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

[momaka]

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

What I think would be good idea is to have a universal voltage for all ICs and components (CPU, RAM, GPU, chipsets, etc). That way, the PSU could go directly down to that voltage, and there would be no VRMs on motherboards.

But VRMs on motherboard are quite efficient. If anything, I'd go with Behemot's suggestion and have just a single 24V rail (and maybe 12V standby) and then have everything else stepped down on the motherboard.
Moreover, you're not going to see things like audio and ethernet chips use anything as low as CPU, especially audio.

[mariushm]

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

Unfortunately, there's just too many usb devices out there, so 5v standby will have to stay. The USB 3.0 still allows 5v but also allows 12v through it with suitable cables.. who knows, maybe in about 10 years the 5v will be deprecated and only 12v devices will be allowed.
Platinum psus (and some gold eff. ones) use relays or some other methods to commute 5vsb to the 12v-to-5v dc-dc converter once the power supply is started

No, I don't think we're going to see soon voltages higher than 12v .. there won't me much higher efficiency gain going from 12 to 24v. In addition, you can use 16v rated capacitors (both electrolytic and polymer) with 12v , while you'd have to use 35v rated caps with 24v, which are much wider and may not even exist (looking at polymer low esr).

12v is also good because once you validate the hardware, you can use it straight from 12v batteries and bypass the ac to dc section of a power supply. 13.8v is not that much higher and a lot of hardware tolerates it (some fans only handle up to 13.5v though, but that's nothing a diode won't fix)

I would like to see 3.3v rail go away, nowadays only some 2.5" ssds are using it, so the mb could have a dc-dc converter in them to provide it, or the ssd itself could just have a dc-dc converter that would support 3.3v-5v from the start.

In a new design, I'd like to see maybe a 10 pin molex like the current ones (power on+ power good, 4 x 5v (don't differentiate between 5vsb and 5v, and all 4 should handle about 15-20A) , 4x12v (again about 5-6a per wire safely).
This would the base connector , 15-20a for 5v , 20-24A for 12v , ideal for mini-itx systems that use maybe 100-150w tops so there's no need for a huge 24pin + 4pin power.
Then for the systems that need more power, just add a second 10 pin molex with 12v only... this should give about 50-60 amps more, which should be enough , considering modern hardware starts to use less and less power.
The power good+power on could basically act like bi-directional i2c/spi and so the mb could basically send the mb info about the power supply, like how many amps it can give though 5vsb, so that the motherboard could inteligently set some usb ports in high power charging mode, or tell it if it got switched from ac input to battery backup (in the case of psus with dc in mode etc etc)

[domas]

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

Marius, I think you are mistaken.

USB3 does not have 12V supply.

eSATAp, does though, but it has to be dual-power eSATA. Most desktops have this output, and very rare laptops have 12v in esata.

[mariushm]

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

it's negotiated by the controller and it's up to the controller to support as many or as little power profiles:

http://semiaccurate.com/2012/07/23/1...delivery-spec/

PD specifies, but does not require absolutely sticking to, six profiles for power. The intent is to deliver between 10W and 100W in steps that roughly double the one before.

These six profiles are cleverly named Profile 0-5, with 0 being reserved.

Profile 1 is 2A@5V or 10W, roughly what Gigabyte delivers now on their 3X power boards.
From there, Profile 2 is 5V@2A or 12V@1.5A,
Profile 3 is 5V@2A or 12V@3A,
Profile 4 adds 20V@3A to it’s predecessor, [ my note: so it includes profile 3 options]
and Profile 5 bumps the 12V and 20V to the full 5A.

We hear that the real reason Profile 0 is reserved is because Apple kept vetoing the PD-AJS (PD-Auto Jump Start) addendum for fear of their precious cables getting dirty in public. The USB Promoter Group will not confirm or deny whether this was the case.

[domas]

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

ok nice.

How do I find out if my computer has 12V output on usb3?
it doesn't have on eSATAp.

it seems that it does not have one on usb3 either:

[kc8adu]

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

at the high currents but low voltages the cpu,gpu,chipset use it would be impractical.
the very heavy wiring and connectors alone would be a show stopper.
going the +24v route would be the most senseable.
then its business as usual with better efficiency.same buck converters at the served load.

