Power supply build quality pictorial. part 2

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

Some notes i forgot to add to the main post due to time restrictions: The main control IC is a TL494L PWM controller with a TP3510 supervisor IC.
All of the wires are 20AWG, with an exception for the PS-ON wire which is 22AWG.
the two smaller transformers are roughly 30mm tall by 20mm wide.
Based off the date codes on components, this thing was made around early 2011.
I'm not a power supply expert, so i'm unsure what the full capability of this thing is, i will say however that the primary caps seem a bit undersized for 350w.
Anyways that's that and sorry for double posting.

The worst thing here is the output rectifier for 12V rail.

I mean, this thing was made in 2010, 10 years after the Pentium 4 cpus that were the first cpus drawing power from 12V rail.

It should be power heavy on 12V and light on the other rails, not the other way.

But the good news is that you could replace the part with a proper 30A schottky diode with V reverse of 60V or more

[momaka]

Re: ISO Model: ISO-400 350w PSU

Hello, I've been wanting to show this PSU for a while, but have not due to a mix of laziness and fear that i would once again make a fool of myself.

Laziness/procrastination, I can understand.
But making a fool of yourself? I mean, you're just posting about a PSU. Have no worries!

The main transformer's size is 36.5mm wide by 45mm tall (i'm guessing this makes it an ERL36).

Should be 35, as the sizes are more or less standard (28, 33, 35, 40, and etc.) Not that this is the only important parameter. Height matters as well. You can have a supposedly ERL35 traffo. But if it's very short, that means less core material and less space for windings as well, which can translate to a hotter-running traffo.

But the one in this ISO looks normal.

Now i'm not sure how capable this is for 350w of power

With the size of those primary caps and output toroid... eh, probably not. With 470 uF, 200V input caps, I'd say the limit is closer to 250-300 Watts tops. The output toroid seems kind of small too, so it probably won't allow for more either. If these two were bigger, the 13009 BJTs on the primary should allow it to do 350 Watts continuous and maybe 400W peak for short periods of time (under a minute.)

the Gun primary caps immediately reminded me of my Winsa PSU, which i suspected at the time was possibly OEMed by CWT (like this PSU), but didn't think CWT would lower their standards to such a level to make such a thing.

Yeah, that Winsa PSU kind of tries to look like a CWT with those green transformers. However, the PCB layout is not something I have seen come out of CWT. So hard to say.

I hope this post is any good

Yessir!
Thanks for sharing your PSU pics with us.

don't think CWT has anything to do with this PoS, likely some fake copy if anything

they make CM6800 based PSUs as the cheapest for way more than a decade now

Nah, Rukycon is correct - this is indeed a CWT PSU. It's just an extremely old platform from the early 2000's, known as the "ISO" line. The TL494 and TPS3510 ICs on the secondary side and the way they are arranged is one of the giveaways. The "Lineplay Pass" sticker on the primary caps is another one. I see that the Winsa PSU has that sticker as well. On a second look, that may be a CWT PSU as well.

Indeed hard to imagine why CWT would make such crap, but you never know. CWT still churns out some cheapo PSUs, so maybe this was an updated replacement for the ISO budget line? Or perhaps an even more stripped down budget PSU meant for 3rd world countries only??

With the large-scale manufacturing of today, though, it's still mind-boggling to try to understand why such POS PSUs are still made.

don't think CWT has anything to do with this PoS, likely some fake copy if anything

they make CM6800 based PSUs as the cheapest for way more than a decade now

Nah, Rukycon is correct - this is indeed a CWT PSU. It's just an extremely old platform from the early 2000's, known as the "ISO" line. The TL494 and TPS3510 ICs on the secondary side and the way they are arranged is one of the giveaways. The "Lineplay Pass" sticker on the primary caps is another one. I see that the Winsa PSU has that sticker as well. On a second look, that may be a CWT PSU as well.

Indeed hard to imagine why CWT would make such crap, but you never know. CWT still churns out some cheapo PSUs, so maybe this was an updated replacement for the ISO budget line? Or perhaps an even more stripped down budget PSU meant for 3rd world countries only??

With the large-scale manufacturing of today, though, it's still mind-boggling to try to understand why such POS PSUs are still made.

Protection? No, we don't have this here

I have noticed several times Linkworld psus with bulging caps and 12V rail going up to even 14V and the psu still working

This is the thing I don't like at all at Linkworld power supplies. But, maybe this is common in many cheap psus, can't tell for sure.

Well, under and over-voltage protections seem to always have been problematic for cheapo PSUs, especially ones based on half-bridge. But I remember reading some bargain basement PSU reviews a while back, and Linkworld PSUs - at least those with their own proprietary ICs - were among the few that would shut down when overloaded (OPP, likely?)

But the good news is that you could replace the part with a proper 30A schottky diode with V reverse of 60V or more

Yup.
But question after that is what will burn next if you try to pull that kind of power? The output toroid and 12V rail snubber already look like they were barely making it by the skin of their teeth. I don't imagine they could take much more.

[RukyCon]

Re: ISO Model: ISO-400 350w PSU

Laziness/procrastination, I can understand.
But making a fool of yourself? I mean, you're just posting about a PSU. Have no worries!

I tend to make super obvious mistakes if i'm left here for too long, something that a lot of people here must have seen first hand at some point, i try my best for it not to happen, but it still happens a lot.

