Power supply build quality pictorial. part 2

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Nothing wrong with CapXon? Could have fooled me.
Are you saying all those bulging\leaking CapXon I've pulled and replaced were OK?

No they were probably not.

But that does not mean that CapXon still is as bad as they were a couple of years ago...

And it's the question if the caps were misused or not.
Especially in those FSP Epsilon units....
But that one was a couple of years ago...

So we have to wait a couple of years to see if those CapXon caps are still failing or not...
They are used in PSUs like Super Flower's HX series or Corsair RM.

My guess would be that they will be fine in the super flower but not so much in the Corsair RM...

But we might not know the operation parameters of those (failed) caps -> how much ripple did they endure, how much heat did they have to endure...

PS: I have a device with some failed CapXon GL caps on it.
It's the PSU of my Philips 200P6IS, 20" 4:3 IPS screen. But I really used that one a lot. So that screen may have even more than 10kh usage endured. So I don't really blame the caps for that. That screen is almost 10 years old (and just a 16ms one)...
So nothing wrong with them failing after ~8 Years of daily use (up to 12h or even more per day, mostly not in standby)...

[Behemot]

Re: Power supply build quality pictorial. part 2

Are you saying all those bulging\leaking CapXon I've pulled and replaced were OK?

Most chinese caps are PoS and CrapXon is leading the shit legions. The only thing in its favor is PSUs today are 5-10 % more effective than 10 years ago so they are not stressed so much. But their time will come! Maybe not in 5 years, than in 7 years.

I have a customer now who wants replacement for those shiny blue (not dark blue) OSTs on one motherboard. How many times I've read how great they are? Bullshit.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

I don't recollect Delta ever using AsiaX, though.

Right, they used Aishi...
I threw them in the same pot as AsiaX, JunFu and the likes..

PS: and there are some Enermax out there with JunFu caps (Triathlor Eco) as well as THermaltake (some Smart Power or the Europe series)...

So I'd rather have 100% CapXon than a japanese primary for marketing and most of the caps are those blue shits...

[Behemot]

Re: Power supply build quality pictorial. part 2

It's like from shit to mud…

[Agent24]

Re: Power supply build quality pictorial. part 2

But we might not know the operation parameters of those (failed) caps -> how much ripple did they endure, how much heat did they have to endure...

PS: I have a device with some failed CapXon GL caps on it.
It's the PSU of my Philips 200P6IS, 20" 4:3 IPS screen. But I really used that one a lot. So that screen may have even more than 10kh usage endured. So I don't really blame the caps for that. That screen is almost 10 years old (and just a 16ms one)...
So nothing wrong with them failing after ~8 Years of daily use (up to 12h or even more per day, mostly not in standby)...

The funny thing is, when I replace CapXons in PSUs which also use Rubycon etc, a pattern emerges: The CapXons are blown and the Rubycons etc are still well within ESR spec.

If the CapXons were so great, surely they would still be OK!

After seeing so many failures of Chinese\Taiwanese capacitors both on the forum and in my own life, I don't trust them anywhere near as much as the Japanese ones. Yes, those will fail too, eventually, but they sure as shit last longer than the others. (with exception of course series like KZG with manufacturing defect)

My motto is "Japanese or go home" and I'm sticking with it.

[c_hegge]

Re: Power supply build quality pictorial. part 2

So nothing wrong with them failing after ~8 Years of daily use (up to 12h or even more per day, mostly not in standby)...

Disagreed. They should last at least 15 years and ideally 20. The Japanese brands have absolutely no trouble with that for the most part. Especially now that an 8 year old C2D E6600 is still perfectly capable of running all of the latest versions of windows, office, IE, and all the programs that everyone uses, and thus still has plenty of useful life left.

btw, that Delta works perfectly now that I've re-capped it. I also oiled the fan, cut out the grille and installed a wire grille in its place. Hopefully it should be fine for another 20 years now.

[fanbus]

Re: Power supply build quality pictorial. part 2

Delta DPS-250AB.

I've got the same PSU. They gave it to me because it wasn't working fine. It turned on with the green-black connection, but not with a computer. I saw that bulging CapXon when I first opened it, but I didn't bother about it until the unit didn't want to turn on. Then I replaced that cap (I don't remember which brand/model I used) and it worked perfectly.

