Source of Problem

Darren

Moderator
Staff member
Actually Denther none of that is true.

Firstly, efficiency doesn't work that way. If a computer requires 400W from a PSU that can deliver a 400W (33A DC) - a total of 480W (2A AC) will be drawn from the wall. Not the other way around ;)

I'm not arguing with you at all but that doesn't really make sense the way you said it. You're saying a 400 watt unit will draw more than that from the wall (480 for example) and will supply 400 of that to the computer?

If that's true then an inefficient PSU would still supply 400 watts even if it used say 550 watts from the wall.
 

Darren

Moderator
Staff member
Alright that reaffirms my point. Then what does efficiency matter if you don't care about how much power is coming out of the wall. It may be inefficient, but you're still getting the same amount of computer applied to the computer (which is what you want).
 

Darren

Moderator
Staff member
You are missing one thing. Energy can not be destroyed, only converted. What does not make it to DC is converted to heat, which is bad for the system and lowers efficiency as it builds up. The less efficient it is the hotter it is and thus even more efficiency is lost.

Yeah I hadn't thought of that. But that's still not my point, does a 400 watt PSU, regardless of "efficiency", provide the same amount of power to the computer compared to another 400 watt PSU. Even if it does produce more heat and use more power from the wall.
 

Darren

Moderator
Staff member
Well that would completely depend on the distribution across the rails. Assuming 2 supplies have say 33 amps on the 12V rail (which will never happen, as that is 398-399 watts right there, then yes. They will both make th esame power. But one that makes say 15 amps on the 5V and 20 on the 12V and say 26 on the 3.3V, then it would never be able to keep up these days. (Note these numbers are just for demonstration).

You need to look at amps and efficiency to get the PSU you need. Mainly these days it is power on the 12V rail, but older systems the CPU pulled from the 5V rail (Pre ATX standard).

Okay thanks. Makes sense. PSU's are probably the most confusing part to breakdown so just recommending a good brand works well enough. :D
 

mr.doom

Member
You are missing one thing. Energy can not be destroyed, only converted. What does not make it to DC is converted to heat, which is bad for the system and lowers efficiency as it builds up. The less efficient it is the hotter it is and thus even more efficiency is lost.

I was about to say that. :p Anyway, points stands across the board - get a new PSU, as soon as it is humanly possible. I keep the fingers crossed that it did not damage your GPU in the process, as this type of damage will not be covered by the warranty (after all they do specify their "safe" minimum for the PSU on the box).
 

Okedokey

Well-Known Member
I'm not arguing with you at all but that doesn't really make sense the way you said it. You're saying a 400 watt unit will draw more than that from the wall (480 for example) and will supply 400 of that to the computer?

If that's true then an inefficient PSU would still supply 400 watts even if it used say 550 watts from the wall.



Exactly.

Ill just add because...

Okay thanks. Makes sense. PSU's are probably the most confusing part to breakdown so just recommending a good brand works well enough. :D

PSU design efficiency costs. 50oC rated capacitors will be good, 105oC caps are military grade. The OPs current caps are rated at 20oC. 5 degrees below standard (25oC rating) and well below operating temp. So instantly de-rate the PSU 20W total below advertised values. Then temperature derate it beyond the standard 25oC correction we made - the inside of PCs get to over 45oC in summer when gaming. That's at least 20oC higher and at 4watt/degree C above 25oC, you have lost another 80W. So yes, as I guessed before, 100W.

You're never going to make your money investing in efficiency because its a useless number unless you care about electricity costs which are tiny compared to the 20 bucks you spent extra to get it.

Why? It is only indicative of a good quality PSU, but an indicator, not necessarily essential to providing 12v stable.

Just means better electronic tolerances (accuracy) and they never die.

This means, as you worked out, that to deliver 400W DC it needs to extract 480W AC which is caused by hysteresis and heat loss as well as basic resistance and inductance (thus active PFC).

Cheap PSUs don't have active pfc (power factor control), nice caps or good voltage regulation circuitry.

Even Active PFC is in the same category. Nice to have, purely useless unless you pay reactive load chargers on your bill. (almost no residents do)

The job of a PSU is to convert AC power (sine wave) to DC power (straight line).

Most importantly for most electronics, 12V DC. All the other rails (e.g. 3V) are simply achieved by voltage dividers (resistors etc)

PSUs job is to maintain 12V DC as accurately as possible REGARDLESS of load.

This variable LOAD, is measured in amps and is the name of the unit for current (A) - a resultant of watts (W) and voltage/the fixed reference point (V) joining forces.

Amps (A) is what you're therefore interested in because 12V amperage is the number that matters.

Keep stable (quiet) 12V DC rail and you have a good PSU.

Also, forget the discussion on rails. Anytime you think about that you're either talking about:

  1. a very early Corsair 1000W (dual coil) or;
  2. you don't understand psus.
99% of domestic PSUs are technically single rail (1 transformer coil), and voltage dividers do the work, but they certainly don't improve matters (thermodynamic). Its simply the cheaper designed psus must have multiple voltage dividing circuitry to keep within the 240AC safety limit (ATX).

Better components and design allow better quality PSUs to deliver 24/7 12V stable amperage 20% above the computer's required max amperage at 80%+ efficiency at 50oC and on a single rail.

That's what you're after.

This discussion is actually why we should all move to high voltage ultra frequency DC decentralised power distribution
 
Last edited:
Top