Should be 35, as the sizes are more or less standard (28, 33, 35, 40, and etc.) Not that this is the only important parameter. Height matters as well. You can have a supposedly ERL35 traffo. But if it's very short, that means less core material and less space for windings as well, which can translate to a hotter-running traffo.

But the one in this ISO looks normal.

I wasn't too familiar with the transformer sizing and the different sizes, so i didn't know if ERL36 was a size.

With the size of those primary caps and output toroid... eh, probably not. With 470 uF, 200V input caps, I'd say the limit is closer to 250-300 Watts tops. The output toroid seems kind of small too, so it probably won't allow for more either. If these two were bigger, the 13009 BJTs on the primary should allow it to do 350 Watts continuous and maybe 400W peak for short periods of time (under a minute.)

I could at the very least put bigger caps on the primary side to improve hold up time, but thats assuming i have a use for it, which at the moment, i do not.

[momaka]

Jewel JS-12030-2E power adapter, take TWO

Remember this Jewel JS-12030-2E power adapter??

That CapXon KF may test okay for now but who knows how long it’s going to last, especially since it’s 12 years old now going by the datecode. CrapXon are still replace-on-sight for me, always have been and always will be.

Proper foreshadowing there!

After maybe about 1 year of occasional use, I plugged this power adapter one day and heard a sizzling / hissing noise and misty smoke pour out of it. I hadn't closed the case, of course, so taking it apart was as simple as pulling the top cover off… which revealed this:
https://www.badcaps.net/forum/attach...1&d=1602447954


Mmmmm… freshly-bulged/leaked CapXon KF. What a delightful smell!
You can also see droplets of electrolyte on the top shell in the upper-left corner – again, courtesy of that CapXon KF, 16V, 1000 uF cap on the output.

I desoldered the CapXon KF cap and did a few checks on the power adapter to make sure nothing else went bad. Everything checked out okay, so I gave it a more proper recap this time:


Basically, I moved the Elite EB, 16V, 1000 uF cap on the spot after the PI coil (where the freshly-bulged CapXon KF was just sitting) and installed a new Rubycon YXJ, 16V, 1000 uF cap in the spot before the PI coil. While at it, I also changed the startup cap (50V, 22 uF) with a Nichicon PW (or PM or PS - I can’t remember which) that I now had in stock.

After this, I plugged the adapter in and checked it with and without a load once again. Output voltage was good in both cases, so we’re back in business with this one.

The only CapXon cap left now is the primary input one. But I don’t have any 400V caps in stock in that size, so it’s going to have to wait. As such, I won’t bother with gluing / sealing the top cover again, because I have a hunch even that input CapXon KM cap will fail, despite primary-connected caps usually not failing that often - at least on SMPS without APFC. But this is CapXon, after all, so one can never know.

Anyone wanna hazard a guess when I’ll be posting back about this adapter once again to report the primary CapXon cap failing? Feel free to cue your votes below.

Oh, and what’s very ironic is I used this adapter quite often with my ESR / bad cap meter. So my cap meter’s power adapter had a bad cap. Imagine that.

[TechGeek]

Re: Power supply build quality pictorial. part 2

I give it roughly 6 months before the main filter expires. I'm also going to guess that when it does, it's going to fail shorted before you can unplug it and trip a breaker or blow a fuse, and it might take out the bridge rectifier in the process.

[momaka]

TRUMPower FSP300-60LG [P/N: TIPC-300FLX-UT]

With presidential elections right around the corner, today’s PSU post is going to be themed around that. More specifically, if our current president (Mr. Trump) could name his own brand of ATX power supplies, what do you think he’d choose?
.
.
If you guessed TRUMPower, you’d be right. GO MURICA!!!


No, that label is not a photoshop joke. This is a real legit power supply. And if the FSP300-LG in the model number is to hint at something, it’s that it is made by FSP. Moreover, those two specific stickers below the label (“Ball Bearing Fan” and “Active PFC”) are another tell-tale that this is an FSP/Sparkle unit, in case anyone didn’t look up the UL number already (E190414).

As for the PSU itself, this is it:


It’s one of those small (ITX?) form factor PSUs. It came from a mini server blade/computer from my previous job. There was just no way to check the PSU at the time and place where the server was, and we were swapping parts desperately to get up the server running as quick as possible again. So to eliminate as many components as possible from being troublesome again, the PSU was swapped on the spot. No one wanted to do anything with the old PSU (the one pictured here), so it was destined for recycling… until I intercepted.

As you can see, there isn’t much to the TRUMPower FPS300-60LG it in terms of cabling: only a 20-pin ATX connector, 4-pin 12V CPU power connector, a single Molex, and a floppy/berg - all 20-AWG size wires too, except the three 3.3V wires on the main 20-pin ATX connector (those are 18-AWG.) That’s it! Not sure why a 300 Watt PSU would have so few connections, but I suspect it was a custom order from FSP specifically designed for the blades we were using, because those are all of the connections they really need.

Moving onto the insides of the PSU… I can’t really show pictures of the primary and the secondary separately, like I usually do. It’s just way too cramped in there. Have a look:




This PSU is almost like a laptop power adapter, except a little more powerful (rated for 300 Watts.) I’m not sure at what temperature that is, though. A slightly older FSP from the older blades we had at work, was pretty much the same identical unit as this one (though, with more output connectors), yet only rated at 270 Watts (@ 40°C) / 300W @ 25°C ambient operation. So I can probably assume the same applies for this TRUMPower unit as well.

