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

[momaka]

^ Exactly!

Too many people online simply just shout numbers without any scientific evidence whatsoever.

[ChaosLegionnaire]

Originally Posted by momaka

Maybe whoever was in control of the conveyor speed for the assembly line accidentally bumped the knob to "high speed" and then the workers tried to compensate by installing everything quickly?

i think u just described exactly the kind of thing that goes on in a chinese sweatshop!

Originally Posted by RJARRRPCGP

A Core 2 Quad, OTOH, exponentially goes up in watts when OC'ed, 250 W easily for the processor alone at 3.3!

Easily like an OC'ed FX, if not more! In fact, I wouldn't be surprised if a 65 nm Quad makes my FX look like a green freak! LOL!

The OC expert folks say 250 W+ easily for CPU alone with 1.4 V Vcore at only 3.3. So, that would easily mean at least a 350 W PSU, if not a 400 W PSU. (And that's with a video card that don't use a power plug, such as a GeForce GT 640 and GeForce GTX 750)

i think their measurement might have been at the wall with a wattmeter which is a meaurement from the ac side and does not take into account the inefficiency of the psu.

a common mistake is them thinking e.g. a 500w psu means 500w max from the ac side which is wrong... its supposed to be 500w max output from the dc side not the input ac side.

their calculation of the cpu's power consumption is also prone to a degree of inaccuracy from trying to reverse engineer how much power the other components (mobo, ram, hdd, gpu) are using and they might have missed out a few other things in their calculation.

i still think those cpu wattage calcs (where u input the stock clocks, voltages and wattage) are a good estimation of cpu power consumption assuming u dont go crazy with the overclocks and overvoltages till u cause the cpu's internal capacitance to totally screw up the tdp calculation!

[momaka]

Originally Posted by ChaosLegionnaire

i think u just described exactly the kind of thing that goes on in a chinese sweatshop!

Well, I was thinking either that... or someone was playing that Macarena song on the radio / over the PA while the workers were on the assembly line.
Explains why the components appeared all "dancy" on the board.

Originally Posted by ChaosLegionnaire

i still think those cpu wattage calcs (where u input the stock clocks, voltages and wattage) are a good estimation of cpu power consumption assuming u dont go crazy with the overclocks and overvoltages till u cause the cpu's internal capacitance to totally screw up the tdp calculation!

I don't know about that. In my experience, those calculators always grossly over-estimate the power needed.
Personally, I prefer to go to cpu-world.com and look up the TDP spec of the CPU I am using, then do a bunch more calcualtions (such as, assume only 90% efficiency for CPU VRM), add 5-15 Watts for RAM, some more for HDD, and if there is to be a add-on graphics card, I try to look up the TDP of that too. That's it. If the CPU is to be over-clocked, I simply look up the CPU spec that is from the same family and has the same core but higher clock, and use the TDP spec from that. Easy! And a lot more accurate.

[TELVM]

Originally Posted by RJARRRPCGP

... The OC expert folks say 250 W+ easily for CPU alone with 1.4 V Vcore at only 3.3 ...

Here's to your "experts":

. . .

http://www.xbitlabs.com/articles/cpu...locking_8.html

[Behemot]

Lol, I just noticed the CPU glued to its heatsink by rock-hard thermal compound, being pulled from locked socket

Hate when that happens…rotating the heatsink while CPU is still in socket often helps.

[goodpsusearch]

Premier DR-8460BTX

SB3045ST @ 5V

SB1060CT @ 12V

S10C45C @ 3.3V

2x 13007 transistors

33 sized transformer

2x 330uF 200V ANODIA caps

2N65 for 5vsb

4 diode treatment for AC rectification

Incomplete EMI input filter

Tiny heatsinks

1x1000uF for every important output

Not a single coil for ripple filtering

Tiny toroid coils

Tiny parts everywhere

But it works and none of the caps are bulging... The fan is moving at full speed. This thing was buried in dust and tobacco residue...

[momaka]

Originally Posted by goodpsusearch

Premier DR-8460BTX
SB3045ST @ 5V
SB1060CT @ 12V
S10C45C @ 3.3V

... And yet the label says the "dual" 12V rail can handle 14 + 16 Amps . Where the fuck do they see that at with a 10 Amp rectifier?

At least they put a nice big label with bright red letters. Probably to warn users: "Warning! This PSU is dangerous!".

Originally Posted by goodpsusearch

2x 13007 transistors
33 sized transformer
2x 330uF 200V ANODIA caps
2N65 for 5vsb
4 diode treatment for AC rectification
Incomplete EMI input filter
Tiny heatsinks

Typical Deer/L&C.
Don't forget those big, low-resistance dummy load resistors on the output. I see the one for the 5V rail is 20 Ohms. Should keep the output caps nice and toasty.

Originally Posted by goodpsusearch

1x1000uF for every important output
Not a single coil for ripple filtering
Tiny toroid coils
Tiny parts everywhere

Woah DUDE, chill. Let's just go surfing! No, DUDE, not surfing the internet. I meant surfing those huge AWESOME ripple tidal waves from the PSU output. Yeah, now wouldn't that be TOTALLY RAD DUDE?
/goofy_surfer_voice

Originally Posted by goodpsusearch

But it works...

Typical Deer/L&C. x2

I guess they are dependable PSUs - that is, you can always depend on them to kill your hardware.

The case looks nice, though. Could be used to house the guts from another PSU if needed.

