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

[momaka]

Originally Posted by Pentium4

I'll recheck tomorrow. I'm pretty sure it was a 30A part for the 5V though

Even better. It's probably a SBL3040PT or SBL3045PT. You can add some serious capacity to the rails of a PSU with this component. The only place where you likely won't be able to use it is on the 12V rail of half-bridge PSUs. Single and double-transistor forward converter design PSUs can use it on the 12V rail though.

[Wester547]

Do half-bridge PSUs have more of a limit in what voltage their rectifiers can handle on each rail or something? ^

[momaka]

Originally Posted by Wester547

Do half-bridge PSUs have more of a limit in what voltage their rectifiers can handle on each rail or something? ^

No, but the reverse voltage the PSU puts across the rectifier is much higher (usually at least 2 times the voltage of the rail on which the rectifier is on - i.e. 10V on the 5V rail, and 24V on the 12V rail). With high load, the voltage spikes can be even higher. Therefore, you need a rectifier that can withstand the high reverse voltage. A SBL2040PT can withstand up to 40V reverse voltage. The reverse voltage spikes on the 12V rail may actually jump that high in some cases under high PSU load.

[Pentium4]

It's 30A. I'll look for a PSU where it can fit, I have a few 30A parts lyin around

[tleu8472]

2theMax VP350R. Seems to have a complete transient filter stage. Pair of Capxon 680uF primaries. Tiny heatsinks though. Junk or a keeper?

[Behemot]

Seems to be able to do like 300-350 W. But the input caps seem rather small for 680 uF to me.

Just wanted to ask about this heatsinsk too. They have rather small area at first sight, but I think I can help it by bending the fins. On the other hand, it is thicker than the usuall thin crap with more fins. Do you think it will be better for cooling?

[rogfanther]

Bending the fins you can improve somewhat the air flux. Any improvement is always good.
The thickness won´t help too much after the thing reachs a steady state of heat transfer. Good for transient variations, though.

[tom66]

Essentially* with fan cooled power supply thermal design, you don't care how thick your heatsink is, because when almost all of the cooling comes from the air flow, you just need more surface area. Bending the fins like that is essentially getting free extra surface area for the same metal cost, plus a little extra mechanical work, but nothing compared to making the heatsink bulkier.

*There's a lot of complex theory that I don't understand but essentially the thickness only helps with fanless (i.e. static) cooling. Obviously it will have a dissipating effect so will work regardless of airflow, but will have limited extra benefit when airflow is introduced. A thicker heatsink may also allow for greater transients as it has higher thermal capacitance (yes, there is such a thing as thermal capacitance, along with thermal resistance.) Greater thermal capacitance means the heatsink takes longer to reach a specific temperature. The energy eventually gets dissipated though, but over a longer period of time.

[Behemot]

Yeah but by bending it you improve heat trasfer into air by a great margin, and also the heat flows with smaller resistance into the fins.

On the thin heatsinks I believe the heat cannot even flow into the fins properly so they are cold all the time and the lower part with transistors and/or rectifiers is hot than.

[tleu8472]

Why does this PSU only have one toroidal coil in the secondary filter stage? I have typically seen two in most power supplies?

[tom66]

Originally Posted by tleu8472

Why does this PSU only have one toroidal coil in the secondary filter stage? I have typically seen two in most power supplies?

I guess it doesn't use a magamp for the 3.3V, but runs it directly off the transformer.

[Behemot]

What are the transistors and rectifiers inside? Seems like 13009 and some TO-220, if they are shottkies, than it's definitelly better than it seems on first look…

[rogfanther]

Originally Posted by Behemot

Yeah but by bending it you improve heat trasfer into air by a great margin, and also the heat flows with smaller resistance into the fins.

In the transfer of heat by convection, the flux is proportional to surface properties, area , and the fluid properties / speed of movement.
Bending the fins so that the fluid ( air ) can move freely over them improves the transfer from the aluminum surface to the air. Fins much close to each other "trap" hot air between them. The movement of heat inside of the fins won´t improve if they are bent ( could be decreased, as the corner would create a resistance, like in electricity ) .

On the thin heatsinks I believe the heat cannot even flow into the fins properly so they are cold all the time and the lower part with transistors and/or rectifiers is hot than.

As the transfer of heat by conduction ( inside the fins ) also depends of its transversal area, a too thin dissipator would also hamper heat transfer ( Q = k.A.DeltaT, where k is a coeficient that depends of the material, and A is the transversal area to the heat flux.

[goodpsusearch]

Definitely a keeper!

[tleu8472]

Primary switcher is a ST P10NK60Z. I can only make out the 3.3v rectifier which is a MBR3045.

[Behemot]

Originally Posted by rogfanther

In the transfer of heat by convection, the flux is proportional to surface properties, area , and the fluid properties / speed of movement.
Bending the fins so that the fluid ( air ) can move freely over them improves the transfer from the aluminum surface to the air. Fins much close to each other "trap" hot air between them. The movement of heat inside of the fins won´t improve if they are bent ( could be decreased, as the corner would create a resistance, like in electricity ) .