In any case, the power per unit volume of this PSU is still pretty amazing. Moreover, the PSU does actually work fine - I tested it eventually. That shouldn’t be too surprising, though, given the good quality components inside. In fact, note that all of the capacitors are Japanese brands (mostly United Chemicon and Rubycon from what I see.) Even the primary cap is Japanese (United Chemicon KMR), which is good to see in a PSU with APFC. And to show that FSP isn’t joking, they used a 450V -rated cap here.

On that note, in order to get this kind of high power output with good efficiency, the TRUMPower PSU uses a monorail output design - i.e. main PS produces only a 12V DC output (through two 60 Amp Schottky rectifiers?) The 3.3V and 5V rails are generated each by its own buck converter from the 12V rail. Both the 3.3V and 5V rails use polymer caps for their filters (Fujitsu FPCAP, IIRC.)

And that more or less sums it up about this PSU. Everything that needs to be there is there. See below for a slightly more detailed parts list.

Input EMI/RFI filter: two X2-class caps, four Y2-class caps, two CM chokes
Input Protection: 8 Amp or 10 Amp fuse; SCK 1R55 NTC thermistor; 1x MOV
Input wiring: 20-AWG, 300V
APFC: 1x 11N60NT(?) MOSFET + 1x STTH8R06FP diode
Input cap: Nippon Chemicon KMR series, 450V, 270 uF, 30x22 mm (dia. x h.)
Main PS: 2x FCPF 11N60 MOSFETs (2-transistor forward design), driven by a CM6800(?) IC
ICs: TNY260PN (5VSB), CM6800 (APFC + main PS), APW7159 (3.3V & 5V buck driver IC), and not sure about protection IC, because I can’t see where it is buried in there.
Transformers: 29 mm core for main PS and 19 mm core for 5VSB
Cooling fan: Protechnic Electric 40 x 40 x 15 mm
Output Rectifiers: two 60A60 (in parallel??) for the 12V rail, can't read info on anything else, though.

Output caps:
1x Rubycon ZLH, 16V, 2200 uF, 10 mm dia. (for 5VSB?)
2x Rubycon ZLH, 16V, 1000 uF, 8 mm dia. (for 12V rail?)
1x United Chemicon KY, 16V, 1000 uF, 8 mm dia. (for??)
1x United Chemicon KZH, 10V, 1000 uF, 8 mm dia. (for 5VSB?)
1x UCC PSF polymer, 16V, 470 uF, 8 mm dia. + 3x other polies (can’t read info, but they look like Fujitsu FPCAPs??)

And that’s all Mr. Trump packed in this one.
Now if only our current politics looked this clean and tidy as this PSU…

[Dan81]

Re: Power supply build quality pictorial. part 2

Looking at that form factor, I think it's TFX as I might have a similar 250W unit (except it actually says FSP on it.) out of a NEC unit that had a burnt mobo.

[Xan03]

Re: Power supply build quality pictorial. part 2

I have a little selection of old AT and ATX PSUs around that I'll try to post here over time, some of them I intermittently use with vintage hardware. I'll start with this Astec AT unit from early 1995.




All Japanese electrolytics, which is kind of rare for a PC PSU of the time from all I've seen, so that's good... but wait, what are those yellow things on the input?




Yep, it's those RIFA Y-caps, and they sure are cracked - an apparently quite well-known issue that I've coincidentally recently learned about. I've seen on pics how those can fail so guess I'll be looking into replacing them sometime, as this is one unit that does see some use. I suspect Astec stopped using them very shortly after this unit was built.

Electrolytic caps as far as I could identify:

2x Nichicon LQ 470uF/200V
NCC SME 47uF/25V, probably 2x
2x Nichicon PF 2200uF/16V
Nichicon PF 1500uF/10V
Nichicon PF 220uF/10V
one unidentified Nichicon PF.

I'm aware the caps do fall under the quaternary ammonium salt category, though so far there is no visible leakage.

[Xan03]

Re: Power supply build quality pictorial. part 2

Before I post some more older PSUs, here's something newer first: a bequiet Straight Power E9 400W that was in use in my main PC from 2012 to 2020. I won't bother with a capacitor list for this as it's probably in some review on the net already.

Why did I stop using it? Well, one day I was cleaning my PC and from what I recall improperly put the graphics card back in the slot without noticing it, and when I went to turn the PC back on it tripped my fuse box; reset the fuse and tried it again (foolishly, perhaps), same thing, and that's when this thing bit the dust. Fortunately no hardware damage otherwise.







Just took it apart for the first time ever, and lo and behold, the primary Teapo LH 220uF/420V (isn't that really undersized for 400W?) is bulging badly. First time seeing a primary cap bulge... That's pretty ironic as during its operation I periodically checked the SCs through the grill and they all look visually fine even now for whatever that's worth. Well, I have already learned from this forum that APFC can thrash a primary cap so it's not a complete surprise though. Mains voltage in my house is high too, btw; measured up to 243VAC when it's supposed to be 230. So that may also have contributed.

I don't know what component exactly caused the fuse tripping, some safety part (fuse?) inside the PSU maybe, but I never bothered fixing it as a group regulated unit with all Teapo isn't that exciting for newer hardware. Actually looking at this now, it's only for the better I replaced it. And incidentally, since the pics of the bulging cap turned out worse than I thought I tried opening it again, but one of those tiny hexagonal screws instantly wore out and won't turn anymore... so in the junk this goes I guess.

