Heroic Register reader battles EXPLODING COMPUTER A heroic Reg reader who battled an exploding computer to save his son's homework (and possibly his life) has written in to share the harrowing tale of a power supply unit gone MAD. Keith, who lives in Germany, wrote in to tell us how he dealt with a flaming desktop PC, and how his quick instincts to throw it out the balcony …
Ugh, why I stay away from them nowadays. I've actually had a similar experience with two no-brand PSUs. Thankfully, unlike this case, the PSU never did spew flames. But I do only buy PSUs of brands deemed trustworthy with reasonably high wattage ratings now.
Protip: The thing about these no-brand PSUs are that their wattage ratings are often very much overrated and their actual maximum output is actually much lower than advertised. For example 250w PSU would often be labelled as 500w one. Net result is that the consumer will buy it and fit it into a PC that needs 500w, and thus stress the PSU and cause it to go out with a bang, fire optional.
A PC that actually actually needs a 500W power supply is a rare thing indeed. Even those requiring in excess of 250W continuous load are few and far between, falling very squarely into the enthusiast category. Any enthusiast installing a no-brand PSU in such a situation* is either broke or not thinking straight. It's not so much that the continuous rating of cheapo supplies is insufficient, more that they are ill-equipped to deal with anything more than the smallest of transient peaks. And that's before you even consider the cheap components and shoddy manufacturing processes that contribute towards non-overload-related random catastrophic failures like the one in this article.
FWIW, having read up on a few PSU manufacturers' processes, the one that most thoroughly torture tests as part of production line QC is Seasonic. I believe they are the OEM manufacturers for a number of high quality PSU brands.
These were Antec TruePower 480W units, don't know who OEMed them, but I had one fail, replaced under RMA (which took an inordinate amount of time too), then the other failed just out of warranty, and then shortly after the replacement unit went too, all with low-ish and wildly unstable 5V rails.
No more Antec for me.
Hear hear.
My daughter bought a replacement laptop PSU from Leeds market for £5. She complained the laptop's charging socket must be broken (I bought her a new HP business laptop to go to uni with) as the thing would charge for 10 minutes then shut out and she had to wiggle it about to get it to start charging again. Annoyingly, none of the official HP PSUs I had shipped to her fitted the laptop physically. For some reason the listed part number was wrong, as was the part number stencilled onto the side of the original PSU.
When she eventually brought it home with her, I put it on my test bench and, sure enough, it charged for 10 minutes then cut out. I went to unplug the cord and burnt my fingers quite badly on the plug. After letting it cool and dissecting the wire, I found the actual copper core was so thin I wouldn't even use it for a doorbell, and the soldering in the plug was so bad that it was heating up the plug and tripping a thermal cutout HP had thoughtfully provided next to the charging port.
Needless to say she got an extra lecture or two once I'd found out exactly where she'd bought the PSU from and how much she'd paid.
Last PSU I bought was a fanless Seasonic for around a hundred quid- over kill at 400W, as it's supplying a system with no dedicated graphics, SSDs and a low-power 'S' variant of a i7 3770 CPU- but blissfully silent. The computer death that prompted this upgrade was an old Dell Optiplex with blown capacitors on the motherboard- leading me to learn about an episode of industrial espionage: http://www.badcaps.net/pages.php?vid=4
reminds me of the death of my treasured Epox 8RDA+ based PC ....... http://spectrum.ieee.org/computing/hardware/leaking-capacitors-muck-up-motherboards
A spot of failed espionage leading to a few million hopeless and doomed capacitors hitting the market..... Failure mode was usually along the lines of capacitors swelling and then failing, but occasionally along the lines of a loud bang, accompanied by a flash and tiny tatters of wax paper and aluminium, foil drifting around...... Very occasionally fire...... I wonder if any of those caps are still in the supply chain....
If I remember rightly when custom PC torture tested a load of PSUs even most of the named ones couldn't handle the load they claimed and one exploded, Seasonic got mentioned because 2 companies withdrew their brands from testing, but Seasonic went back and asked Custom PC what was wrong and then went and redesigned their PSUs.
I have to agree btw my Seasonic PSU was a little more pricey but is damn good.
No-Brand PSU's use cheap parts that last nowhere near as long and often have very poor power efficency. If its generating more heat than DC power then it'll not only cost more on the leccy bill but often fail taking out anything its powering.
