the gutless, bloated, and fried power supply hall of shame

[Pentium4]

Originally Posted by Behemot

The questino is, are they both really the same except for peak current?

Okay, well...maybe not. They are a MOSPEC F16C20C and a MOSPEC F12C20C

[goodpsusearch]

This is not a good idea. I would put 2xF16C20C in parallel. Actually I have done it many times in the past with success.

[kc8adu]

heres another one

Originally Posted by momaka

Those Chinese PSU names are funny. Though still not as funny as PowMax, IMO.

[mariushm]

here's the power supply that died in my father's computer - see this thread if you don't know what I'm talking about : http://www.badcaps.net/forum/showthread.php?t=33718

Claims 450w , 3.3v @ 24a , 5v @ 35a , 12v @ 16a .. uses 16a diodes on secondary for all rails.
Primary capacitors claim to be 330uF - I don't care enough to measure them, I'm sure they're probably 220uF. 13007 transistors in theory could give it 250-300 watts but no way with that main transformer.

Power supply still 5v standby but when i try to power it on, there's a fain fizz (barely audible) and it doesn't turn on. Assume over voltage protection or some diode shorted.. not sure if I even want to spend time to fix it.

Shame about the nice black pcb with nice silkscreen.

Not much else to say... see the pictures below:

1.jpg
2.jpg
3.jpg
4.jpg
5.jpg
7.jpg
8.jpg
9.jpg
10.jpg

[PeteS in CA]

That thing's so wimpy that 200W might be pushing your luck, and I'm not sure what the worst weak point is. The AC I/P rectifiers are probably rated for 1A. Even if they really are 330uF, those I/P lytics are nowhere near what they should be for 450W. The heatsinks are somewhere between painfully thin and anorexic. Even the number of output wires on the high current outputs looks worrisome.

But at least you didn't have lots of beefy components and wires to obstruct getting good pictures! And you don't have to worry about unsafe X and Y caps.

[mariushm]

I actually checked those diodes after taking the shots. I think it says RL257 on those and in that case they're rated for 2.5 A each:

http://www.rectron.com/data_sheets/rl251-rl257.pdf

It could be that the manufacturer is overly optimistic or... yeah, datasheet shows them rated at 2.5 A for up to 75c .. at 100c they can only do 1.5 amps.

It's sad these things are sold in my country and people like my father without electronics knowledge fall for these (and often in small towns with just one IT store you don't have alternatives). He simply went with the whole computer to the store and asked them to change the power supply after previous one failed.

This power supply was 50 RON, which is about 15.4$ and that includes our country's 24% VAT.

So the distributor probably brings them into the country for around 9-10$ each...

later edit: what's interesting is that this power supply as weak as it is, it uses that WG606P which apparently is a TNY267 clone. I was expecting the standard two transistor job for 5vSB or something cheap, but there's even a basic pi filter on 5v sb.

[TELVM]

Originally Posted by mariushm

... not sure if I even want to spend time to fix it ...

Not much worth it IMHO. Even if you upgrade everything else, output filtering lacks PI coils and there is only one cap per rail . There are no empty parking slots to improve that.

In Clube do Hardware's 'Produtos Bomba' section (a heaven for the gutless wonder scholar ) every turd with that "output filtering configuration" shows hopeless levels of ripple:

Example 1

Example 2

Example 3

[PeteS in CA]

I actually checked those diodes after taking the shots. I think it says RL257 on those and in that case they're rated for 2.5 A each:

http://www.rectron.com/data_sheets/rl251-rl257.pdf

It could be that the manufacturer is overly optimistic or... yeah, datasheet shows them rated at 2.5 A for up to 75c .. at 100c they can only do 1.5 amps.

Nah, I just misjudged the rectifiers from the picture. At 2.5A they're less under-sized than I thought, but still probably not good - with reasonable derating - for more than 400W. Not having PFC, the peak currents those things conduct at the peaks of each half cycle must be pretty high (looks like it can handle on the order of 12A repetitive peaks).

It's sad these things are sold in my country and people like my father without electronics knowledge fall for these (and often in small towns with just one IT store you don't have alternatives). He simply went with the whole computer to the store and asked them to change the power supply after previous one failed.

This power supply was 50 RON, which is about 15.4$ and that includes our country's 24% VAT.

So the distributor probably brings them into the country for around 9-10$ each...

