EVGA W1 500W... Oops!

[ashiekh]

I can tell you, from experience, that it's not.

Then what are the specs saying other than heat is more easily lost down the leads than through the case? the leads are metal, the case is plastic.

Seems to me one must be very careful before announcing "A Major Design Flaw?".

[Jon Gerow]

I don't know. Why does an OnSemi TL431 have a wider variance of output voltage depending on operating temperature, compared to the equivalent LiteOn part in the same circuit, despite the part operating well within specified operating temperatures? On paper, the behavior should be the same between the two. But they're not.

If we could design EVERYTHING based on what a part is supposed to do "on paper" everyone's job would be a lot easier. :-D

[ashiekh]

I agree, but dumping heat to the circuit board is quite acceptable. If there were room in the design, I'd bend the bridge diode over and bolt it to the board if only to make it more physically robust.

P.S. hot diodes are more efficient than cold ones ;-) although efficiency is probably not the aim for this power supply.

[Jon Gerow]

If there were room in the design

Here's the crux. EVGA has no engineers that know the difference.

[Jon Gerow]

If there were room in the design

Here's the crux. EVGA has no engineers that know the difference.

[ashiekh]

So this isn't just a standard platform that EVGA put their name on?

[Tator Tot]

It's probably a semi-standard design, but component selection might be the lowest possible on the AVL to meet the price point EVGA wants. Essentially EVGA may have cheaped out on components without knowing the detriment to the design.

[Juular]

It's probably a semi-standard design, but component selection might be the lowest possible on the AVL to meet the price point EVGA wants. Essentially EVGA may have cheaped out on components without knowing the detriment to the design.

Or they've cheaped out knowing that people will still buy their shit just because it has 'EVGA' on it and everyone and their grandma are running around yelling that EVGA's support is out of this world. Even with good support they'll have margins because even through they're selling 200-300W PSU as 500W, you wouldn't find this PSU in a system that really draws 500W on a persistent basis, well ... it's possible but you should be a complete idiot to put this in a such system so hopefully there aren't many such cases.

[Tator Tot]

Or they've cheaped out knowing that people will still buy their shit just because it has 'EVGA' on it and everyone and their grandma are running around yelling that EVGA's support is out of this world. Even with good support they'll have margins because even through they're selling 200-300W PSU as 500W, you wouldn't find this PSU in a system that really draws 500W on a persistent basis, well ... it's possible but you should be a complete idiot to put this in a such system so hopefully there aren't many such cases.

Given the lack of engineers and a bit of background knowledge in how these things go, the conversation most likely went like this:

EVGA - "We need a 500w unit that can be sold for $xx USD"
OEM - "We have this, here are our documents on the product, here's where you can trim cost."
EVGA - "Trim all the costs so we actually make some margin on this."
OEM - "Okay."

In the scenario there might be an Engineer at the OEM who really understands what they're selling and it's flaws but he's also most likely not a part of the sales team communicating with EVGA and EVGA doesn't have any Engineers for this product to do in house testing either. They're also not gonna catch any potential problems with a unit like this unless they start seeing massive amounts of returns (either from customers or retailers.)

It's generally wise to always assume ignorance before malice.

[crmaris]

In diodes as the temperature rises the voltage drops on them decrease so less energy gets wasted. That said, you need to be extra sure before you install a diode like this, even in a low to mid-capacity platform. In some other PSUs that the manufacturer wanted to achieve this, they installed two diodes to half the Amperage, bolted together (without a heatsink in between).

In general, you need to ask for a full thermal analysis in every stage of manufacturing, but such an analysis lasts long (it can run for 40h straight) and you have to trust the manufacturer's results, too. And you pay extra of course unless you ask kindly Moreover, you have to have people able to interpret the results! Because getting a thermal analysis without any knowledge on tolerances etc. is meaningless.

Since I doubt many of the results provided by factories, I will invest in a high-end Keysight DAQ to conduct them on my own in some selected samples.