This was replaced with a Straight Power 11 450W from the same company, a DC-DC unit with all Japanese caps and very quiet like this one. Also noticed that the voltage sags on load I was getting on 12V with the old unit are gone now. I really wanted to go Seasonic originally but was put off by their semi-passive designs and numerous stories about grinding fans.

Edit: just noticed the other thread about fried PSUs where this thing would rather belong, but can't move the post now.

[momaka]

Re: Power supply build quality pictorial. part 2

I'll start with this Astec AT unit from early 1995.
...
All Japanese electrolytics, which is kind of rare for a PC PSU of the time from all I've seen, so that's good...

Yes, Astec was one of the few (if not the only) PSU manufacturer that used Japanese caps only in their PSUs. Even much later in the ATX era when the likes of Delta and LiteOn started switching to Taiwanese stuff, Astec didn't change its way... and perhaps that's why they stopped making PC PSUs?? If I had to guess, they most likely just weren't as price-competitive as the other manufacturers that used cheaper caps.

Aside from those RIFA caps, you can always rely on an Astec PSU to work properly. I've always liked their designs, even if their fan-controlled PSUs were a little on the loud side.

Well, one day I was cleaning my PC and from what I recall improperly put the graphics card back in the slot without noticing it, and when I went to turn the PC back on it tripped my fuse box; reset the fuse and tried it again (foolishly, perhaps), same thing, and that's when this thing bit the dust. Fortunately no hardware damage otherwise.

I suspect the GPU not inserted all the way was NOT the issue for the PSU to blow. More than likely, it was just a coincidence. APFC, as you've mentioned correctly, does put a lot more stress on the primary caps, and so they fail quite often, actually - especially non-Japanese brands and GP/85C -rated ones. It's actually nice to see the manufacturer use 420V caps here instead of cheaper 400V units. But still, like you noted, probably the small capacity was the other reason why it blew up. After all, smaller capacity means smaller cap size, which means the cap can handle less ripple current for a given load... and ultimately that's what gets them. If you've ever looked at any high-end 500-600 Watt Silverstone PSU inside, you'll probably notice that the primary caps are "huge" for the rated power output of the PSU, compared to other cheaper PSUs that are rated for much more Watts. And that's why Silverstone often offers 10 year warranty - they know they have good caps and a good design in the PSU, so they know it will likely last. But then again, most Silverstone units I've seen cost 2 to 3x more per Watt... so in the end, you only get what you pay for (and that's if you're lucky! )

Mains voltage in my house is high too, btw; measured up to 243VAC when it's supposed to be 230. So that may also have contributed.

Actually, that probably helped the cap a little, since the APFC has to work less to boost the voltage... but in the end, it's the high-frequency of the APFC combined with the power draw of the system that determines how much stress that primary cap will see.

I don't know what component exactly caused the fuse tripping, some safety part (fuse?) inside the PSU maybe,

Well, with that primary Teapo cap bulging like that, it probably has gone open-circuit... which means the energy stored in the APFC boost coil will have nowhere to go after the APFC MOSFETs "pump" it with current. So the usual casualties are either the APFC MOSFETs, NTC thermistor, or primary switchers... and sometimes any combination of these, including the 5VSB switch/IC, because the energy from the APFC coil will generate an extremely high voltage that has nowhere to go (since the primary cap is open) which then over-volts the primary side and kills whatever component is the weakest.

so in the junk this goes I guess.

If you fix electronics, it may be a good idea to save it for the parts. In fact, that Astec with the bad RIFA caps... just take the X2 and Y2 caps from this units to fix that one - simple as that!
There are also other useful components you can scavenge over time... or just put it on your local classifieds for free (you never know when someone who likes to thinker and fix stuff... like me ... might just be happy to take it to play with )

This was replaced with a Straight Power 11 450W from the same company, a DC-DC unit with all Japanese caps and very quiet like this one.

But it probably still has APFC, so check the size and ratings of the primary cap inside that one. Even if it's a Japanese brand, it too will likely eventually fail like this one if the primary cap is undersized.

I've also started experimentally adding metal film / polypropylene caps in parallel with the primary cap - typically 1 to 4.7 uF - to help with the high-frequency ripple, hopefully. But I have no idea if that will actually help or not. I guess time might tell... or maybe not, since the few APFC units I have don't get much use. Whenever possible, I use oldschool PSUs with classic 2-cap setup on the primary with no PFC of just PPFC. These just last for AGES, even with the most dubious and obscure cap brands I've seen. Fuhjyyu? No problems - still good!

[Xan03]

Re: Power supply build quality pictorial. part 2

Yeah, the APFC+undersized primary cap combination definitely smells like a planned obsolescence mechanism; you can even see by the radial mark on the pictures how there's room for a much bigger cap diameter there. In my current PSU, the primary cap there is a blue sleeved Hitachi 105°C one - not 100% sure on the specs but I seem to recall it's 420uF/450V or thereabouts. Would still be not fantastic but nevertheless an upgrade over this. There's no reviews of the 450W version unfortunately, but the 550W has a 470uF/420V Nichicon GN.

In hindsight it would have made sense to go for a higher watt version instead of saving a bit, but I had never really read about the primary cap being a failure point before; contrarily reviews I've seen always seemed to suggest that it's not highly stressed at all, compared to the secondary ones. But in any case, only time will tell, but I'll try to look at it once a year or so.