This broaches a couple of things that really bug me. Painting with a very broad brush ... North Americans and Western Euro-Folk are freaking spoiled when it comes to having good stuff available - speaking as a North American. Eastern EuroFolk, and folks from Oz and KiwiLand have it less easy, and much of Asia, Africa, South America and Oceania, good stuff is hard to get, if at all. Or at least that's my perception. Those who make the good stuff go where the $$ is, and less prosperous nations are left wide open to shadier operators.

When it comes to safety agencies, North American and Western EuroLand agencies really are legit - creating meaningful regs and having enforcement behind them. To be blunt, I think "safety" agencies in many/most third world countries are means of extorting $$ from companies that make good stuff, but lack the enforcement teeth to keep out unsafe @#$% from shady fly-by-nighters (and seriously, who's the bigger target, multi-B-$$ companies like HP and Sony? Or No-Name companies making a few million $$ a year making @#$%?).

And then there's the grey areas of not-so-good stuff, brought into whatever country a bit shadily, sold by retailers who may or may not know what quality looks like, and sold to non-technical customers who trust the retailer to sell products that can safely do what they claim to do. That's a plague affecting every country to some degree.

I wish I could say that I see an good solution, but government regulation easily becomes weapons to stifle innovation and competition, and to pay off cronies (e.g. Solyndra and more) while punishing political enemies (e.g. Gibson guitars). OTOH, there are plenty of shady and dishonest people eager to make a few quick $$ off of people's trust and ignorance.

[rogfanther]

Originally Posted by PeteS in CA

This broaches a couple of things that really bug me. Painting with a very broad brush ... North Americans and Western Euro-Folk are freaking spoiled when it comes to having good stuff available - speaking as a North American. Eastern EuroFolk, and folks from Oz and KiwiLand have it less easy, and much of Asia, Africa, South America and Oceania, good stuff is hard to get, if at all. Or at least that's my perception. Those who make the good stuff go where the $$ is, and less prosperous nations are left wide open to shadier operators.

When it comes to safety agencies, North American and Western EuroLand agencies really are legit - creating meaningful regs and having enforcement behind them. To be blunt, I think "safety" agencies in many/most third world countries are means of extorting $$ from companies that make good stuff, but lack the enforcement teeth to keep out unsafe @#$% from shady fly-by-nighters (and seriously, who's the bigger target, multi-B-$$ companies like HP and Sony? Or No-Name companies making a few million $$ a year making @#$%?).

And then there's the grey areas of not-so-good stuff, brought into whatever country a bit shadily, sold by retailers who may or may not know what quality looks like, and sold to non-technical customers who trust the retailer to sell products that can safely do what they claim to do. That's a plague affecting every country to some degree.

Yeah, at least here, good stuff is somehow available, but at prices much higher, comparatively, that in other places.

Also, too many sellers go just by the brand, so an entry-level Corsair psu, for example, is sold at almost the same prices as the top level ones, with the former being touted as a "special offer " to justify why its price is a little lower.

And there are not good, reliable middle-level options for the "casual" build. The cheap ones, at about US$ 15-20, are crap. Then there are the others, at about uS$ 50, that use to give stable 350W when sold as a 500W. A good 500W or similar runs up to US$ 100 or more. So, when replacing a psu for a low-spec machine, like ones used in a pizza place or like that, the decision is between using a crappy psu, hoping it will endure at least 6 months, or recomending a good one, and having the customer think we charge too much and going to another place , that will install the same crappy psu and say it has "Nominal 500W". And the customer thus allow himself to be fooled, and will blame ligthning or the energy company when his pc is reduced to ashes.

I´ve been trying to keep some spares from good brands, most obtained from dead HP or Dell machines, and recapping them. Almost the same price as a new shitty one, and at least I can have some confidence that it will not go up in flames ....

[Behemot]

Originally Posted by mariushm

later edit: what's interesting is that this power supply as weak as it is, it uses that WG606P which apparently is a TNY267 clone. I was expecting the standard two transistor job for 5vSB or something cheap, but there's even a basic pi filter on 5v sb.

Cannot do with two-transistor forward! Rememer EUs legislatioin, as far as I remember, Romania part of ESSR so it has to comply. For quite some time they demand less than 1 W stand-by drain (with no load) so even cheapest craps since had to do with those PWM-with-integrated-transistor.

FSP Group uses 2 transistor still, but they are able to do it so it does not burn power for nothing and even has some nice efficiency under load.

