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

[ben7]

Originally Posted by mariushm

later edit 2 : TELVM if you contribute to that site, why the hell does it not allow google translate to translate the forum?

It translated it for me just fine ...

Looks like the typical cheap-ish PSU. Those thin heatsinks with holes on the top seem fairly common. And, typical 3A diode bridge rectifier... not bad for low wattage but for higher wattage you want to go with a better bridge rectifier.

[TELVM]

I love the 'prize' that is 'awarded' to gutless wonders in that site:


"This product is the bomb!"

[rogfanther]

Considering the "special effects" that occur when one powers up a computer with one of these things, it is very adequate.

[c_hegge]

The funny part is that if you type "Produto Bomba" into google translate, you get "Product Pump"

[TELVM]

^ Yeah in portuguese and spanish the same word bomba is used both for bomb and pump.

This leads to the funny situation where firemen are called bomberos ('bombers' ).

[TELVM]

A fearsome squad of bomberas :

[momaka]

Originally Posted by TELVM

"This product is the bomb!"

... only if it's assembled by a group of terrorists.

Originally Posted by TELVM

This leads to the funny situation where firemen are called bomberos ('bombers' ).

[rogfanther]

Also causes some problems in airport security, when people don´t know the correct name for something ( like asthma pumps ) and call it "bombs" ...

[TELVM]

Another even worse 'TOOQ' gutless wonder :

Behold its mighty rectifier bridge :

(Should I 'upgrade' to a four-diode treatment?)

Somebody messed up and put 13009s instead of the traditional 13007s:

Armoured extra thick heatsinks:

Exquisite output filtering courtesy of state-of-the-art, stealth technology PI coils:

And not a single SATA connector !

.

[goodpsusearch]

270w max

[Pentium4]

Seen a Worldwide PSU before, but it was a rebranded CWT ISO series PSU. This ones UL number leads to "MITACHI TAIWAN CO LTD" and the fan pretty much confirms this. For the 300W label it looks alright.

First time I've ever seen Su'scon on the voltage doubler of a PSU, 560µF isn't bad. 4A bridge, switchers are your standard 12A 13009 transistors, and 2 transistor 5VSB. The input filter is decent minus the fact that those aren't safety approved Y caps and the spots for the MOV's were not filled.

35 transformer, decent sized 5VSB transformer.

Here's what I don't understand about this PSU. Where is the toroid for the 3.3V regulation, and the 3.3V output?! Also, instead of your average 16-20A ultra fast, there's a 40A MOSFET on the 3.3V rail. Why would they use this method? 16A ultra fast for 12V and 30A schottky for 5V. Both the 12V and 5V have room for another rectifier in parallel. All of the caps minus the bulk caps are JEE, and surprisingly only one of them on the 5VSB is bulging. Although I wouldn't be surprised if some of the other ones had high ESR. According to the PWM chip, this thing was manufactured sometime in late 2006.

Also really glad that this things fan doesn't run at 100% all the time!

[c_hegge]

The 3.3V rail uses linear DC-DC conversion from the 5V rail, which doesn't require a toroid coil. It's used because it's cheaper than using a separate rectification circuit for it.

[Pentium4]

Originally Posted by c_hegge

The 3.3V rail uses linear DC-DC conversion from the 5V rail, which doesn't require a toroid coil. It's used because it's cheaper than using a separate rectification circuit for it.

Cool, thanks! Does it yield any benefits such has higher efficiency, or does it just reduce the ripple on the 3.3V rail?

[Th3_uN1Qu3]

It's just cheap. Linear regulation is the cleanest and simplest, but also the most inefficient form of regulation, as the difference between input voltage and output voltage is dropped across the regulator element and wasted as heat. Basically the regulating element behaves like an automated variable resistor.

[Behemot]

It's 34 watts at 20 A? OMG, what is efficiency of this thing, 40 %?

[Pentium4]

Originally Posted by Th3_uN1Qu3

It's just cheap. Linear regulation is the cleanest and simplest, but also the most inefficient form of regulation, as the difference between input voltage and output voltage is dropped across the regulator element and wasted as heat. Basically the regulating element behaves like an automated variable resistor.

Interesting. Does this create more heat on the only toroid since it's generated from the 5V? So it probably doesn't even need a PI coil for that rail to stay low on ripple. Luckily the 3.3V rail isn't stressed much at all in modern systems. I have this thing powering a Pentium D (90nm powerhog) with DDR RAM and a GT220, surprisingly even with that Pentium D at full load the 12V rail is staying steady at 11.97V. The fan did kick up quite a bit, but still pretty quiet.

Originally Posted by Behemot

It's 34 watts at 20 A? OMG, what is efficiency of this thing, 40 %?

Are you talking about the 3.3V rectifier?

[Behemot]

Whole supply, in worst case scenario it burns 34 watts just at +3,3 V, don't rather think how much it burns on other rails

[Pentium4]

Originally Posted by Behemot

Whole supply, in worst case scenario it burns 34 watts just at +3,3 V, don't rather think how much it burns on other rails

Haha When it was running that Pentium D system it was pulling about 150W and the secondary heatsink + toroid were 35C, 36C respectively. Doesn't seem too bad. This thing did not take a 30A schottky rectifier on the 12V, even one with 100V reverse voltage. Can I put a 12A ultra fast in parallel with the original 16A ultra fast? They're both MOSPEC with the same exact specs except the peak amperage. If not, I think two 12A ultra fasts in parallel would be better than a single 16A especially for efficiency, and combined probably good for ~19A

[momaka]

Originally Posted by Th3_uN1Qu3

It's just cheap. Linear regulation is the cleanest and simplest, but also the most inefficient form of regulation, as the difference between input voltage and output voltage is dropped across the regulator element and wasted as heat.

Indeed.
But if you don't have much of a load on the 3.3V rail (which most systems nowadays don't), then this is nothing to worry about. In fact, I prefer this linear regulation method as you have the choice to use any caps you want on the output of the 3.3V rail and also the 3.3V rail outputs very very clean power.

As far as I know, only old video cards like the Radeon 9500/9700, and 9800 used the 3.3V rail more heavily (like 2 to 5 Amps).

[Behemot]

Originally Posted by Pentium4

Can I put a 12A ultra fast in parallel with the original 16A ultra fast? They're both MOSPEC with the same exact specs except the peak amperage. If not, I think two 12A ultra fasts in parallel would be better than a single 16A especially for efficiency, and combined probably good for ~19A

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