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

[Pentium4]

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

This thing has potential. Obviously horribly overrated from the label but not bad after some upgrades...it has two 2SC2625 transistors, 30A schottky for 5V, 20A fast recovery for 3.3V and 20A schottky for 12V. Not a bad transformer. Filtering is terrible but that's cause of all those tiny Jenpo caps, primary's are 470uF Teapo, definitely in need of being increased. On the secondary side it has room for big sized schottky's, thinking of putting a 30A part for the 3.3V and 12V

[Behemot]

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

Half bridge with some iteration of TL494? You seen my thread I guess. I was able to get 300 W from that with more or less being in spec, after all modifications possible, only +5 V SB left with out-of-spec ripple. After that, efficiency goes down the toilet, ripple goes up the chimney.

I do not suggest shottky for +3,3 V, these things seem to have wiring beefy enough for loses on superfast. If you swap it, you get too much of a boost on this rail. I suggest two superfasts in parallel at most. Than maybe two 20-25A shottkies in parallel on +5 V.

[Wester547]

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

On the subject of the 20 vs. 18AWG cables of the Thermaltake - not only are the cables 20AWG but they're sleeved (further reducing what they can handle, I believe) and they're not all connected to the main (20/24-pin) connector but to smaller connectors, which may further hurt it, especially if the cables aren't safety-approved...

[momaka]

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

^ I see some toasty resistors in that Velocity PSU. Check the 5VSB output caps and the 5VSB critical cap. Sometimes those can cause the auxiliary voltage on the secondary side of the 5VSB traffo to go way higher that it's supposed to. I've seen that take out the PWM controller a few times. If that's not it, then trace what those resistors go to. Something is not right with the amount of power they are dissipating.

On the subject of the 20 vs. 18AWG cables of the Thermaltake - not only are the cables 20AWG but they're sleeved (further reducing what they can handle, I believe)

I don't see how the sleeves hurt the capacity of the wires. The wires may get a little less cooling, but I doubt that will make any difference at all. Or am I missing something here?

and they're not all connected to the main (20/24-pin) connector but to smaller connectors, which may further hurt it, especially if the cables aren't safety-approved...

Again, I don't see how. It all really depends on which connectors they used the 20 AWG wires. For example, drives don't use much power so 20 AWG for those won't make a difference. On the P4/CPU or PCI-E power connector, though, it might since there can be a lot of current going through these, depending on what hardware is attached to them.

[Behemot]

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

Again, I don't see how. It all really depends on which connectors they used the 20 AWG wires. For example, drives don't use much power so 20 AWG for those won't make a difference. On the P4/CPU or PCI-E power connector, though, it might since there can be a lot of current going through these, depending on what hardware is attached to them.

It can very much matter on older AGP graphics, with peripheral molex or even FDD molex to power them. If they suck pwoer from +5 V, they can take huge current.

[momaka]

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

^ I think the Radeon 9700 and 9800 series were one of the last cards to use power from the 3.3V and 5V line for GPU, with the 9700 being the last to use the FDD connector. According to Xbit labs, the 9800 pulls about 4 A from the 5V line on the connector so I'm guessing the 9700 pulls more or less the same as well. Yeah, 4A through that kind of wire is considerable, but still far from burning it. However, let's be fair: how many people are using such old GPUs anyways (besides myself that is.... I have 4x 9700s ).

[Behemot]

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

Would guess them more…

Have 9700 Pro in my parents comp

[mariushm]

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

Even with PCI Express it's not that much current.... 75w per 6 pin connector, so we're talking of about 6 amps shared between three wires.

The PCI Express 8 pin connector is double that.. so 12 amps shared between those 3 wires

[Behemot]

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

Yeah but you got 12 V. These older cards use 5 V. I would guess them 50+ W, thats over 10 A, too.

[momaka]

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

These older cards use 5 V. I would guess them 50+ W, thats over 10 A, too.

Hardly at all. According to Xbits labs, the Radeon 9800 Pro uses just a tad over 47W at full load. That 47W is split mostly between 3.3V and 5V rail, with the AGP connector also providing a significant amount of that power.

You can easily guesstimate a video card's power consumption by the size of its cooler. Obviously if it's got a tiny cooler with a tiny fan, it's not going to need a "400W" power supply like the manufacturers always like to recommend.
Heck, most computers don't use even 200W at full load (even the mid-range gaming ones). Cheap PSU manufacturers know that and that's how they get away with selling such crappy PSUs.

[Behemot]

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

I got Silencer 5 on mine, that's hardly small cooler…

As for todays graphics, many times it has PCIe connector, it uses small or no power from PCIe. Why shouldn't it be the same with AGP…you get better efficiency with 5 V and there's not much of that in AGP.

[cheese007]

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

^ I think the Radeon 9700 and 9800 series were one of the last cards to use power from the 3.3V and 5V line for GPU, with the 9700 being the last to use the FDD connector. .

I have an x1650 Pro that takes in power from the FDD connector, but I don't remember seeing other cards of the same gen using it.

[Pentium4]

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

I just noticed that the only non Jenpo caps are JPCE-TUR on the 5VSB, still going to replace them all. I have an x1650 Pro AGP and it doesn't use a lot of power but that son of a bitch gets HOT!