[momaka]

Re: Power supply build quality pictorial. part 2

Disagreed. They should last at least 15 years and ideally 20. The Japanese brands have absolutely no trouble with that for the most part.

+1

I have many Japanese 85C-rated caps from the early 90's. Most of them came from scrap CRT TV/monitors I found. Just about every one tested in spec for both ESR and capacitance. I mean, most of these are 15-25 year old caps! Now that is some good caps.

[Compgeke]

Re: Power supply build quality pictorial. part 2

As for how many connectors it has and the rail distribution, I don't know. I haven't seen an 800LB before.

I don't have pics of the connectors here but I do have gut pics. I don't own the system anymore or I'd go take more pics.
https://www.badcaps.net/forum/showth...747#post512747

You can see the motherboard connectors on pics of an XW8600's motherboard. 8 pin for CPU power, 6 pin for RAM and 24 for power. I believe the CPU was non-standard.

[Compgeke]

Re: Power supply build quality pictorial. part 2

Working with an old Time IPS timecard system thing and inside it's just an ITX board running Linux.

Power supply though is quite nice looking.

Channel Well PSM200A-89. 200W, ATX connectors.


Nice and packed, as one might expect from such a small PSU.


Really really packed.


From what I can see this is all Hitachi and Koshin caps. Never run across Koshins before, seem to be common in HK audio gear from web searches but little other than that.




And finally, the PCB model matches that on the casing.

[ben7]

Re: Power supply build quality pictorial. part 2

^ Big blue cap is Hitachi, small caps are Koshin and fuhjyyu!!!

If you look closely, the Hitachi and fuhjyyu caps do have different logos. They might look the same from a distance though, kinda like "Robicon" and "Rudoycon".

[momaka]

ePower EP-350XP-24 (P022-E platform) (page 118)

Alright, today I decided to post something a little different from the usual HiPro and Macron PSUs:
ePower Xtrem-350 (model EP-350XP-24 based on P022-E platform).

Before I begin my "book", first I’d like to express my gratitude towards Pentium4 here, as he is the one that gave me this cool PSU.

Okay, now let's get down to brass tacks, the real deal, the actual PSU and its guts . Most of you PSU "maniacs" reading this thread know what I mean. Starting with some case pictures...
https://www.badcaps.net/forum/attach...1&d=1432949412
https://www.badcaps.net/forum/attach...1&d=1432949412
Nothing extraordinary here, but you can see that we have us a wire fan grille rather than cheap stamp cutouts in the case, so I think we are off to a good start. Also some sleeved and twisted output cables. And perhaps another notable feature is that some of the cables for the main rails on the ATX connector are 16 AWG rather than 18 AWG. So we are definitely NOT dealing with a cheapie "hong-kong-fly-appart LLC" PSU here .

What the label says…
https://www.badcaps.net/forum/attach...1&d=1432949412
Ah. It is a bit disappointing with all of those different wattages. It's like ePower really didn't want to print different labels for their PSUs, so they just did a "one-label-fits-all" here. But anyways, this is the 350W version, so that's what we'll be looking at…
Label claims two separate 12V rails capable of 18A each, but combined total is 20A, or 240W of power on the 12V rail alone. That should be plenty for a 90W TDP CPU and a medium-power graphics card, though. And one last bit of info we get from the label is who made this power supply. UL number E1390843 means this PSU was made by Topower.

Let's remove some screws now to see the guts…
https://www.badcaps.net/forum/attach...1&d=1432949412
Oh my! Did I just have a deja vu or something?... *drinks some cold water* Yep, that is a double warranty sticker . But I don't think I can get double the warranty here .

Anyways… Moving onto the real guts. Top view:

We see four transformers here (the one on the upper right is a CM choke for the AC line). Definitely quite unique design, as you will see in a bit (at least nothing that I have seen before). It's basically a half-bridge PSU with an extra transformer that seems to be used for over-power protection/sensing (OPP). As for the sizes, the main traffo appears slightly bigger than your typical "35". I think it is a "39"-size.
Other things to note: there is an actual AC connector on the board and the fuse has its own holder. The switch on the back of the PSU is a DPST. All mains-wiring is 18 AWG rated for 600V. And, of course, nice, big, black heatsinks .