Don't spend £30 on something which will provide power to parts worth at least £200.
cornz1,
Thanks for the tip. I shall upgrade my production line immediately, by the introduction of small water tanks to my PSUs to add weight and fool expert PC builders. What could possibly go wrong...
El Cheapo PSUs: Power To You (possibly directly if the guy doing earthing was hung over that day).
[small print: We absolutely guarantee that our PSUs won't kill you. We will pay £1million to any person who has been killed by any of our shoddyfine products. The victim simply needs to apply in person to our head office, to receive their pay-out]
A bit clunky, a water tank on top, could really mess up the innards as well.
Better might be a little cylinder of Halon, poof and the fire is out.
Mind you, Halon doesn't cool, so you'll need a second or third shot of Halon available, means money.
More elegant, possibly least weight, might be a small air pump attached somewhere. The air pump suction should be at an opening in the casing. Start the airpump and voila!, the resultant vacuum starves the flame. The little airpump can run and run. make sure the airpump suction is the only opening in the casing.
Perhaps you could have a reversible cooling fan!
Or cast iron computer innards, don't care about a little excess heat.
Just speculating....
I thought it was the PC that caught fire - not just another iffy PSU.
I think I'd have just rescued the hard drive and not have poured water everywhere.
Wouldn't have bothered to try and save the rest of it - data is more valuable than a few chips.
(my last PSU began smelling of fish and then popped with a lovely spike that took out the RAM and Mboard)
No legitimately CE marked PSU could fail like this unless there is a manufacturing defect, so either it's a bad one and you get your money back, or report the supplier to Trading Standards (or local equivalent) for selling dangerous goods with an improper CE mark.
You'll know which when you ask for your money back, and request incidental damage payments as well although I'm not certain of the law in this case.
- Note that it is the importer, not the manufacturer that's responsible for the CE mark.
Even "reputable" PC PSUs are so variable it's crazy - I've seen some dangerously shoddy soldering inside some well-known brands.
> I've seen some dangerously shoddy soldering inside some well-known brands.
It's worth checking the approvals on the case. If it is UL listed and you can read the logos of some of the approving agencies in Europe, it is worth contacting them and seeing if they want the PSU back for investigation. Test agencies really love having exhibits for their black museums, as it helps justify their existence.
Kick up a bit of stink because you should get a new computer at least out of the PSU supplier. Remember that even if a PSU is overloaded it is required to degrade gracefully - unless you actually stuck a screwdriver inside and waggled it around, there is simply no excuse for catching fire like this.
And yes, I have worked with approvals bodies in the past, and been involved in electrical safety testing. You probably have at least one product I have been involved with in your house.
Five years old, but still valid.
Hmmm, I recall an associate attempting to convince our boss that since the smoked had escaped from said boss's computer that the electronics would no longer function. We assured him that electronics ran on smoke and that once the smoke had escaped, it was nearly impossible to to put back. In this case it was a harddrive whose onctroller board suddenly emitted a "snap" and a puff of smoke. From what we could tell, the only unbacked-up data was his stash of scanned pr0n. Everything else was backed up daily, weekly and monthly. He was in deeo denial for a bit, then acquired his own personal scanner.
With PSU failures it's often difficult to know if it was faulty - or just an extreme reaction to another component failing.
My worst was a friend's PC which had been switched off for a few weeks in a move. He flipped the mains switch a few times in frustration when it didn't power-up - and then reported a bright blue flash.
On inspection - the Enermax PSU fuse internals had vaporised. Component testing showed that the motherboard was also dead - including the ram and video card. Unexpectedly both hard disks, DVD-writer, and floppy were also Norwegian parrots. The only electrical bit that had survived was the cpu fan and case leds.
There should be NO extreme reaction to another failure. Power supplies are a part of the system that are REQUIRED (sorry about the caps but I don't think I can do html here) to withstand electrical abuse. They are the guardians of the downstream equipment.