Lets keep it simple, nobody is interested in convection here, we are talking about fan-cooled PSUs. What I am talking about here is with this type of heatsinks, half of their surface almost does not get in contact with air, so it cannot transfer any heat to air, right? If you bent it, you make obstacle and slow down the air a little, on the other hand, transfer much more heat into it because you simple add surface area which was almost unusable before.

Originally Posted by rogfanther

As the transfer of heat by conduction ( inside the fins ) also depends of its transversal area, a too thin dissipator would also hamper heat transfer ( Q = k.A.DeltaT, where k is a coeficient that depends of the material, and A is the transversal area to the heat flux.

One part of the area (lenght) is same in both cases, but the thickness can be double or even tripple. That means double or tripple heat conduction. Or otherwise, thinner heatsinks mean worse conduction, so, we can keep the fins cool (as they have big surface enough and we pump enough of air through it), but as long as heat couldn't be transferred so effectivelly from source (transistor/bridge), temeperature (delta T) on the hot end increases. Because the amount if heat (Q) is the same. (or, in reality, even higher the higher the delta T).

So, what I am basically dealing with is, whether like twice or tripple the crosscut has higher impact on hot end than 50 % higher fin area. In reality, the crosscut may be much greater than 2-3times on thick heatsinks, because if you have seen latest crappy heatinks, the part with fins is made from 50% thinner material than the rest (where transistors/bridges are attached to)!!

[rogfanther]

Originally Posted by Behemot

Lets keep it simple, nobody is interested in convection here, we are talking about fan-cooled PSUs. What I am talking about here is with this type of heatsinks, half of their surface almost does not get in contact with air, so it cannot transfer any heat to air, right? If you bent it, you make obstacle and slow down the air a little, on the other hand, transfer much more heat into it because you simple add surface area which was almost unusable before.

Yes, we are saying the same thing in different ways. While the external area of the heatsink remains the same, with the fins alls grouped near themselves the air cannot move between them and take the heat away. Bending them so that air is free to move around and away of them improves heat transfer.

So, what I am basically dealing with is, whether like twice or tripple the crosscut has higher impact on hot end than 50 % higher fin area. In reality, the crosscut may be much greater than 2-3times on thick heatsinks, because if you have seen latest crappy heatinks, the part with fins is made from 50% thinner material than the rest (where transistors/bridges are attached to)!!

The higher fin area has higher impact in the final heat transfer. The precise amount would need to be calculated.

[tom66]

The weird thing is, there's a limit to the number of fins for optimal efficiency. It turns out that heatsinks wouldn't be best with a theoretically infinite number of fins - they'd work worse.

[Wester547]

I'm interested in convection cooling!

So thick heatsinks are ultimately the best for heat transfer in PSUs, like those found in older Hipros, Newton Powers, Liteons, and Deltas? I would think that thinner heatsinks would get hotter because while they do transfer heat they do not do so as effectively by way of more limited surface area, and as stated before me, too many fins isn't necessarily good. If anything else, though, I thought thicker heatsinks hugely help mediocre capacitors last much longer than usual, though well built voltage filter circuits and overbuilt rectifiers can do just that, along with fans that blow copious amounts of air. I also think vents on the side and large vents on the back and bottom help too....

It's true though, I think, airflow from the fan helps worlds more than heatsinks do, that's why PSUs overheats and fail so hard when the fan fails, even components mounted to heatsinks...

My only issue with thick heatsinks is that sometimes they might almost take "too much" space: so much so that other components not mounted to the heatsink or hidden under the thickness of it may not be as easy to cool. Not only that, but they might limit how much air is blown out of the case from the PSU fan. Given that PSUs are built in less space now adays, I can understand the move to thinner heatsinks but better airflow (120mm-140mm fans mounted on the bottom for more effectual cooling, though that's a problem for sleeve bearing fans). Ultimately, more or less, it's about increasing efficiency so less power and heat is "squandered" or "lost". The same reason why FSP have 80%+ efficient PSUs with unimpressive heatsinks.

As for the PSU, I think one toroidal coil and two ferrite coils for the PI filter is fine: it's a transistor forward design so it's more effective. Remember that toroidal coils can get warm, and it seems there is a decent amount of coils on the primary.

I've always wondered, though... how would a pair of 13009s (TO-220 or TO-3P/TO-247 as some of them are) compare to the rectifier shown (STP10NK60Z)? Especially since this is a forward rather than bridge PSU...

[c_hegge]

I have a PSU just like that (a 430W Thermaltake). That PSU actually uses DC-DC conversion from the 5V rail, which is why there is no separate toroid coil for the 3.3V rail. The part for it on the heat sink is a MOSFET.