Here are a few more internal shots:
https://www.badcaps.net/forum/attach...1&d=1432949412
https://www.badcaps.net/forum/attach...1&d=1432949412
https://www.badcaps.net/forum/attach...1&d=1432949412

And finally, a solder-side and fan picture – all for your viewing pleasure .
https://www.badcaps.net/forum/attach...1&d=1432949412
https://www.badcaps.net/forum/attach...1&d=1432949412
The soldering is quite good on this PSU and the board was cleaned very well from leftover flux. Also, note the fan: although it is a sleeve bearing Globe Fan, it has a much bigger rotor assembly and the plastic is shiny, smooth, and well-formed. The fan also spins very "smoothly". So I think this is definitely a higher quality Globe Fan. Plus, it is temperature controlled, and the controller itself actually has connectors for two addition fans. Cool!

Now, these pictures still can't really tell you how unique this PSU is, so I will do a detailed breakdown of the components next.

Primary Side
2x single-mode chokes on AC receptacle + 1x CM choke
1x 0.33 uF + 1x 0.1 uF safety X-caps between L-N. Also 5x Y-safety caps.
GBU8J bridge rectifier (8A, 600V) and has its own small heatsink.
2x Teapo LXK 200 V, 680 uF primary bulk caps, 21 x 44 mm (d x h)
2x 2SC2625 NPN BJTs for the main PS + 2.2 uF film cap
2x YihCon RK 50 V, 4.7 uF caps for the BJT base drive
1x 2SK2645 N-Ch MOSFET for the 5VSB

Controllers and ICs
This is where it gets interesting. There is a UC3842b PWM controller. However, this is not for the main PS – it is for the 5VSB. Yes, a whole PWM controller just for the 5VSB! Why Topower didn't use a PWM-FET combined IC? I don't know either. But this is still much more enlightening than a 2-transistor design. As for the main PS, that uses a standard TL494 controller in conjunction with a LM339N quad comparator.
BUT WAIT THERE IS MORE!
The 3.3V rail does not appear to be linear or mag-amp regulated like you might expect. Instead, it is (I think) a single-ended DC-DC buck-converter (high-side MOSFET + free wheeling diode + output coil), with its source being the 5V rail. Now, what I am not sure about is what controls this. There is a KEL321 chip on the board from ST micro (which I ASSume is equivalent to a TL321 dual op-amp and is what drives the 3.3V rail MOSFET). Finally, the -5V rail (yes, there is one on this PSU) uses a 7905 IC and it is attached to the secondary heatsink.

Secondary Side
5V RAIL:
2x D83-004 (30A, 40V) schottky rectifiers in parallel, 100 Ohm load resistor
2x Teapo SC 6.3 V, 3300 uF, 10 x 25 mm (d x h) caps with 3-turn, 6 mm PI coil in between them.
1x Fuhjyyu TMR 10 V, 2200 uF, 10 x 25 mm, with a 5-turn, 6mm coil separating it from the 5V rail. This is the connected to the (upper) MOSFET for the 3.3V rail buck regulator.

3.3V RAIL:
No rectifier (shared capacity of the two D83-004 on the 5V rail), 100 Ohm load resistor
CEP8030 or CEP8060 MOSFET for the buck regulator (not sure of label, since I could see only the letters in bold.
SBL2040CT free-wheeling diode for the buck regulator
2x Fuhjyyu TMR 10 V, 2200 uF, 10 x 25 mm (output of buck regulator), with 6-turn, 5 mm air-core PI coil in between these caps

12V RAIL:
2x BYQ28E fast recovery rectifiers (10A peak per diode, 200V) in parallel
330 Ohm load resistor
1x Fuhjyyu TMR 16 V, 2200 uF, 10 x 30 mm, with 18-turn, 6 mm PI coil before it

-12V RAIL:
Two 2 A (?) diodes, 220 Ohm load resistor
1x YihCon RK 16 V, 220 uF, 6.3 x 11 mm and 1x Fuhjyyu TNR 16 V, 470 uF, 8 x 13 mm
With ~36-turn, 4-5 mm PI coil after the YihCon cap above.