That said, repeated switch flipping can have unexpected effects. Sometimes rcbos trip because the suppressor capacitors draw too much asymmetric current at switch on. Because they hold their charge for a second or two, two flips in a short interval can sometimes get a shutdown ring main up without having to disconnect everything. But with many cheap PSUs, charging the live side capacitor can result in a large surge current. Even in the days of 286 and 386 computers, early computer PSUs could easily spike 40-50A during the first half cycle. Each time you switched the thing on, failure came a bit nearer. Since repeated surges through time delay fuses gradually weaken the fuse, it isn't that hard to design a circuit so that the fuse should blow before another component is likely to fail due to repeated thermal or magnetic field shock. And the capacitor should fail open or short; in the first case the thing simply doesn't work, in the second the fuse blows. Even if the capacitor blows up, the shrapnel should be retained by the case and if anything gets shorted the fuse should blow. Glass fuses can blow with a flash. So can some other overcurrent protectors.
So, answering your point, PSUs can indeed fail due to external component failure but usually fail due to their own highly stressed components going wrong. But they should never fail dangerously.
Tried using a power supply to charge a laptop, it didn't like it even with a motherboard attached as a dummy load.
Ran for about an hour then KABLAMMO!
Had "failed" power supplies caused by bad capacitors on the mains side, these have a tendency to go O/C for no apparent reason causing audible whining noises and eventually blowing the fuse.
Same with output ones (more common)
Many years ago, a colleague and I had the task of commissioning a roomful of shonky XT clones. Being the sorts who just had to find a way of making this more interesting, we scrounged every set of DOS diskettes we could lay our paws on, loaded all the diskette drives and took half the room each. Last to finish buys the beer.
My side went <Clack>, whirrr, <Clack>, whirrr all the way along. His got half way and produced <Clack>, <BANG!!>. That'll be the one with its PSU switched to 110 volts.
Also there's this. There are some things made by man in this business that even fire will not kill.
"That'll be the one with its PSU switched to 110 volts."
A PC came in for repair with failure symptoms that sounded like an instantaneous PSU pop. It transpired that the owner had been raided by the police on the periphery of a larger investigation that was going nowhere. After examining his PC for several weeks they found nothing to substantiate their hopsuspicions - so there was no further action.
Unfortunately while being returned the PC had mysteriouaccidentally had its PSU switched to 110 volt mode. Surprisingly it had survived a few weeks before popping.
A decent PSU will often not contain Acid paste electrolytic capacitors, so not fail so messily; reasonably priced PSUs can now be found with solid capacitors, so will not cause this mess. The ironic thing is, the most stressed capacitors are on the AC side of the switching transformer, so there is no reason why any powered components should be taken out by a failed PSU!
The mobo and CPU would have been fine if a powder or better a CO2 fire extinguisher had been used; any sensible home owner should have at least one fire extinguisher suitable for electrical fires; these are regularly on sale in discount shops, often for cars, so no excuses!
"[...] any sensible home owner should have at least one fire extinguisher suitable for electrical fires; these are regularly on sale in discount shops, often for cars, so no excuses!"
It seems strange when advocating avoiding cheap PSUs - to then recommend buying cheap fire extinguishers. A one-off emergency device needs to be guaranteed to work reliably. Even apparently branded items from a discounted source could be fakes.
A survey a few years ago showed that many powder extinguishers sold in car shops had insufficient capacity for a fuel fire.
You can buy kitemarked powder extinguishers at very reasonable prices from marine chandlers, as they are a legal requirement on boats.
Afloat they have a shelf life of 5 years because of corrosion and powder compaction due to vibration. Ashore, just checkweigh the carbon dioxide cylinder periodically.
Infernoz,
Powder extinguishers need regular maintenance, or they refuse the work. The powder gets compacted, and won't come out when you press the button. I'm not sure whether this maintenance is simply turning them upside down for a bit though. There's also a bit of a knack to making powder work, which you're probably not going to get with your first go, and only a small extinguisher.
CO2 has it's uses in the home, but there are plenty of places you wouldn't want to use it in the kitchen. Unless you fancy a face full of burning fat.
Apart from the obvious point mentioned above, that you're advocating buying cheap, crap safety equipment from discount stores - you're often more of a risk to yourself than to the fire when it comes to first-aid firefighting. I'd suggest that the best thing to have in the home is a fire blanket. Anything you can't easily put out with that, or just turning the power off, or sometimes a bucket of water, is probably dangerous enough that your best bet is to run away and call the fire brigade.
"CO2 has it's uses in the home, but there are plenty of places you wouldn't want to use it in the kitchen. Unless you fancy a face full of burning fat."
