Power supply build quality pictorial. part 2

[Behemot]

Re: Power supply build quality pictorial. part 2

Samxon are good caps.

As for quality, I am more or less the only person importign them into Czech republic ATM. And even though I also want to bring the price down, I rather still use quality caps, even though for example old. Got 1000 pcs of 7 years old Rubycon MBZs 1000 uF/6,3 V, real capacity is 750 uF. Still much better than those original craps I swap Rubycons for and I got them for very good price.

[b700029]

Re: Power supply build quality pictorial. part 2

That LC-B400ATX is definitely L&C/Deer. Although it's a max 350W design (as the PCB shows), it's unusually full for a Deer. The 2003 controller in SMD format is also a bit unusual. Butterfly on the logo looks like the Superflower one...

I don't think the Xilence is a Leadman/Powmax. They're usually more pathetic, like this. I see "WIN-12??" in the corner, so it might be some other OEM.

[mockingbird]

Re: Power supply build quality pictorial. part 2

Power Man IP-P350AJ2-0 350W

I originally relegated this one to the scrap PSU heap, but being a 24-pin PSU, I thought I'd give it another look. I originally thought this was an FSP, but it most certainly isn't. There's actually a logo on the PCB (Sorry, you can't see it in the pic) and it says "300W REV 1.2" in the top left corner. Anyone recognize the design?

Infineon ice2a0565z for the 5VSB circuit. 2 x caps, 2 y caps.

Secondary FETs are:
2 x STPS4045CW
1 X LT5216

Supply looks a little anemic component wise (Though I am delighted that it has a standby IC for the 5VSB), not to mention the puny heatsinks, but aside from all that, do you think this is a good candidate for re-capping?

[mariushm]

Re: Power supply build quality pictorial. part 2

At least the label is honest, "inspired by in win" ... the UL number belongs to IN WIN

In Win E193791

.. and the model is listed on ul.com and there's an InWin model with the same code:

http://www.newegg.com/Product/Produc...82E16817103149

There's even a review of this psu here: http://www.xbitlabs.com/articles/cas...2_6.html#sect0

It looks ok to me but I'm no expert. I'd recap it.

[c_hegge]

Re: Power supply build quality pictorial. part 2

That is a decent PSU. In Win build their own PSUs now, and they know what they are doing. That PSU will more than likely deliver 350W.

[mockingbird]

Re: Power supply build quality pictorial. part 2

You see the pi filter with the big one and small little ones that almost looks like a square? Those three little ones are 16V 470uF. Should I beef that up?

[c_hegge]

Re: Power supply build quality pictorial. part 2

Depends which rails they're on. At least one looks like its on the -12V rail (with another one possibly on the -5V), where 470uF is plenty. If the other two are on the main rails (5v, 12v or 3.3v), then you could try beefing them up to 1000uF or so, but I don't know how tolerant the feedback circuit will be.

[Th3_uN1Qu3]

Re: Power supply build quality pictorial. part 2

The feedback is always taken before the pi filter, so feel free to add bigger caps after it. Also, in a group regulated design, the dominant rail in the feedback is the 5v rail, so unless you do something really stupid on 5v, or use it with no load on 5v (impossible in a computer), it'll work just fine regardless of caps.

[c_hegge]

Re: Power supply build quality pictorial. part 2

Enlight GPS-300AB-100 V



Looks like a typical 300W Delta design. Heat sinks are nice and thick.


Samxon 560uF Primaries. This is a single transistor forward design with a Fairchild FQA8N80 as the switching transistor.


Lousy Trash Electrolytic Capacitors are used throughout the secondary. Rectifiers are 30A on the 5v and 3.3v rails and 16A on the 12V. Obviously, it's an older design.


Fan is a Yate Loon sleeve bearing. Not too bad, but a Delta fan would have been nice.


As usual, Delta's soldering is great, but there's some heat discolouration near the 5vsb area. Most of it seems to be just before the transformer (on the primary side, near R902 and D951).


And yes, that diode is practically touching the cap. Considering that there is discolouration around the 5vsb input cap (which is the only capXon in this PSU), I'm surprised it's not cooked yet.

As for mods, I'll probably beef the 12V rail up to 30A or so, and replace the 5vsb diode with a TO-220 one and mount it on the secondary heatsink, so it doesn't cook the cap. I'll probably also recap it with pannies and Nichicons.

[Wester547]

Re: Power supply build quality pictorial. part 2

Lousy Trash Electrolytic Capacitors are used throughout the secondary. Rectifiers are 30A on the 5v and 3.3v rails and 16A on the 12V. Obviously, it's an older design.

All the Enlights (Delta or Sirtec) I see (of that age) usually have the most power regulated to the +5V and +3.3V rails and the least share of power to the +12V rail. The heatsinks are good, though I think it would have done better with larger (and thicker) heatsinks, but that's just me. But given that this is a single-transistor forward design, does that limit the theoretical current of the rectifiers more? Hardware secrets always says that, by their arithmetic, single and two transistor forward topologies divide the limit of the rectifier by a 30% duty cycle (and subtract it by 1), though I suppose that is just theoretical?