-5V RAIL:
Can't see what rectifiers it has… 560 Ohm load resistor
7905 linear regulator
1x 1x Fuhjyyu TNR 16 V, 470 uF, 8 x 13 mm after the regulator

5VSB RAIL:
SB540 (?) (5A?) schottky rectifier, 220 Ohm load resistor
2x Fuhjyyu TNR 10 V, 1000 uF (?) caps with a PI coil in between them

Finally, I did some brief 5VSB tests.
- With no load on 5VSB, the PSU pulls 3.1W from the wall @ 0.42 PF and sits at 5.06V
- With 1.1A load ( ~5.5W) on 5VSB , the PSU draws 10.3W @ 0.51 PF and is exactly 5.00V. The efficiency is about 53% with that load, which is quite acceptable compared to many other PSUs I've seen (especially some 2-transistor designs).

And that concludes my unofficial review of this thing. I still haven't decided in which one of my future build PCs I'll put it yet (I have several socket 939 and P4 motherboards sitting around). Not sure if I will recap it yet either, since it is NEW! Generally, for my own stuff, my rule is: no Fuhjyyu cap gets a break until it bulges, leaks, or goes high-ESR. So I might just use it as-is for a while.

Anyways... as always, big thank you if you read through my wall-of-text . (and again if you didn't )

[TELVM]

Re: Power supply build quality pictorial. part 2

For the archaeological interest here's a friend's 20+ year old AT Seasonic SSA-200G still alive and kicking powering retro PCs:







I told him to leave the Pannys and Rubys soldier on, and just recap the CEHTRs and check the fan.

... If you look closely, the Hitachi and fuhjyyu caps do have different logos. They might look the same from a distance though, kinda like "Robicon" and "Rudoycon".

[Pentium4]

Re: Power supply build quality pictorial. part 2

Alright, today I decided to post something a little different from the usual HiPro and Macron PSUs:
ePower Xtrem-350 (model EP-350XP-24 based on P022-E platform).

Before I begin my "book", first I’d like to express my gratitude towards Pentium4 here, as he is the one that gave me this cool PSU.

Nice, you disassembled it! Yes, these PSUs are really awesome. I think they're the best designed half bridge units out there, along with the Super Flower units I've seen. I have this exact unit (although it's a little different, more on that later) and I've put 12,000 running hours on it since I've recapped it.

some of the cables for the main rails on the ATX connector are 16 AWG

Isn't that nice for a cheap PSU? By the way, when I picked this up on newegg, it was only $12.99 with free shipping. It was borderline theft for what you're getting

UL number E1390843 means this PSU was made by Topower.

Yeah, one that was actually made by Topower. They use Leadman and Zumax now

Yep, that is a double warranty sticker

Whoa! Never seen that before. That's hilarious! However, I have seen them miss the screw, so that you could disassemble it and keep the "warranty" in tact

It's basically a half-bridge PSU with an extra transformer that seems to be used for over-power protection/sensing (OPP).

Very interesting. I've never seen that either. How does this work? Is it connected to any of the IC's?

As for the sizes, the main traffo appears slightly bigger than your typical "35". I think it is a "39"-size.

I think it's at least 39, maybe even 40. It's pretty large. The pictures don't do justice, because those heatinks are HUGE.

The soldering is quite good on this PSU and the board was cleaned very well from leftover flux.

They definitely improved it quite a bit. I have the same platform PSU that were made in 2004/2005 and the soldering is pretty bad. They improved mostly on the output wire soldering.

Also, note the fan: although it is a sleeve bearing Globe Fan, it has a much bigger rotor assembly and the plastic is shiny, smooth, and well-formed. The fan also spins very "smoothly". So I think this is definitely a higher quality Globe Fan.

I noticed this too. However, check under the seal! Mine had a low amount of drying grease. I cleaned out the grease and put oil in there. It spun even better, and a nicer "rock" to it when coming to a stop. I checked it at roughly around 11,000 running hours and it spins the same as when I oiled it, even with a fair amount of dust collected on it.

Plus, it is temperature controlled, and the controller itself actually has connectors for two addition fans. Cool!

If you don't want to put more fans inside the PSU (which you probably couldn't) if you solder fan or molex headers to those extra fan pins, you can have them go out the back of the PSU. Those headers get a minimum of 5.75V, and slowly climb as the temperature rises. You could use regular 12V molex fans and have them not be unbearably loud without even having to do a wire mod.