Reminds me of our training session by a visiting fire brigade crew. We all had a play at dealing with a petrol fire in a large metal tray using CO2. They showed us the correct technique to smother the fire. It worked ok for a few people until the tray became hot - and then the petrol it would instantly re-ignite. A powder extinguisher would have prevented that.
A week later our team leader lit his pipe, shook the match to extinguish it - and dropped it in his waste paper bin as usual. Unfortunately in the bright sunshine he hadn't noticed that the match was still burning. The bin was full of plastic coffee cups which proceeded to burn vigorously.
A great chance to put our training into practice. Grab the CO2 extinguisher from the corridor - remember to remove the safety pin - point the nozzle at the waste paper bin - and give it a quick burst. The fire was out instantly.
However the mistake was to put the nozzle very close to the burning mass - and aimed into the bin. The CO2 burst ejected the contents high into the air - and the burning plastic had produced lots of small sooty particles that gently descended on the furniture like black snow.
Damn, you must be having some seriously cheap hardware there.
First, I've managed probably thousands of PC's, 50% of which I was supporting rather than specifying and most of which have been 3-5 years old or even more. I've sometimes inherited the most cheap networks, with the most horrendous repairs/bodges known to man (how about a major UK educational supplier who advise you, when the caps blow on a motherboard, to buy PCI cards in order, because they know what order the caps will blow and what functionality you will lose over time as they do so?!).
In over ten years, I have had precisely one PSU do something a bit odd (and also the closest experience I've ever had to "fire" from a PC - by my statistics, NiCd battery chargers are about 10 times more dangerous). A PC was reported to have a "burning" smell that was present whenever the computer was turned on and, undeniably, was coming from the PSU. No smoke, no flame, no boom, no pop, hell the computer just kept on working. But the PC was nearly 15 years old and only used because it ran some ancient piece of software that only run on Windows 95.
And, to be honest, a PC is a sealed metal box. Sure, if you have things close enough to block the vents then a fire could be a severe hazard to your health (and I was concerned enough about the burning smell to instantly condemn the machine above) but otherwise, it'll just burn to the point that the fuse blows in the plug (you *did* have a fused plug and outlet?) or the PSU, and then pretty much stop. The only source of ignition is the electricity, the only fuel a flame-retardant motherboard and some plastic on the cables / fans (which, by dying, take out the main source of oxygen), the only way to escape out of the vents at the rear which you should be keeping WELL clear anyway if you don't want a fire to start.
Out of all the electronics I use every day, and all the office equipment I see used, the PC's and their PSU's (even if replaced with ones from 10-year-younger PC's, etc. when they go wrong, which I have personally witnessed) are probably the least risky things. Hell, block a printers vents and you have an instant fire hazard as several call I've received in my office will testify to. Lovely heated element, little ventilation, huge temptation to stack paper over the fans and vents and, believe it or not, a huge fuel source that gets fed through the device by an automated process. Just lovely. (Yes, I have received at least two calls along the panicked lines of "The printer is smoking, what should I do?" - SWITCH THE DAMN THING OFF AND GET AWAY FROM IT, MAYBE?).
Hell, much more of a risk than anything else I see are probably things like extension leads, dodgy cables that have been run over by office chairs (though regular PAT testing tends to eliminate such problems over time), electric heaters and printers. And I'm much more personally worried about UPS's than PSU's. Those things scare me. Some of them have car batteries in them, provide lethal voltage, and they DON'T SHUT OFF. If one of those goes wrong, you'll find me pushing the fire alarm, not faffing about trying to get it outside.
P.S. Picking up a metal case that's on fire, and throwing water on a metal / plastic / electrical fire probably isn't the best thing either. Probably would have done better just switching it off and covering it with a blanket.
"Picking up a metal case that's on fire, and throwing water on a metal / plastic / electrical fire probably isn't the best thing either."
You're probably right, but it worked. I could visually see that it was burning in the back and not in the front, the front is where I picked it up when I tossed it out on the balcony. The balcony is also where I doused the little flaming demon.
I have a 10kg CO2 extinguisher in a cupboard by the front door. This puts out fires without making a mess on the carpet. Before getting one read up on the effects of CO2 poisoning, or if you prefer, die.
Paris, because she lights my fire and no amount of CO2 would put it out. Cold water would work much better for this special case.
And so expose it to the outside, via high-powered cables that have to run external to the box (and thus subject to wear and tear), outside of the nice largely-inflammable, and properly-earthed box it's already in? Yes, you can if you want. Mine will stay where it is.