What I think would be good idea is to have a universal voltage for all ICs and components (CPU, RAM, GPU, chipsets, etc). That way, the PSU could go directly down to that voltage, and there would be no VRMs on motherboards.

[cheapie]

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

What I think would be good idea is to have a universal voltage for all ICs and components (CPU, RAM, GPU, chipsets, etc). That way, the PSU could go directly down to that voltage, and there would be no VRMs on motherboards.

But then you couldn't overvolt, and the required cables for the current that a modern CPU takes would be huge - it's something like 100A at 1V.

[mariushm]

Re: Power supply build quality pictorial. part 2

And here's the shitty power supply that the power supply above is gonna replace.
This one in theory still works but I'm a bit concerned about the 12.5v output. After opening it up, I realize it's actually worse than the above power supply.

Just some Y caps that I doubt they're safety rated on the primary side,
puny rectifier diodes good for probably 1A each,
330uf/200v YC brand capacitors that in reality are 216 uF each (and about 0.15 ohm esr not that it matters here)
didn't bother to write down the mosfet/transistors on primary, i though they can be read from pictures... can't be bothered to open it again
ei-33 (in theory) transformer

SBL1640CT (16a/40v max) schottky on 5v, SBL1040CT (10a, 40v) or 3.3v, two big diode treatment on 12v (who knows, maybe 4-6 amps? )
Some YC caps on secondary side, some Su'scon (probably on 5v and 3.3v) , I think maybe a G-Luxon on 12v? 470uf/16 on -5v , 1000uf/16 otherwise as far as i can see
no secondary side inductors in sight, just a couple of minimum load resistors (3.3v and -5v?)

Weak, really weak...

[c_hegge]

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

^
Copied from PSU Build quality pictorial thread

[Heihachi_73]

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

Another typical L&C. That thing looks like it would struggle to power a clock radio.

I still have an L&C diode bracket (poor man's MOSFET) modern art piece in my diode/transistor collection.

[domas]

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

on the other hand, this p/s is very easy to work on because it is easy to access components that are there due to "forgotten" components. Damn its empty.
Typical

[Agent24]

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

I still have an L&C diode bracket (poor man's MOSFET)

More like Chinese fake MOSFET.

[TELVM]

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



'REAL POWER: 250W'

[rogfanther]

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

Probably the real "real" is around 175-200W ...

[momaka]

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

on the other hand, this p/s is very easy to work on...

Indeed.
The fact that it uses leaded solder makes it even easier.
It seems that these older crappy PSUs are great when it comes to cooking your own sub-250W adjustable PSU.

'REAL POWER: 250W'

Yeah, I think they meant UNreal power.
And next to where it says "500W" on the sticker, it should say, gamers's overclocked supper power!

[Pentium4]

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

Not fair! You uploaded a label shot and no internal pictures? Better get on that so the criticisms can begin

[TELVM]

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

Not fair! You uploaded a label shot and no internal pictures? Better get on that so the criticisms can begin

Here you go: Warning - Not for the faint hearted!

[mariushm]

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

Surprisingly not that bad. Compared to my psu posted above, this one's much better. Might actually do 250-300 watts provided the loads are evenly spread on all voltages.

later edit: Ah, I see it didn't do so well with ~ 230w load.

later edit 2 : TELVM if you contribute to that site, why the hell does it not allow google translate to translate the forum?

[TELVM]

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

... why the hell does it not allow google translate to translate the forum?

Here's Google translation into english of that article: http://translate.google.com/translat...DX500SE%2F2799

The reviewer is Gabriel Torres, the same guy behind Hardware Secrets .

[rogfanther]

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

That is just another of the weak psus that are sold here in Brazil with description of "Nominal 500W ". The common joke nowadays when the seller says that is that the 500W is in the name.

But some are learning ( slowly ) . Some places are already calling them a "Real" 200W psu. Given some margin of error, that would amount to a 175W useful, but many old machines won´t complain ( much ) .