Weird PSU Issues...Thoughts?

So, a good friend of mine texted me last night and told me that the PSU on her computer went out last night, completely with smoke. This is the FOURTH time that’s happened.

Last time it happened, I replaced the PSU and recommended she invest in a good surge protector, which she did. It’s been going fine for months.

But last night, kaput…again. I am REASONABLY certain that’s it’s not the electrical system at her place, as it’s definitely up to code…but SOMETHING is going on.

It’s been suggested it might be the power cable, or environmental conditions, but I’m wondering what other suggestions there might be.

(Loyd and Jason, thanks for your suggestions via Twitter, but this would go better with more than 140 characters per message!!)

Code says nothing about wiring for transistor safety. Code is only about human safety. Furthermore, a building wired to code can actually not be wired to code. Meanwhile, even 1930 wiring must not harm a properly designed supply. Supplies contain layers of protection to make numerous anomalies irrelevant.

You don’t invest in a power strip protector. You waste money on it. If a protector is adjacent to an appliance, then it does not claim to protect from surges. And sometimes can make surge damage easier. Read protector spec numbers to see what is obvious. At best, that protector protects from transients that typically cause no damage. Nothing more.

Now, supplies must contain many functions that make RFI/EMC, low voltages, excessive harmonics, fire, etc all irrelevant. But a power supply need not meet those requirements. Only person 100% responsible for a supply being sufficient is the computer assembler. Most have no idea how electricity works let alone know only they are responsible for meeting those requirements.

How to increase profits on a power supply? Remove required functions and sell it for less money. Most assume it is good only because a computer boots. Says much about technical knowledge of many computer assemblers.

So, does that supply have all necessary functions? Well, if it comes without a full sheet of numeric specs, then suspect the worst. No specs is how a supply can sell without criticism by the few who actually know this stuff. No specs is how to sell a supply intentionally missing required functions. That supply will still boot a computer - which massively increase profits.

A supply should be at least $60 retail. That does not say a $70 supply is sufficient. Please read with care.

Are surges causing damage? Destructive surges occur typically once every seven years. A supply must be so robust as to even make irrelevant transients that are generated by a UPS in battery backup mode. Yes, some of the ‘dirtiest’ power comes from a UPS that can harm power strip protectors. And contrary to popular hearsay. Just another reason why a supply must be so robust.

So, why has the supply failed? Nothing inside a computer can cause that failure. Best is to list / define what parts inside a supply failed. A dead body is always the best evidence. Then those who know (even designed) supplies can provide useful ideas.

First suspect is design quality. Too many will buy only on price and watts rather than on what is necessary to make an informed decision. With so many technically naive computer assemblers, the market is ripe with inferior supplies. Supplies that cannot even meet industry standards such as UL.

Meanwhile, if she has not earthed one ‘whole house’ protector, then everything in the building has no effective protection. Nothing even if she has purchased 100 power strip protectors.
[FONT=Times New Roman][SIZE=3] [/SIZE][/FONT]