And on a side-note, I have just taken delivery of 25+ all-in-one touchscreen PC's for my workplace. They use 19V external power adaptors, the same as a laptop, but have pretty meaty CPU's in them (their graphics are pretty dumb and basic). They get incredibly hot and connect via a single 19v DC cable to the computer itself, which is liable to pulling out, yanking, tearing, etc. and then has to be stepped down to the internal voltages of a normal motherboard and SATA drive inside the machine anyway. They also arc if you don't push the three-leaf-clover cable into them fully. And that's before you even get into trying to make them for high-end gamers PC's where 500-1000W isn't unusual (good lucky getting at least 500W of DC down a single cable that's exposed to users - that's about 5 times your most powerful laptop PSU). And then the step-down circuitry would mean either a huge PSU in the machine anyway, or a huge one in the external PSU and a very thick cable to transport them all, correctly isolated, to the machine itself.
The number one cause of power problems with laptops that I get are: broken cables, "dropped" power supplies (i.e. knock them off the desk, usually knocking the coils off the internal boards and killing them outright), and overheating power supplies (because they tuck them down the backs of sofas / desks without thinking.
The PSU is NOT anywhere near the most dangerous PC component. Go look at your printer and the vents on it, then wonder what happens when you block the vents with stacks of paper, on a device designed to print on stacks of paper by automatically feeding large amounts of said paper through a heating element that runs hot enough to char paper if it stops mid-way...
Yes, I would, thanks. (I'm not being sarcastic, this kind of thing is how I earn money).
If you're worried about the horrors of our libel system, give me an obfuscated email or something here and I'll write you and ask. I don't use my real name so I can comment freely. I can't use Linkedin because they have 12 of you.
Thanks.
I'll post my most-verified one so as to not make a mistake concerning manufacturers, etc.
Samsung CLP-300 series, including the Xerox I-can't-remember-the-model that's a rebranded version of it in a white case, also including the "300N" versions that are networked and the CLX- derivative with the flatbed scanner attached. (all use the same internals on different machines).
Have, in the past three years, consigned ten or more of those to the scrap heap because they allow the paper exit area to block the paper (when someone puts something on it / paper gets jammed normally / people just print too much and don't collect it from the exit tray), which then allows the paper to REFOLD back into the paper exit area roller. You end up with a circular tube of paper that sits right on top of the heater elements, which DO NOT SWITCH OFF and allow the paper to first char (to the point the paper you get to remove is crispy and the lettering burned off), then smoke (which is the point at which users usually phone up) and AT NO POINT SHUT OFF (because the heater element isn't actually overheating, it's just the paper above it that does).
I have personally removed AT LEAST ten burned circular tubes of paper from printers that jammed or had their paper exit area blocked, and subsequently smoked to the point that users noticed before anything dangerous happened (P.S. I work for schools, and printing unattended is fortunately very, very rare!) but which private tests confirm would happily go on until the risk of ignition was more than significant (especially if the paper exit area was blocked by something flammable like... paper). Additionally, at no point was any vent blocked because they are on the side but that MAY cause a thermal fuse to trip if you did that, but it wasn't necessary to cause a hazard. They didn't trip in any of the cases I witnessed and the only "stop" was when the user switched it off (the printer itself tends to just keep on printing and putting more paper onto the roller until the unit jams or the roller stops turning, which in itself raises the risk because the paper goes from being on a heater rotisserie to being directly grilled). In some cases, this was literally MINUTES later when they had witnessed the smoke, thought it strange, sniffed, checked, found the source, called me, I'd screamed at them to power it off and when I ask, the printer was STILL smoking and still feeding paper right up until they'd turned it off.
Very popular model from a few years ago as it was the cheapest "big name" colour laser and takes very cheap toner "pots" rather than huge cartridges. Now "discontinued" but not, from what I can tell, because of those problems. We've replaced them now, because of such problems and their discontinued nature, but I know a lot of places still use them.
Thanks for the info, though I have to be honest and say I would never have recommended the CLP300 to anyone - it is actually very expensive per page.
As someone long involved with electrical safety, cost reduction and printing, my home printer is a large Ricoh gel ink printer, which uses hardly any power, is cheap to run, and is most unlikely to catch fire. But nobody buys them.
