others who measured their systems. They also discovered the 700 watt supply feeding a computer that only consumed maybe 200:
http://forums.anandtech.com/showthread.php...91#post34917091

This stuff is quite easy. Simply ignore anything recommended without hard numbers. Also ignore calculators that do not use numbers unique to each hardware device. Many just ballpark the number, double that estimation, and demand a power supply easily double what you need. Their output numbers are only as good as the number you first provide.

Any layman or junior high school student can use a Kill-A-Watt or the many other equivalent devices. Then numbers based in reality demonstrate what you really needed. And then the fewest who actually know this stuff can post even better recommendations. Remember, those who best know your answers are silenced by hard numbers you did not provide.

Why are brand name systems so reliable? View their power supply numbers. Typically around 300 watts. There is zero relationship between reliability and "more power".