4kW PSU in 20cm by 20cm by 8cm?

[Siwastaja]

Well, I am not a power engineer, but if you take a ballpark 80% efficiency

The one thing you definitely would NOT do, if tasked with design of new high power density PSU, is to accept such poor efficiency target.

Recently it's been understood that clever use of modern technology, and enough silicon, is actually cheaper than to spend money on aluminum heatsinks and fans. This is even just considering BOM. Energy savings for the customer are an added benefit.

Given that transportation costs have skyrocketed in recent years, especially last year or so, power density pays back there, too.

And to get that power density, you just can't arbitrarily increase cooling, you have to increase efficiency.

90% is really a modern-day decent target but for such high power density, better aim higher, and you STILL need some rather noisy fan cooling, as you realistically can't get the dissipation below 200-300W no matter how good you are or design/BOM budget you have.

[jonpaul]

Rebonjour, to the moderators (Simon) and  veterans EEVblog super contribs...


I am very curious...who  is our friend Faringdon?

The  questions and responses seem very odd for just a DIY hobbyist,

My advise is perhaps a waste of my limited time (very old guy here...)

(:-:)

Jon

[Siwastaja]

Older long-time username treez.

Seems to actually know something about switch mode converter design, and can do SPICE simulations. Has claimed to work professionally designing said supplies, usually in companies where management wants to pitch pennies in everything. Also does not like electrolytic capacitors, Chinese design, and loss of UK design jobs. Sometimes posts political rants as a result.

Asks for advice for every possible small design detail. Usually has some weird idea about what others think is "normal" way of doing things. Instead of using a simple Google search to show this is not the case, he spends days discussing the strawman idea here, obviously not getting much work done as a result.

Some can't stand him at all, some take him quite seriously. I think he's OK, most of the threads are technical and "to the point", even if a tad weird, but still good starting points for technical discussions. Too many threads? Just ignore most of them, and only look at one when you feel like you could get a small treez dose (recommended daily intake max 1 thread, with coffee). Some don't like the fact he's asking so much guidance for paid professional job. Others question if he's actually paid, or have ever been. Interesting character, nevertheless.

[Gyro]

Rebonjour, to the moderators (Simon) and  veterans EEVblog super contribs...


I am very curious...who  is our friend Faringdon?

The  questions and responses seem very odd for just a DIY hobbyist,

My advise is perhaps a waste of my limited time (very old guy here...)

(:-:)

Jon

Bonjour Jon,

As you have clearly noticed, there is some 'history'. The initial identity of treez was changed to 'ocset' (he spawned several other identities but these were locked - other than 'Faringdon' which resulted in his original treez / ocset identity being locked). No, he is not a hobbyist, despite some recent claims, he has worked (actually contracted) for several dozen companies, including a rather undistinguished spell working on outdoor LED lighting (which is where much of the eye-rolling has stemmed from).

I sympathise with your concern about your limited time. In fairness, it is probably simplest if you take a random browse though the previous thread title history and make up your own mind... https://www.eevblog.com/forum/profile/?area=showposts;sa=topics;u=139928

Bon chance,

Chris.

[janoc]

Well, I am not a power engineer, but if you take a ballpark 80% efficiency

The one thing you definitely would NOT do, if tasked with design of new high power density PSU, is to accept such poor efficiency target.

...

90% is really a modern-day decent target but for such high power density, better aim higher, and you STILL need some rather noisy fan cooling, as you realistically can't get the dissipation below 200-300W no matter how good you are or design/BOM budget you have.

Oh I have never suggested that that 80% should be the goal. I merely took that number as a common ballpark figure to estimate the cooling requirements because the OP didn't specify anything at all about the supply, apart from the dimensions. Who knows, maybe more wouldn't be achievable because of some (unstated) constraints.

Even at 90% the required dissipation would be massive for that volume - at least 450W or so without any margins.

[jonpaul]

janoc, based on my experience in the epoch 1970s...1990s:

 SMPS efficiency is a function of line and load, PFC, physical size and cost  trade-offs,

Linear supplies  can be 40..70% but most SMPS are 80..95% efficient.

Modern design topologies, type of circuit, switch frequency,  choice of  GaN or SiC switches ,soft switching and optimization of magnetics are all part of an energy efficient design.

Of course a commodity consumer supply, e.g.  from China,  will be optimized ifor lowest possible cost, thus worse efficiency. 

See IEEE power electronics and industry application proceedings, PCIM, PELS, PESC and other power electronics conference proceedings over the years for state of the art development.

bon courage et bon chance

Jon

[janoc]

janoc, based on my experience in the epoch 1970s...1990s:

 SMPS efficiency is a function of line and load, PFC, physical size and cost  trade-offs,

Linear supplies  can be 40..70% but most SMPS are 80..95% efficient.

Modern design topologies, type of circuit, switch frequency,  choice of  GaN or SiC switches ,soft switching and optimization of magnetics are all part of an energy efficient design.

...

Jon, I am not arguing with you I understand all that. You are of course right - that's why I have mentioned also the 95% value in my original comment.

I was only trying to explain how I came up with that 80% number, as a the kind of lowest reasonably expectable limit - I certainly don't think it would be a good target to aim for when designing a supply like that, especially not for a 4kW one!

[pardo-bsso]

Hey Treez, how's it hanging there?

I'm in a bit of a rush but Google sponsored a competition to design a server power supply with very similar specifications.

You should really look into the entries and the winners