I can respect the problem was largely because of the budget psu, however im also thinking that the getto fan install on the side of the case was very likey what contributed to its death.
Next time buy a decent case that capable of exhausting the heat instead of getto fitting fans into your old case because its running hot, causing pressure issues, hot spots due to bad air paths, recirculation and side turbulance in the intake of fans causing an ineffective exhaust.
I've seen several cases that come with vent holes, fan mounting points and those funnels that are meant to end up over the CPU, and at least one that looks exactly like the case pictured here, so strike the 'ghetto fan install'. The only thing people need to be aware of is that there should be enough air intake capacity, to match the exhaust capacity of the fans blowing outwards. Which means either a large grille in the front, or that side fan blowing inwards.
Looks like a perfectly regular case fan which has been fitted to the case using the ventilation slot provided by the case manufacturers for an additional fan. Most cases, including the very high end ones (which, admittedly, this one isn't) don't come with fan slots fully populated, instead they provide locations for enthusiasts to mount their own fans as required beyond the basic number which they provide with the case.
I doubt very much that the 120mm case fan was the cause of the fire, and I don't really see why you are calling it 'getto' (sic). If the ventilation was (badly) self-drilled or the fan was stuck to the case with sellotape that would be 'ghetto', but I can see nothing particularly out of the ordinary in those pictures.
Actually looking at the picture i suspect the side fan, which is mounted so it will exhaust is mounted over what would normally be a passive intake, the holes are normally much larger for force air (fan) mounting points.
So you have the cpu intake (which is designed to feed a stock cooler and not an oversized cooler like the one fitted) drawing air in, the fan next to it is blowing out, which means like some recirculation.
Plus because of where it was exhausting from your fighting the rear fans for the hot air and brining hot air forward into the top of the case instead of trying to keep to the golden cold in the front hot out the back.
Working in a building that's an annex to a hospital, our mains supply gets interrupted as rudely as possible one a month whilst they test the emergency supply. The off part isn't the problem. When they throw the breakers at the substation to turn it all back on, there appears to be a massive spike that travels along the neutral wire attempting to reach ground wherever it can. That appears to be very often through the PSUs of PCs whose owners have decided that switching it off at the mains socket is just as good as unplugging it, like they are repeatedly told to do. The switch mode controller gets stressed by this and gives up the ghost at some point in the next few days. The usual symptom is that it starts with my PSU tester connected and sometimes if you disconnect the HDDs etc. A swap out to a known good usually settles the matter.
Without fail I have two or three bad PSUs a month. Oddly enough, I've only once had an Apple go down to this, and that was one of the blue and grey G3s that used modified generic PC PSUs. Maybe Apple owners have just learnt that it takes forever to repair one and costs a bomb?
Anyway, so I keep a stock of three OCZ 600Ws on the shelf. I can get a PC up and going again within 15 minutes. Or so I thought.
Whip off the case of someone's 6 month old Fujitsu... f***ing 12V only PSU. I don't have a tester for it, and I haven't got a spare 12V. Three weeks later, the replacement part arrives, costing around £60 + delivery. And can you get a pre-built PC with an ATX supply nowadays? Unlikely. No matter how much I rant at the suppliers they are determined to go down this route where you cannot replace a PSU without going to them, and you cannot uprate the PSU (say to match a GPU board) at all.
Be warned!
Over- AND under-voltage protection (OVP, UVP)
Over-current protection (OCP)
Overload, overtemperature and short circuit protection (OLP, OTP, SCP)
No-Load operation protection (NLO)
AND - if it doesn't have a digital protection chip it doesn't really have proper SCP or OTP no matter what it says on the tin.
If the PSU doesn't have ALL of these don't buy it.
If you can't identify the PSU and check it's specs on-line don't buy it.
And remember 550W can often mean <400W - you need to be checking the 12v rail total cos that's what all your bits are drawing on. Unless you have a real esoteric set-up you are never going to be undersupplied on 5v or 3.3v rails. Check your GPU and CPU load requirements, add 50 for your board, another 50 for RAM, drives etc and check that against the 12v total for full load running - if you are running a high end gaming rig you could end up well north of 500w.
And all of that is before we get to the effect of a cheap PSU on your leccy bill.
Cheap PSUs will burn - it's just a matter of time - and while it is more likely to happen under load conditions while in use and therefore attended only put one of these in your HTPC if your house insurance is up to date.