Primary Side
2x single-mode chokes on AC receptacle + 1x CM choke
1x 0.33 uF + 1x 0.1 uF safety X-caps between L-N. Also 5x Y-safety caps.
GBU8J bridge rectifier (8A, 600V) and has its own small heatsink.
2x Teapo LXK 200 V, 680 uF primary bulk caps, 21 x 44 mm (d x h)
2x 2SC2625 NPN BJTs for the main PS + 2.2 uF film cap
2x YihCon RK 50 V, 4.7 uF caps for the BJT base drive
1x 2SK2645 N-Ch MOSFET for the 5VSB

Pretty beefy primary. However, I think you got shafted here! Mine came with 820uF 200V Teapo LXK. Weird, since mine was made fairly closely to yours. The serial number was only 150 numbers away from yours.

There is a UC3842b PWM controller. However, this is not for the main PS – it is for the 5VSB. Yes, a whole PWM controller just for the 5VSB! Why Topower didn't use a PWM-FET combined IC? I don't know either. But this is still much more enlightening than a 2-transistor design

Very cool! So it uses the 2SK2645 FET for 5VSB switching, and the UC3842B for output and regulation?

The 3.3V rail does not appear to be linear or mag-amp regulated like you might expect. Instead, it is (I think) a single-ended DC-DC buck-converter (high-side MOSFET + free wheeling diode + output coil), with its source being the 5V rail. Now, what I am not sure about is what controls this. There is a KEL321 chip on the board from ST micro (which I ASSume is equivalent to a TL321 dual op-amp and is what drives the 3.3V rail MOSFET). Finally, the -5V rail (yes, there is one on this PSU) uses a 7905 IC and it is attached to the secondary heatsink.

Hmmm. This thing is just full of surprises! Is it acting similar to a VRM on a ATX motherboard? Stepping the voltage down in a very efficient and clean way? Seems like a much better method than linear conversion.

12V RAIL:
2x BYQ28E fast recovery rectifiers (10A peak per diode, 200V) in parallel
330 Ohm load resistor
1x Fuhjyyu TMR 16 V, 2200 uF, 10 x 30 mm, with 18-turn, 6 mm PI coil before it

You got jacked here, too! Mine came with two 20A ultra fasts in parallel. I put 20A load on the 12V, and the 12V sagged to 11.64V. I think this is because there was a low load on the 5V and 3.3V. I imagine if those had more current going through them, the 12V would come back up a bit, like in c_hegge's review.

What impresses me is their choice in minimum load resistors for all the rails. 100Ω being the lowest, nice!

Finally, I did some brief 5VSB tests.
- With no load on 5VSB, the PSU pulls 3.1W from the wall @ 0.42 PF and sits at 5.06V
- With 1.1A load ( ~5.5W) on 5VSB , the PSU draws 10.3W @ 0.51 PF and is exactly 5.00V. The efficiency is about 53% with that load, which is quite acceptable compared to many other PSUs I've seen (especially some 2-transistor designs).

Thanks for sharing your test results. Yeah, efficiency shouldn't matter too much at such low loads. And with the large heatsink, and no critical caps, it shouldn't be an issue.

One of my favorite things about this PSU besides the silence, and heatsinks, is the size of the output inductors. Look at those things!

I bought a total of 4 of these things, 3 of them with consecutive serial numbers. Remind me of yours, and I'll eventually open up the other two and see if there are any differences.

Not sure if I will recap it yet either, since it is NEW!

Keep in mind that this thing is sadly ~5 years old or more. Since it was made by Topower themselves, and they ceased PSU production in ~February, 2010.

[Wester547]

Re: Power supply build quality pictorial. part 2

Hmmm. This thing is just full of surprises! Is it acting similar to a VRM on a ATX motherboard? Stepping the voltage down in a very efficient and clean way? Seems like a much better method than linear conversion.

Linear regulation results in a much cleaner and more stable output than a buck converter - that is the one advantage of linear regulation. However, buck converters are much more efficient than linear regulators so they are ultimately preferable, especially for high wattage PSUs.

[momaka]

Re: Power supply build quality pictorial. part 2

Isn't that nice for a cheap PSU? By the way, when I picked this up on newegg, it was only $12.99 with free shipping. It was borderline theft for what you're getting

That is pretty amazing. Most PSUs on Newegg in that price range are... well, not so impressive to say the least.