About keeping the dead PSUs around for parts, you're right, there is no harm in chucking them right back into the junk parts box. But reusing the safety caps though, I'm not sure about that... they still have eight years of use on them and this link suggests that a failed X-class cap can trip a fuse/circuit breaker, which is exactly what happened in my case. Now, that may well not be the cause here, seeing the visible failure on the primary cap, but still... the main argument for buying new parts though would be that these seem to be fairly cheap, around 1€ per piece, so I could just order them next time alongside whatever else I need, and while I'm at it match capacitance to the original parts. That PSU still works fine, to be clear - I just don't know how much is left in these RIFA caps anymore, so it would be a semi-preventative replacement I suppose.

I'll post another Astec unit soon, so stay tuned for that

[Xan03]

Re: Power supply build quality pictorial. part 2

Yes, Astec was one of the few (if not the only) PSU manufacturer that used Japanese caps only in their PSUs. Even much later in the ATX era when the likes of Delta and LiteOn started switching to Taiwanese stuff, Astec didn't change its way... and perhaps that's why they stopped making PC PSUs?? If I had to guess, they most likely just weren't as price-competitive as the other manufacturers that used cheaper caps.

Well, I may or may not have just the right tool at hand to prove you wrong... enter the Astec AA21790 200W ATX PSU from 2000






Just kidding of course, but it does seem like they had at least a little stint with cheaping out a bit on capacitors... that's especially strange here as this particular unit feels very well built with some serious weight to it - most of that will be from what looks to be the passive PFC here, but even the outer metal casing looks to be sturdier than other PSUs from the time, and soldering quality seems quite good too. Interesting setup here with the two daughterboards. Can anyone spot the typo on the silkscreen?

Jamicon perhaps was far from the worst choice, and I've seen similar looking caps of that make in a Sony Trinitron TV from the same time, but still kind of surprising... I'm pretty sure they are original too - I've added a pic of one of them with the datecode visible, if anybody wants to verify.

Problem with this particular unit is it was badly packed by the seller and the PCB got cracked during shipment, especially in one corner... I've actually still plugged it into a PSU tester once and it did come up fine, but yeah, no point in trying to use it with that kind of damage. A bit of a shame seeing how remarkably clean it is, too.

Electrolytic caps I could spot:

2x Nichicon LQ 470uF/200V
NCC LXV 2200uF
Jamicon WG 2200uF/10V
4x Jamicon WG 2200uF/6.3V
2x Jamicon WG 470uF/16V
NCC KMA 33uF/25V, probably 2x

[TechGeek]

Re: Power supply build quality pictorial. part 2

Check the bottom of the PCB where it's cracked. If there's not any traces running through there, it should be 100% safe to use.

[Xan03]

Re: Power supply build quality pictorial. part 2

Here's a classic AT Seasonic from 1995 - a SSG-200G. Another cheap pickup that is still untested as of now.





Electrolytics, as far as I could identify:

2x TOSIN LG2 330uF/200V
9x Rubycon MH 1uF-10uF/50V, biggest one unidentified
Rubycon CE W 470uF/16V
CEHTR 105° 1000uF/16V
PCE-TUR TU WV (?) 1000uF/10V
PCE-TUR 2200uF/10V

That's an interesting mix to say the least... I think Seasonic is one of those makers that is known for using only Japanese caps now, seems that wasn't the case 25 years ago though. Also note the case sticker advertising a temperature controlled fan.

[momaka]

Re: Power supply build quality pictorial. part 2

In hindsight it would have made sense to go for a higher watt version instead of saving a bit, but I had never really read about the primary cap being a failure point before; contrarily reviews I've seen always seemed to suggest that it's not highly stressed at all, compared to the secondary ones.

Indeed, it is not stressed as much as the secondary caps, but still quite a bit more than a regular PPFC or no-PFC PSU with the typical voltage doubler / series 200V caps. Many reviews also don't score a PSU so much based on the brand of the caps either - and that really is the one thing that matters a lot. Crap caps = failure down the road, especially if the PSU is loaded hard or uses passive or semi-passive cooling.

But reusing the safety caps though, I'm not sure about that... they still have eight years of use on them and this link suggests that a failed X-class cap can trip a fuse/circuit breaker, which is exactly what happened in my case.

Yes, X2 caps can fail short, but they can also fail open. There are actually a number of things in that article that are *not* true. Biggest one I saw was this:

"If a Class-Y capacitor, also known as the "line to ground capacitor" or "the line bypass capacitor"—the capacitor placed between line and ground—fails short, this could lead to a fatal electric shock due to the loss of the ground connection."

^ This is completely false.
Ground connection is NOT dependent on the Y caps. It is either there or it's not. For devices that have exposed metal parts connected to their circuit's ground, a grounded plug *must* be used and the house wiring should also have a functional ground.
That being said... on grounded PSUs, ground is always connected to earth in your house, so no matter what live-connected component fails shorted to ground, it will *not* create a shock hazard, because the current will go through the ground conductor and trip the breaker. Only exception is if a grounded PSU has a faulty ground *and* one of the Y2-class cap between Live and Ground happens to short out... which it shouldn't, because Y2 caps are designed to fail open-circuit. But in any case, this in an unlikely scenario and moreover, a faulty ground in your house is always a potential for shock. So the argument about Y2 caps failing shorted is unimportant.