And yes, that diode is practically touching the cap. Considering that there is discolouration around the 5vsb input cap (which is the only capXon in this PSU), I'm surprised it's not cooked yet.

Disappointing to know that Delta doesn't hesitate to improperly place diodes and use conductive glue in their PSUs. Not sure when Delta and Bestec will learn.

[Agent24]

Re: Power supply build quality pictorial. part 2

Disappointing to know that Delta doesn't hesitate to improperly place diodes and use conductive glue in their PSUs. Not sure when Delta and Bestec will learn.

They know exactly what they're doing. It's an inbuilt failure mechanism to ensure the thing breaks sooner rather than later so you'll buy another one.

[Th3_uN1Qu3]

Re: Power supply build quality pictorial. part 2

Hardware secrets always says that, by their arithmetic, single and two transistor forward topologies divide the limit of the rectifier by a 30% duty cycle (and subtract it by 1), though I suppose that is just theoretical?

I dunno where they picked that up - the maximum duty cycle is still 50% (even more with some tricks, but more on that some other time).

What is true however, is that single forward uses half-wave rectification, and the second diode in the double diode packs is only used as freewheeling diode (to discharge the output inductor and prevent the output voltage from spiking negative), so the two diodes do not share the load evenly, the freewheeling diode contributes less. That is why in single and double forward topologies, as opposed to bridge, the output current rating is always lower than that of the double rectifier diodes.

[Wester547]

Re: Power supply build quality pictorial. part 2

How much lower, though? Or does that vary greatly upon use/conditions? And I know they design their units to fail. I just think they make shoddy decisions. And some rectifiers are rated at 25C.

[c_hegge]

Re: Power supply build quality pictorial. part 2

The other thing to bear in mind is that, even if that formula Hardware Secrets use is true in theory, it has one big problem - The datasheet ratings are what it will deliver at 110*C +. In a PSU, they're unlikely to get that hot, so they will often deliver more. I generally just assume the rating of the rectifier as the theoretcal maximum, regardless of the topology.

[Wester547]

Re: Power supply build quality pictorial. part 2

The other thing to bear in mind is that, even if that formula Hardware Secrets use is true in theory, it has one big problem - The datasheet ratings are what it will deliver at 110*C +. In a PSU, they're unlikely to get that hot, so they will often deliver more. I generally just assume the rating of the rectifier as the theoretcal maximum, regardless of the topology.

Some are rated at 25C. Albeit without a heatsink - with a heatsink and with a fan, they can indeed do significantly more. What's the rating of the rectifying bridge used in that Delta/Enlight?

[c_hegge]

Re: Power supply build quality pictorial. part 2

It's an 8A. Some bridge rectifiers can be rated at 25*C, but for TO-220 and TO-247 rectifiers which you find on heat sinks, it's always 110*C or higher

[c_hegge]

Re: Power supply build quality pictorial. part 2

That is why in single and double forward topologies, as opposed to bridge, the output current rating is always lower than that of the double rectifier diodes.

Gabriel disagrees. He recons that in a half bridge, both diodes are used, but only one at a time.

[Behemot]

Re: Power supply build quality pictorial. part 2

And true is that most diodes have these currents for low frequencies. At some us level they state being capable of many times more, and if we are talking about frequencies in 40-100 kHz interval, just do the math, it's 2,5us period at most, that means 1,25 us half-period.

[Th3_uN1Qu3]

Re: Power supply build quality pictorial. part 2

Gabriel disagrees. He recons that in a half bridge, both diodes are used, but only one at a time.

Just like in a regular mains frequency full-wave rectifier, the diodes in a half bridge PSU are used one at a time, because the signal alternates between the two polarities. That means that the maximum current rating of each rail in a halfbridge PSU is the rating of the double rectifier diode (of course, if you can keep it cool enough), since each diode in the pack does half the work, half the time and the currents add up, and that's what Gabe says too.

if we are talking about frequencies in 40-100 kHz interval, just do the math, it's 2,5us period at most, that means 1,25 us half-period.

You got your math wrong by one decimal place. 100kHz is 10us, 40kHz is 25us.

[c_hegge]

Re: Power supply build quality pictorial. part 2

Just like in a regular mains frequency full-wave rectifier, the diodes in a half bridge PSU are used one at a time, because the signal alternates between the two polarities. That means that the maximum current rating of each rail in a halfbridge PSU is the rating of the double rectifier diode (of course, if you can keep it cool enough), since each diode in the pack does half the work, half the time and the currents add up, and that's what Gabe says too.

Nope. He reckons that the limit comes from half bridges having a 50% duty cycle, and that only one diode is used in the math, like in a forward design. If we used both diodes and his formula (with the same 50% duty cycle), we get double the rectifier's rating.