Check tom's for this excellent article on how to buy a PSU.
http://www.tomshardware.com/reviews/power-supply-psu-review,2916-9.html
I've had a desktop PSU blow up when not running. A friend had the same -- his left a scorch mark on the wall. Safest to switch off at the wall socket or rocker switch on PSU (which some still have).
But my biggest gripe is the 2 pence fans they put in most PSUs. Within 12 months they can start to squeal and oscillate on the bearings so that they run below proper speed. At least you get a warning and replacement is simple and cheap enough -- but why not put in slightly better fans originally ?
" but why not put in slightly better fans originally ?"
In my distant youth an old timer explained how domestic radio sets were made. The prototype was designed and built according to the best practice and quality. Then selected components were removed/downgraded until it stopped working. This established the most economical production build.
A friend of mine designed the electronics for videorecorders, back when they were more than a big chip, a smaller (but still biggish) chip, and a handful of discrete components sprinkled around them. He was told not to bother whether he could save on transistors; his time thinking about such trivialities was more valuable than those parts. Once he was done, the schematic and prototypes would be scrutinized by the Parts Decimation Team, who would do what their (unofficial) name suggests while trying to keep the functionality and quality more or less unaffected.
All this talk of powder fire extinguishers and not one mention that for a small fire a packet of baking soda is all you need, even for a fat fire (though I'd go with the tried and tested damp tea-towel myself in that event. Have done, in fact.
Baking soda also cheaper than fire extinguisher (and so can be changed more often) and can also be used to deodorize the fridge (which a fire extinguisher can't). You can even cook with it, which is most assuredly contra-indicated for most fire-retardant chemicals.
This message brought to you by the Small Fire Extinguishing Council, a subsidiary of The Amalgamated Conglomo Baking Soda Mining Consortium.
I, to, have had experience in dousing faulty components. In my case, though, it was a no-name motherboard with a short that was feeding 12v into a component that was designed for 3.3v. Exactly how that led to a fire instead of just a burned out component I'm not sure, but I have never bought any component from a brand I don't recognize since.
Unfortunately in my case nothing survived because the fire started in the bottom of the case and ate everything above it. Thankfully I had all me important data backed up.
A friend had a PC which had had problems since he got it. Prone to crashing, etc.
After a couple of years of torment, he asked me to check to see if I could get it going again for him to extract some of his old files.
The first thing that struck me as I took it out of the car at home was that it rattled quit a bit more than a PC should, even when turning it slowly. Once I had it on the table and the side dropped off, I gently rocked the case and isolated the rattle to the PSU. As it was never in warranty, I removed the PSU and cracked it open. Trying to discharge the caps to avoid an unpleasant shock, I noticed that the components moved a bit under the probe's pressure.
So out came the long-nosed pliers and I pulled on a component .. which came clear off the board. And so did the next. And the next. Releasing the PCB from the PSU chassis revealed that there was no solder at all. Electrical contact existed purely by pressure of component leads onto the copper of the PCB. Looks like it didn't get wet when sailing across the solder seas.
I'd cut through the QC sticker when opening the PSU case. Obviously, placement of that sticker had been quality controlled.
Replacing the PSU, brought the machine up and running a rescue system without a problem. The machine was riddled with malware which was knobbled/removed before booting the installed OS. The installed OS turned out to be unlicenced and was probably a pirated version (being from a different region) so no updates/patches were possible, nor was, as a consequence, the installation of recent versions of anti-virus software.
You buy a three Euro PSU for a enthusiasts PC and that is often what happens be it the fan or the PSU power circuit.
BTW, I would never use a dry chemical fire extinguisher on any electronics, stove, car, etc. That's what they make halon and newer, safer chemicals for. A quality extinguisher will last a lifetime and is rechargeable. They won't create more damage than already exists as dry chemical does.
I once had a really good quality £100 PSU blow up (one of the ones with the aviation modular connectors) but luckily my motorboard survived. Nothing to do with the PSU though.
I had a picture hanging above the computer on a nail and a month or so before the the event the picture fell off the wall with the nail lost in the rats nest of cables. Turns out the power socket had worked its way loose from the surge protector strip and there was a 3mm or so gab between plug and socket. Some how the nail found its way in between the plug and socket and caused a short the BANG gave me quite a fright it did!!! ;)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