Second) X-class caps don't always fail short-circuit. In fact, X-class caps are nothing but specialized metal film capacitors. As such, like most metal film caps, zapping them with over-voltage should actually cause little areas between their plates to short out and burn out, thus sustaining no more damage than loosing capacity slightly. Over time, a very abused film cap (due to voltage spikes) will loose its capacitance. Of course, this does depend on the construction of the X-cap, and some will go short-circuit rather than loose capacitance (and go O/C eventually.) For this, manufacturer's datasheet of the cap should be consulted, as it will tell what is the expected failure mode (though not all datasheets will mention this.)

In any case, X and Y caps are safe to reuse if their capacitance is still within spec.

As for your PSU tripping the breaker - that likely isn't from an X cap gone short. Rather, it's probably either the primary cap is shorted (see this thread) or primary cap is O/C, causing excess energy from APFC coil to make either APFC or main switch MOSFETs stuck On and tripping your breaker from overcurrent (but why the PSU fuse doesn't blow, that's a good question... though I've seen it happen with more sensitive house breakers - they will trip before the fuse goes out, which isn't too bad of a thing )

That PSU still works fine, to be clear - I just don't know how much is left in these RIFA caps anymore, so it would be a semi-preventative replacement I suppose.

Oh yeah, those RIFA caps are notorious for going S/C, so it is better to replace them as soon as possible.

Well, I may or may not have just the right tool at hand to prove you wrong... enter the Astec AA21790 200W ATX PSU from 2000
...

Well now, I'll be darned! That's the first Astec I've seen with so many non-Japanese caps. Perhaps they had them build different for different markets?
In any case, at least they used Jamicon and not something more inferior like CapXon... but I'm still disappointed.

Can anyone spot the typo on the silkscreen?

"(C)2000 Astec Cutom Power"
Oops!

I'll take that over perfect spelling and a crappy PSU, though. Then again, with really cheapo PSUs, you get neither good spelling nor good quality PSU. But hey - it's got an LED fan and cool paint job for $15. That's all it needs to sell well.

Jamicon perhaps was far from the worst choice, and I've seen similar looking caps of that make in a Sony Trinitron TV from the same time, but still kind of surprising...

Interesting. That's another first for me. I mean, I've seen Sony Trinitron TVs and monitors use the occasional Jamicon cap here and there on usually non-important stuff... but again, it was mostly Rubycon and Nichicon everywhere. So again, I wonder if perhaps there were different built specifications for different regions where these were sold (in which case, you being in Germany... I'd imagine they' have the better built stuff. Then again, Europe has a lot of different regions, so something could have been imported from elsewhere.)

Problem with this particular unit is it was badly packed by the seller and the PCB got cracked during shipment, especially in one corner... I've actually still plugged it into a PSU tester once and it did come up fine, but yeah, no point in trying to use it with that kind of damage. A bit of a shame seeing how remarkably clean it is, too.

Eh, that's nothing. Could be repaired easily if you have the time. Shouldn't take more than a bit of RTV and mending traces with solder and jumpers. I've fixed a lot worse, like this Sony CRT TV PSU board... And that one wasn't even the worst. The worst was a toy that someone brought to us to repair way back when I was working in one repair shop. The toy was worthless really, but the mother said her kid really liked it and she wouldn't mind paying more, even if she could find a used one somehow online. Anyways, not sure what exactly what had happened to that toy, but the PCB inside it - about 10x10 cm worth of electrical tracks and some components for a moving train thingy on top - was literally in small bits and pieces the size of coins. It took me about an hour of scraping solder mask, adding jumpers and flowing with solder. But I got it to work and the kid was happy. We even gave it the usual 30-day warranty that we did to other stuff, and it never came back, so we can assume it lasted at least that long. I actually went a little crazy with the hot glue on that one and probably made the PCB tougher than it was originally in some spots, lol.

Check the bottom of the PCB where it's cracked. If there's not any traces running through there, it should be 100% safe to use.

+1
Though it may be a good idea to epoxy/RTV the PCB and make sure it is still held sturdy onto the case, because a sagging/moving PCB can short to the case, especially if there is no plastic isolator sheet.

I think Seasonic is one of those makers that is known for using only Japanese caps now, seems that wasn't the case 25 years ago though.

From what I've seen, they use whatever they are told to (in case of building PSUs for OEMs, such as some Antec-built PSUs, for example) or what they budget their PSU for.

Again, at least they used semi-decent crap cap brands: looks like CEC Holdings, for the most part (i.e PCE-TU/TUR/TUV). These range from GP to basic entry-level low-ESR range, respectively (IIRC). Given the slow switching frequencies of those H-bridge topology PSUs, that didn't really stress those caps too much, especially with the low load they were usually tasked with (let's be honest, old PCs didn't use much power or produce too much heat. )

[Xan03]

Re: Power supply build quality pictorial. part 2

"(C)2000 Astec Cutom Power"
Oops!

Yes, good eye... especially considering the blurry pictures

Interesting. That's another first for me. I mean, I've seen Sony Trinitron TVs and monitors use the occasional Jamicon cap here and there on usually non-important stuff... but again, it was mostly Rubycon and Nichicon everywhere.

Same, and actually, nevermind my statement about Jamicon in Trinitrons - I think I got that confused with a Philips TV I had once checked.