Very interesting. I've never seen that either. How does this work? Is it connected to any of the IC's?

I didn't trace it (as it is a bit cramped and probably a bit over my head). But my best guess is, that feedback traffo has a winding on the primary that is in series with the main PS winding. Every time one of the transistors turns ON, the pulse also goes through that feedback traffo and is sent to the secondary side, where it probably connects to one of the comparator inputs on the LM339. But again, that is a best guess.

I noticed this too. However, check under the seal! Mine had a low amount of drying grease. I cleaned out the grease and put oil in there.

Noted, will do.

If you don't want to put more fans inside the PSU (which you probably couldn't) if you solder fan or molex headers to those extra fan pins, you can have them go out the back of the PSU. Those headers get a minimum of 5.75V, and slowly climb as the temperature rises. You could use regular 12V molex fans and have them not be unbearably loud without even having to do a wire mod.

Yeah, that's the same exact thoughts I had too when I saw the fan controller. I actually do have quite a few 3-pin male fan headers. The real question is, which PC will I use this with?

Pretty beefy primary. However, I think you got shafted here! Mine came with 820uF 200V Teapo LXK. Weird, since mine was made fairly closely to yours. The serial number was only 150 numbers away from yours.

I looked at my pictures and notes again.
I really don't know why I wrote 680 uF when the capacitors in there are 820 uF indeed .
I hate when my mind plays me tricks like this.

Very cool! So it uses the 2SK2645 FET for 5VSB switching, and the UC3842B for output and regulation?

UC3842B generates the PWM pulses that drive the 2SK2645 FET. 5VSB regulation is achieved by a 431 adjustable shunt on the secondary side (as with all other flyback topologies) that sends feedback to UC3842B controller on primary side via optocoupler.

You got jacked here, too! Mine came with two 20A ultra fasts in parallel.

If I understand it right from the datasheet, the BYQ28E are 20A FR rectifiers (10A per side... i.e. 10A per diode). So should be the same thing.

Thanks for sharing your test results. Yeah, efficiency shouldn't matter too much at such low loads.

No problem.
Reason I measure 5VSB efficiency is because when the PC is OFF or in Standby (all of my PCs are like that when not used actively), the 5VSB can generate quite a bit of heat if it is inefficient and kill its output caps on 5VSB. I consider 3-4 Watts without load to be acceptable. Any more, and it's a waste of power and generating too much heat. Can't remember which one of my PSUs was the worst, but its 5VSB drew something like 10W withOUT load.

[Wester547]

Re: ePower EP-350XP-24 (P022-E platform) (page 118)

Not sure if I will recap it yet either, since it is NEW!

Those Fuhjyyu capacitors aren't really new in the sense that one would think. In terms of shelf life, they are in fact over five years old, or well over 40,000 hours old. Just that amount of time in storage is enough time for the defective electrolyte in those to break down and for hydrogen gas to generate (or just for them to silently fail as hydrogen gas generation is just one byproduct of capacitors that have internal issues). Heat only accelerates the inevitable defects in the poor quality raw materials. That's exactly why the likes of KZG and KZJ bloat in storage. Of course, there's always a chance that they (the Fuhjyyus, though they have been known to fail in storage before) may be fine as in your case, but I don't really consider electrolytic capacitors that have sat for a long time to be new, especially those in need of reforming. Just my view.

And no, BYQ28E-200s are 10A ultrafasts, 5A per diode. You're thinking of BYV32E-200s, those are 20A ultrafasts. BYQ30E-200s are 16A ultrafasts.

[Behemot]

Re: Power supply build quality pictorial. part 2

That is pretty amazing. Most PSUs on Newegg in that price range are... well, not so impressive to say the least.

Yeah but but you can get PSUs like FSP400-60GLN for free with bloated caps. You need to recap this thing anyway…guess which one is better than.

[Pentium4]

Re: Power supply build quality pictorial. part 2

Yeah but but you can get PSUs like FSP400-60GLN for free with bloated caps. You need to recap this thing anyway…guess which one is better than.

Guess which one is more interesting though

[Behemot]

Re: Power supply build quality pictorial. part 2

Fortron.