[mariushm]

You must be joking. This power supply doesn't care about regulations.

Legally, it can not be sold either way because any power supply above around 150w sold in EU must have at least Passive PFC...

Well, technically - I'm not sure it has changed in the meantime, probably did change - but a few years ago the legislation only said that the amount of emi and other noise related stuff must be below a threshold and it just so happens passive pfc or active pfc was the easiest way to achieve that.

But this article says it's a requirement nowadays: http://www.edn.com/electronics-news/...power-supplies

Nobody checks these power supplies to see if they conform to standards here in Romania.

I don't know. My guess is they got a sweet deal on that chinese tny clone and figured it's cheaper than getting a to-220 mosfet plus insulator sheet plus washer and screw.

And btw, about that idle power rule... I think it's something like <0.15w idle in the latest versions of the standard.

[Behemot]

I am typing like animal, should go sleep…

Well, most cheap craps around I've seen lately at least comply with this one, though it's worse with power factor correction. It is good for PR when you state low idle consumption and really, price for those chips are very low lately, jsut everybody makes them for couple cents. Latest version of that european BS is more strict, it was under 0,5 W, maybe even less lately.

As for the PFC, I think it still states treshold of harmonics. Ofc crap without any correction will have very bad harmonics just as natural fact, tried once to report such to some burreau responsible for checking products here, mister "engineer" there was only talking BS how my consumer grade equipment cannot measure anything of that and some other stuff. Not sure if he was just lazy or bribed, but one thing went out of him when I told him I don't see any PFC in there - those harmonics, no norm says anything about PFC explicitly. I looked into that subject afterwards and it was true.

They may have changed it though but since the article you posted is somewhat old, I think author just got it wrong like 95 % of other such writers who never actually bumped into the norm itself.

[TELVM]

From the UK:

Maplin 001.jpg bx87.jpg 1h0c.jpg w49c.jpg jbbt.jpg bdpv.jpg

[Pentium4]

That looks like a rebranded Coolmax unit, or at least the same OEM. Maybe 200W with the ripple in spec. What brand is the fan that it uses?

[PeteS in CA]

Other than the 300W and +12V current ratings and the rather empty looking secondary side, that isn't too awful. It even has EMI filtering with what looks like agency-rated parts. 200W seems pretty realistic.

[TELVM]

Sorry P4, don't have the PSU around and can't check the fan.

Yep the input filtering is passable but the output is a crime, not a PI coil on sight and just one small 1000uF cap on +12V, that's guaranteed boatloads of ripple.

[370forlife]

Originally Posted by Pentium4

That looks like a rebranded Coolmax unit, or at least the same OEM. Maybe 200W with the ripple in spec. What brand is the fan that it uses?

Looks like a GoldenTiger. Most Coolmax's are Longyi, they are both very similar, usually only told apart by the part number. GoldenTiger = GT, Longyi = LY.

[momaka]

Originally Posted by PeteS in CA

200W seems pretty realistic.

Arguable.
Those heat sinks are paper-thin. I'd rate it around 100 Watts max continuous if you want to keep parts from overheating in there. With those output filters like that, the ripple will be out of spec on at least one rail even at over 50W.
Realistically, I would rate it.... trash

It uses some interesting red glue, though. Beats older Bestecs in that regard.

By the way, isn't this very similar to the PSU that just died in mariushm's father's PC? post $864 above

[c_hegge]

Golden Tiger like to use that red glue. I recon it could probably do 250W in spec, or maybe a little less if the fan stays real quiet. Even with those thin heat sinks, I've successfully pulled that from a few PSUs before. I've even manged 300W from one or two. It actually looks similar to a Bliss "350W" which will be in the 2013 el cheapo roundup. That one didn't blow up until I asked for 350W.

[momaka]

Originally Posted by c_hegge

Even with those thin heat sinks, I've successfully pulled that from a few PSUs before. I've even manged 300W from one or two.

But for how long is the question?
A few seconds to a few minutes isn't satisactory for me. If you ran it on the load tester for a day, then I say that's acceptable.
Also, how much heat is the PSU kicking out at that power level and how hot are the heat sinks? I bet you the output caps will bake quickly if the PSU is ran at that much power over an extended period of time. And I'm almost certain you can't get even a 100W out of the PSU without one of the rails getting out-of-spec ripple.