Eh, that's nothing. Could be repaired easily if you have the time. Shouldn't take more than a bit of RTV and mending traces with solder and jumpers. I've fixed a lot worse, like this Sony CRT TV PSU board...

Nice repair (as for the toy as well, btw, which really sounds like solving a jigsaw puzzle in a weird way), at a glance I can barely even make out where that PCB was cracked... and yeah, I'm well aware the PSU is totally fixable with the proper time/skills; I might attempt that myself one day.

[momaka]

HiPro HP-235NLXAK REV-01 235 Watt ATX PSU

Since we are still on the topic of old AT and ATX PSUs, I suppose I can continue with one unit I haven't posted before: a HiPro HP-235NLXAK 235 Watt ATX PSU. This one is from a socket 7 “CyberMax” PC with an AMD K6-2.


^ Classic thick steel case with that oldschool -styled stamped fan exhaust. You can see the case even has a cutout for AC power switch wires (if an AT PSU was fitted inside.)
Wiring is standard 20-pin ATX connector, 4 Molex drive connectors, and 2x floppy drive connectors. Also, note the big sticker with all those QC ink stamps - I'm pretty certain this was properly QC checked at the factory.

Next, the label…

^ Classic 5V-heavy design, with 25 Amps on the 5V rail and 14 on the 3V3. The 12V rail is not too shabby, though - rated for a whopping 8 Amps. We also see HiPro's UL number: E143709… among other safety marks.

Now let's see it inside.

Look at those neatly tied cables! And of course, we see more QC stickers and a Yate Loon fan. Fuse is placed in a holder and AC input wires are not soldered but rather have a connector (a rare sight these day for PC PSUs. )

Two more:
https://www.badcaps.net/forum/attach...1&d=1612922672
https://www.badcaps.net/forum/attach...1&d=1612922672
Apart from the crappy tan glue (that goes conductive with heat) spilled here and there, overall PCB layout is well-done for a PSU from its time. Design is half-bridge group-regulated, of course, with the 3.3V rail being mag-amp regulated. Both output inductor cores are made from -26 core material, but this was standard back then (and OK, since these half-bridge PSUs use relatively low switching frequency.) Caps are *not* Japanese - “Tosin” -brand for the two 200V 470 uF primaries, and Su'scon just about everywhere else. Despite being non-Japanese brands, IMO these are far from the worst. In fact, I haven't seen Su'scon fail as much in ATX PSUs as other (supposedly equivalent quality) brands. Some of you may have also spotted that darkened spot on the PCB next to a big resistor – I believe that is part of the 3.3V rail mag-amp circuit, along with the TO-92L transistor next to it. Not sure why it's burned like that, as I haven't tested the PSU or investigated if there's anything wrong (I got this computer from a repair shop that was trying to clean up space.) However, I do suspect the PSU's cooling may have contributed to that. If you look back at the first two pictures of the case, you may notice that the air intake vents on the PSU are mostly on the top, and there are very few on the back. It's quite possible that this resistor is just sitting on a hot spot. When I get around to cleaning that PC, I'll check it out though. It should make for a great classic Windows 3.1 or 95 (or maybe 98?) PC.

Anyways, let's look at the 5VSB circuit in more details:
https://www.badcaps.net/forum/attach...1&d=1612922672
It uses a 2-transistor self-oscillating design with feedback (opto is a little dusty and hard to see, but it's there) – nothing out of the ordinary for that era of PSUs. It probably could do more than the 1 Amp the label claims with a bigger output diode and caps (and maybe a heatsink on that primary BJT.) In the above picture, you can also see there is a 4th tiny transformer right next to the main one. It has only a single turn on the primary but a lot more on the secondary. I imagine that's used for some kind of a current or power-sensing protection?

Now let's take a look at some secondary-side parts…
https://www.badcaps.net/forum/attach...1&d=1612922672
https://www.badcaps.net/forum/attach...1&d=1612922672
5V rail is generated from a 30 Amp, 40V Schottky rectifier in TO-247 case – this should be enough for the 25 Amp rating on the label (with maybe even some room to spare), given that this is a half-bridge design. As for the caps – they are a bit crammed in there under the wires, but still plenty of space for them (some caps are 12.5 mm diameter.) We also see the PWM chip and protections are all implemented on a daughter board with a TL494 IC and another chip (that I couldn't identify, due to sticker on top… but likely a LM339, if I had to guess.)

Speaking of the daughterboard, here are two more pictures:
https://www.badcaps.net/forum/attach...1&d=1612922688
https://www.badcaps.net/forum/attach...1&d=1612922688

And finally, the solder side of the PCB:

We see nice clean HiPro soldering, as usual. Note that most component leads on the primary are bent away from neighboring traces (suggesting a PSU assembled by competent workers that weren't probably as rushed as in some basement sweat shop.)

~~~~~~~~
Component summary…

Primary Side

* Input filtering: 2x MKP 0.47 uF X2-class caps, 3x 4.7 nF Y2-class caps, 1 big + 1 smaller CM chokes
* F6A 250V glass fuse, 13SP 005 NTC thermistor, 2x MOVs (across primary caps)
* PBU605 bridge rectifier; input wires rated for 600V, 18 AWG, by Wonderful-D
* input caps: 2x Tosin LG2 200V, 470 uF, 85°C, 22 x 36 mm (ø x h)
* 2x SEC E13009 BJTs for main PS (half-bridge), 1x 2SC4020 BJT for 5VSB (flyback)
* 2x Su'scon SL, 50V, 1 uF, 5 x 11 mm caps for main PS BJT drive
* main PS snubber network: RC-type with 2x 100-Ohm, 2W resistors in parallel + 1 nF, 1 kV ceramic disk cap
* transformer cores: 35 mm for main PS, 16 mm for 5VSB, 16 mm for BJT drive, and a tiny transformer for OPP detection (?)

ICs
* TL494 (main PS PWM controller) + LM339 (for protections… OVP?/UVP?/SCP)
* KA393 (for??)
* 7905 regulator (-5V rail output regulator)

Secondary Side
3.3 V rail (mag-amp regulated):
* STPS2045CT (TO-220) Schottky rectifier
* 1x Su'scon SD, 10V, 3300 uF, 12.5 x 25 mm before PI coil
* 1x Su'scon SD, 10V, 2200 uF, 10 x 25 mm after PI coil
* PI coil: 6 turns, 6 mm core, 18 AWG

5 V rail:
* SBL3040PT (TO-247) Schottky rectifier
* same capacitor arrangement as the 3.3V rail
* PI coil: 8 turns, 8 mm core, 14 AWG

12 V rail:
* ON semi C25M-02C (TO-220 isolated tab) 200V, 10 Amp FR rectifier
* 1x Su'scon SD, 16V, 2200 uF, 10 x 30 mm before PI coil
* 1x Su'scon SD, 16V, 470 uF, 8 x 13 mm after PI coil
* PI coil: 14 turns, 6 mm core, 18 AWG

-12 V rail:
* 2x 1-1.5 Amp diodes as rectifiers
* 1x G-Luxon GR, 16V, 470 uF, 8 x 13 mm after PI coil

-5 V rail:
* 7905 linear regulator for output (-12V rail as input)
* 1x Su'scon SK, 16V, 220 uF, 6.3 x 13 mm on 7905 reg. output

5 VSB rail:
* SB___ 1.5-2 Amp Schottky diode as a rectifier
* 1x G-Luxon GR, 10V, 470 uF, 8 x 11 mm before PI coil (in a 10 mm cap spot)
* 1x Su'scon SL, 10V, 470 uF, 8 x 11 mm after PI coil
* PI coil: 15 turns, 5 mm core, 20 AWG

Aux. Sec. (PWM controller power):
* 1x Su'scon SL, 50V, 47 uF, 8 x 11 mm
Output wires: 300V, 18 AWG, (except for floppy), Unicab brand
Connectors: 20-pin ATX, 4x Molex drive, 2x floppy drive
Fan: Yate-Loon D80SH-12 (12V, 0.19 Amp), running straight from 12V rail

Other: burned resistor R129 (8.8 Ohms), used for 3.3V rail mag-amp? .. with 2SA928 (Q9) TO-92L BJT

… that's all folks.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Built in early 1999, so a 1998 or earlier design. The 1A 5VSB was ordinary in that time frame. It would not surprise me if Hipro designed in a little margin in the 5VSB circuit to allow a higher current (soon after 5VSB current ratings moved up to 2A-3A).

I think your guess about the little transformer with a single-turn primary is correct. It's sensing the primary current.

Is Iskra, the manufacturer of the polypropylene X-caps, still around? Or did the break-up of Yugoslavia and later armed conflict kill of the company?

A TL494-based half bridge design, while not spectacular in performance, was quite adequate for the ATX spec of the time. The switch frequency probably would have been around 30KHz and venerable -26 material would have been fine. The bridge rectifier and 470uF input caps are pretty conservative, good for longer life and performance. Similarly, the MKE13009 clones are a pretty conservative choice. The heatsinks look OK for the rated power. Hipro wanted repeat business.

[Xan03]

Re: Power supply build quality pictorial. part 2

About TOSIN, on that Japanese site (http://capacitor.web.fc2.com/t.html#tosin) it seems to be mentioned that they were in fact a Japanese maker in the 90s that cooperated with Fuhjyyu.

[momaka]

Re: Power supply build quality pictorial. part 2

Is Iskra, the manufacturer of the polypropylene X-caps, still around? Or did the break-up of Yugoslavia and later armed conflict kill of the company?

Is that what brand those "MKP" capacitors are in this HiPro??

According to this website...
https://www.iskra.eu/en/Capacitor/
... looks like they are still being made (now in Slovenia, according to the ".si" domain, if you click on the other flag... though maybe Germany and Macedonia too, based on the other two flags on that page??)

In any case, it would be funny to think the X caps in that HiPro, made in the former Yugoslavia, were cheaper to buy and send over to China (where this HiPro was made, according to the label), then send it all the way back to the US for sale. LOL!

Then again... and probably on a tangent here... I saw hazelnuts in the grocery store once here, where it said "Product of Turkey", but then it said packaged in Thailand or India or China, and then finally imported in the US. Makes sense, right.

Hipro wanted repeat business.

And probably why they are still around.

About TOSIN, on that Japanese site (http://capacitor.web.fc2.com/t.html#tosin) it seems to be mentioned that they were in fact a Japanese maker in the 90s that cooperated with Fuhjyyu.

Well, Japanese or not... any association with Fuhjyyu kind of blows their credibility as a reliable brand, IMO.
Of course, these being on the primary, I think they will be fine for many many years to come.
Heck, I have several (non-APFC) PSUs with Fuhjyyu primary caps, and all of those are still doing OK.