[RJARRRPCGP]

Originally Posted by TELVM

Here's to your "experts":

. . .

http://www.xbitlabs.com/articles/cpu...locking_8.html

That was for 65nm quads. The 45 nms are much better.

[Stefan Payne]

the Q9505 is a 45nm CPU...

And you can't do anything about physics...
If you increase the voltage, even a little bit, the powerconsumption skyrockets...

[Viperel]

And here is a Micro ATX (SFX) 1.3 which came with a Deluxe case so im guessing the PSU is also made by them because there is no UL code.

All caps seem to be Chengx but at least they are rated 105degrees.
No heatshrink anywhere, a probably underpowered 10A schottky, the fuse is soldered directly to the board, but at least there seems to be a PWM IC CG8010DX16 supporting UVP/OVP

[PeteS in CA]

450W? For how many milliseconds?

[Dan81]

Originally Posted by PeteS in CA

450W? For how many milliseconds?

Actually,that might be another 0W PSU,as there are absolutely no PI coils (or ANY SPACE printed on the PCB for them,let alone holes),thus it won't do a watt in spec. Good only for smashing.

[Viperel]

Originally Posted by PeteS in CA

450W? For how many milliseconds?

I tried for the fun of it to hook up an 80W/12V bulb and the +12v rail showed 10.1V

[Behemot]

It's +5V heavy, you should load the rail as well. At least couple amps.

[everell]

From the looks of those pictures, looks like the black Line wire is going to Neutral and the white neutral wire is going to Line. Be careful with that one and don't get electrocuted - or start a fire!

[Viperel]

Originally Posted by Behemot

It's +5V heavy, you should load the rail as well. At least couple amps.

Well the same 80W bulb takes the 5V line down to 4.8v which is actually in the +/- 5% specs but i would not trust it hooked to a PC.

[momaka]

Originally Posted by Viperel

Well the same 80W bulb takes the 5V line down to 4.8v which is actually in the +/- 5% specs but i would not trust it hooked to a PC.

Ay, that PSU is no good then - not for powering a computer anyways. Maybe okay for playing with LEDs or small motors that don't care about smooth voltage. After all, those caps on the output are indeed too small, as Dan81 pointed out.

Originally Posted by Dan81

Good only for smashing.

Well, not exactly. There are still good parts in that power supply. They are just not good enough to make a functional PC PSU. But if you take the "crappy" schottky rectifiers from this PSU and put them in say, a small 10-20 Watt power adapter that uses regular diodes for the rectification, you would get a lot less heat and much better efficiency. Or, if you're game, you can build your own custom SMPS for something like an amplifier, like our other fellow Romanian, Th3_uN1Qu3, did many ages ago :
http://www.badcaps.net/forum/showthread.php?t=10491

Originally Posted by everell

From the looks of those pictures, looks like the black Line wire is going to Neutral and the white neutral wire is going to Line. Be careful with that one and don't get electrocuted - or start a fire!

It actually doesn't matter that much where he lives. Many countries in Europe use the type-F Schuko plug. And with those, Live-Neutral polarity is ambiguous (i.e. L/N can be swapped, depending on which way you insert the plug).
https://en.wikipedia.org/wiki/Schuko

As long as you are aware of that (especially if working on live stuff), it's not too much of a problem. And of course, if designing something with a line switch, you should always use a double-pole switch so that both Live and Neutral are disconnected.

[TechGeek]

Originally Posted by momaka

ay, that psu is no good then - not for powering a computer anyways. Maybe okay for playing with leds or small motors that don't care about smooth voltage. After all, those caps on the output are indeed too small, as dan81 pointed out.


Well, not exactly. There are still good parts in that power supply. They are just not good enough to make a functional pc psu. But if you take the "crappy" schottky rectifiers from this psu and put them in say, a small 10-20 watt power adapter that uses regular diodes for the rectification, you would get a lot less heat and much better efficiency. Or, if you're game, you can build your own custom smps for something like an amplifier, like our other fellow romanian, th3_un1qu3, did many ages ago :
http://www.badcaps.net/forum/showthread.php?t=10491


it actually doesn't matter that much where he lives. Many countries in europe use the type-f schuko plug. And with those, live-neutral polarity is ambiguous (i.e. L/n can be swapped, depending on which way you insert the plug).
https://en.wikipedia.org/wiki/schuko

as long as you are aware of that (especially if working on live stuff), it's not too much of a problem. And of course, if designing something with a line switch, you should always use a double-pole switch so that both live and neutral are disconnected.

+10

[Behemot]

Well the point is to load both rails *at the same time*. Loading only one at a time is called crossload and group-designs don't like that. Group regulated units are even worse.

[PeteS in CA]

Originally Posted by Dan81

Actually,that might be another 0W PSU,as there are absolutely no PI coils (or ANY SPACE printed on the PCB for them,let alone holes),thus it won't do a watt in spec. Good only for smashing.

Speaking more seriously, I've worked in power electronics for over 35 years. Those semiconductors, heatsinks, and magnetics are probably good for 200W continuous. The output noise would probably have horrible noise spikes (due to the lack of what you called "Pi coils"), and the conducted noise would probably be horrible, due to the lack of input inductors.

But if loaded to 450W, its life would be minutes or even seconds. Those input rectifiers are probably not good for that much current; the MJE13007-clone transistors are not good for that much current (and their beta would probably be less than 5!); the heatsinks aren't adequate for that much heat dissipation, the transformer core might saturate, and the output rectifiers probably aren't good for that much current.