Found a brand new Comp USA PSU

[Pentium4]

Re: Found a brand new Comp USA PSU

True, but with only an ultra fast on the 12V it may hit the 70's with a 5V based system.

[Behemot]

Re: Found a brand new Comp USA PSU

OK show me 300 W for an hour…

[Pentium4]

Re: Found a brand new Comp USA PSU

I don't think it can do 300W. I'm just guessing man

[ben7]

Re: Found a brand new Comp USA PSU

I don't think it can do 300W. I'm just guessing man

Fine.

300mW then?

[Behemot]

Re: Found a brand new Comp USA PSU

What ripple would it have with such output?

[ben7]

Re: Found a brand new Comp USA PSU

What ripple would it have with such output?

300kV

[tom66]

Re: Found a brand new Comp USA PSU

5VSB transistor usually dissipates no more than 1W with 5V, 3A load. Two-transistor 5VSB is actually pretty efficient, around 75% to 80% (which is good for a small, low cost flyback.) And it's possible to make it without the killer cap, but you may as well use a controller IC because all of the tweaks will likely cost more than a cheap TNY-type switcher.

[kc8adu]

Re: Found a brand new Comp USA PSU

OK show me 300 W for an hour…

[Pentium4]

Re: Found a brand new Comp USA PSU

So it's pretty safe to say that if a PSU is using a 2 transistor circuit then it is a very old PCB design?

[370forlife]

Re: Found a brand new Comp USA PSU

Yes, This design looks particularly old. The 5vsb transistor not being attached to the heatsink is a good indicator that this was originally an AT design that was adapted to ATX.

[Pentium4]

Re: Found a brand new Comp USA PSU

Wow, that's pretty lousy. But I guess tons of manufacturers do the same type of thing when they just beef up the 12V on old half bridge ATX 1.3 designs

[370forlife]

Re: Found a brand new Comp USA PSU

Its more obvious on the designs where they use a transistor to generate the 3.3v rail from the 5v rail.

[Behemot]

Re: Found a brand new Comp USA PSU

5VSB transistor usually dissipates no more than 1W with 5V, 3A load. Two-transistor 5VSB is actually pretty efficient, around 75% to 80% (which is good for a small, low cost flyback.) And it's possible to make it without the killer cap, but you may as well use a controller IC because all of the tweaks will likely cost more than a cheap TNY-type switcher.

Do you have any real numbers or you are just making it out? I have never achieved even 80 % with 80+ Platinum Power Supplies with the best PWM combo chips, best result with Fractal Design Newton R3 600 W was 73,9 % on overload.

Two transistor usualy barely achieves 65 %. It burns insane power, mostly on the rectifying diode which, if not enough sized and/or superfast used, is sometimes capable of baking itself and everything around. They don't usually even count with 60+ % long-term output on these things so if that happenes, it won't survive that for long.

[Pentium4]

Re: Found a brand new Comp USA PSU

I believe on one of JG's reviews, an el cheapo he reviewed was something crazy like 59% efficiency in stand by with a 2 transistor design, maybe even lower

Edit: Found it, look at that, it was 41% at some point! http://www.jonnyguru.com/modules.php...tory3&reid=324

[tom66]

Re: Found a brand new Comp USA PSU

Do you have any real numbers or you are just making it out? I have never achieved even 80 % with 80+ Platinum Power Supplies with the best PWM combo chips, best result with Fractal Design Newton R3 600 W was 73,9 % on overload.

Two transistor usualy barely achieves 65 %. It burns insane power, mostly on the rectifying diode which, if not enough sized and/or superfast used, is sometimes capable of baking itself and everything around. They don't usually even count with 60+ % long-term output on these things so if that happenes, it won't survive that for long.

I'm talking about the 5VSB, not the main switcher; for the 5VSB 80% efficiency is possible due to the single switch, lower losses and small transformer. Bigger output powers do not lend themselves to flyback designs, you may be confusing it with a two-transistor NPN bipolar design (half bridge I think?) which does have poor efficiency.

You have to remember with those tests involving the 5VSB in an integrated design is the 5VSB transformer often powers the TL494 or similar chip on a separate rail plus other control logic. As I said it's possible to make them efficient, but more costly than a good TNY- single chip design.

[Behemot]

Re: Found a brand new Comp USA PSU

I am talking about +5 V SB, my own load tests. I have never seen 80 % efficiency on that. Remember that you have the whole primary side to feed even if just the +5 V SB rail is operating.

Will have another 80 PLUS Platinum PSU in next days reviewed and later some 80 PLUS Gold ones in next weeks (Fortron Aurum should arrive). Anyone wanna bet on crossing the 80% barrier?

[tom66]

Re: Found a brand new Comp USA PSU

What do you mean whole primary side? The only thing active would be bleed resistors and small leakage currents.

[Behemot]

Re: Found a brand new Comp USA PSU

Input filtering is not active? Rectifier, PFC and capacitor is not active? Only inactive thing in there are switching transistors, everything else works. These days almost only Champion Micro's PFC+PWM combo chips are used and PFC part of them is active, just the PWM not.

[tom66]

Re: Found a brand new Comp USA PSU

Rectifier draws only power as required by standby circuit and losses are very low at the 10's of mA drawn by the standby circuit. The X capacitor draws reactive power, which is not included in real power calculations. The bulk capacitor has negligible leakage current. 99% goes to standby power circuit.

[Behemot]

Re: Found a brand new Comp USA PSU

Sure but still there are loses.

There is also possibility of very inaccurate measurement of drawed power because of low power factor (usually <0,1), but much more ppl than me have the same results so statistically…