[Behemot]

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

Those are CEC Holding I think, not THAT bad being taiwaneese I think.

[Wester547]

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

One last note on the Thermaltake (or sort of regarding it, at least)...

It's a single transistor forward design save by the +12V rectifier and that whole circuit of the PSU. It says so in the review. So I think, while forward topology is more efficient than bridge (that's why it can get away with only using one main switcher instead of two), that the primary capacitors, rectifying bridge, transformer, and rest of the primary probably aren't any more efficient (unless cooling is considerably improved) in either topology. Does this, however, dispel the "divide the microfarad rating by two to get the wattage" rule in terms of watts and total output capacitance (for non-APFC PSUs)? Some Macron MPT-301 PSUs, for an example, only have two 470uF/200V primary capacitors and yet they do 300W just fine in a half-bridge configuration with a temperature-controlled fan, and with very good voltage regulation.

[Behemot]

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

If you remember that Eurocase of mine…editor in chief wanted me to compare it to un-modified version. Well, we can finaly call this crap a crap, cause we have it confimred, right? That's the freedom of speech in practice.

So I have bought brand new one. It is so bad that I am putting it right here, it's not worth PSU built "quality" pictorial. Five good things in there are: almost full filtration (most of it on separate board soldered to mains connector; only thermistor missing), insulation foil under PCB, ViPER 22 (but they had to use something like this to comply that EUs <1 W standby stuff), all schottky rectifiers and full output filtering.

The rest is pure horror. But I am quite surprised that the tiny 2A bridge rectifier on input actually survived that. On the other hand, at the very moment it would definitelly start to draw over 2 A, the voltages just went down the tube. I am somewhat suspicious whether they didn't make it on purpose so it won't blow. If not, than it's the tiny transformers fault.

All in all, look at the results, they speak for themselves. I suspect the shottkies for the huge voltages on both +12 and +3,3 V, I think they made the transformer wirings for ultrafasts once gain, but ended with putting in schottkies and here is the result.

I haven't even be able to load it fully, as the voltages dropped also the current through resistors dropped so I ended in like 17 A for both +5 and +3,3 V. So I wanted to add load on +12 V (where it also dropped to aprox. 12 A) when it suddenly shut down. It was still working afterwards but started to shut down around just over 150 W, for that reason I don't have results for power draw, temperatures and ripple. But considering the previous results, is it even important? Ripple was out-of-spec from the very begining anyway.

[ben7]

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

If you remember that Eurocase of mine…editor in chief wanted me to compare it to un-modified version. Well, we can finaly call this crap a crap, cause we have it confimred, right? That's the freedom of speech in practice.

So I have bought brand new one. It is so bad that I am putting it right here, it's not worth PSU built "quality" pictorial. Five good things in there are: almost full filtration (most of it on separate board soldered to mains connector; only thermistor missing), insulation foil under PCB, ViPER 22 (but they had to use something like this to comply that EUs <1 W standby stuff), all schottky rectifiers and full output filtering.

The rest is pure horror. But I am quite surprised that the tiny 2A bridge rectifier on input actually survived that. On the other hand, at the very moment it would definitelly start to draw over 2 A, the voltages just went down the tube. I am somewhat suspicious whether they didn't make it on purpose so it won't blow. If not, than it's the tiny transformers fault.

All in all, look at the results, they speak for themselves. I suspect the shottkies for the huge voltages on both +12 and +3,3 V, I think they made the transformer wirings for ultrafasts once gain, but ended with putting in schottkies and here is the result.

I haven't even be able to load it fully, as the voltages dropped also the current through resistors dropped so I ended in like 17 A for both +5 and +3,3 V. So I wanted to add load on +12 V (where it also dropped to aprox. 12 A) when it suddenly shut down. It was still working afterwards but started to shut down around just over 150 W, for that reason I don't have results for power draw, temperatures and ripple. But considering the previous results, is it even important? Ripple was out-of-spec from the very begining anyway.

Whats the capacitance of the primaries? They say 470uF, but I'm guessing they are like 220 or 100uF!

[Behemot]

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

I don't possess meter, sorry. But I think they are big enough for that capacitance.

Look at Fortrons, they got input caps U woudl never say being 820 uF, but I don't think FSP would use fakes. Seems evolution goes forward, though slowly

[Behemot]

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

I am writing the rest of the review ATM, and one thing I have run into is, what the hell is driving this thing's primary?? The PS222S is secondary-monitoring only. There is no big PWM, I am sure. I can only see some miniature 3pin devices in TO-92 or extended packages. There are also not any bigger capacitor typical for two-transistor forward designs.

And if that's not it (any of the small 3pins), than there is only thing which strikes me - some kind of self-resonant circuitry. Perhaps that small coil with cap between primaries and heatsink?

[ben7]

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

I am writing the rest of the review ATM, and one thing I have run into is, what the hell is driving this thing's primary?? The PS222S is secondary-monitoring only. There is no big PWM, I am sure. I can only see some miniature 3pin devices in TO-92 or extended packages. And if that's not it, than only thing which strikes me is some kind of self-resonant circuitry. There are also not any bigger capacitor typical for two-transistor forward designs.

Looks like there is no feedback transformer, but there IS 2 extra opto-isolators! I bet they drive the switching transistors. There is a PWM controller, that DIP case style chip.