Power supply build quality pictorial. part 2

[razvanu]

Hi everyone I have 5 PSUs for spare parts, but when they do power on.
Which of them it's a good to keep ?

[momaka]

Originally Posted by razvanu

Hi everyone I have 5 PSUs for spare parts, but when they do power on.
Which of them it's a good to keep ?

Not enough pictures to show really... but here's my 2 cents anyways....

Keep:
https://www.badcaps.net/forum/attach...3&d=1647766851

Probably keep (can't see caps and other components to tell more certainly):
https://www.badcaps.net/forum/attach...5&d=1647766851

Keep... but add missing input filtering and maybe new caps on output:
https://www.badcaps.net/forum/attach...4&d=1647766851

[RukyCon]

So i finally got some replacement caps for this thing, so here is the recapping, plus the new cables for this thing.

5VSB - Teapo SH 1000uf 16, Teapo SEK 470uf 16v ⇒ Rubycon YXJ 1000uf 16v, NCC KYB 470uf 16v.
12V - Teapo SEK 10uf 50v, Teapo SC 2200uf 16v -> Rubycon YXG 10uf 50v, NCC KZE 2200uf 16v.
5v - 2x Teapo SC 3300uf 10v -> 2x Panasonic FS 2700uf 10v.
3.3v - 2x Teapo SC 3300uf 6.3 -> 2x Panasonic FS 2700uf 10v.
-12v - Teapo SEK 470uf 16v -> Rubycon ZLH 680uf 16v.
-5V - Teapo SEK 10uf 50v -> Rubycon YXG 10uf 50v.
IMG_20220415_224724[1].jpgIMG_20220414_134918[1].jpg
I also replaced pretty much all the smaller caps on both the primary and secondary side (forgot to document them all), and out of all the Teapo caps i replaced, only about 2 of them were still what i'd consider to be good, most of the smaller ones had high ESR, and the 3300uf ones had high leakage.

And lastly, the new cables i attached came from a couple of damaged PSUs, and the cables are as follows:
20-pin ATX cable (18AWG).
2x Molex cable (18AWG)
4-pin 12v motherboard cable (20AWG)
1x Molux plus 2x SATA cable (20AWG)
IMG_20220415_224907[1].jpg
If it doesn't seem like a lot, then just know that it was all what i had lying around at the time, and i could add more if needed, but for right now, it should do.

[RukyCon]

Forgot to add one last thing, I forgot where i had put one of the two fans that were in this PSU, so i substituted it with one of these Sanyo Denki fans i had gotten from a surplus store a long time ago.

[Dan81]

Originally Posted by momaka

Keep... but add missing input filtering and maybe new caps on output:
https://www.badcaps.net/forum/attach...4&d=1647766851

If I were him I'd replace the fuse as well. No way in hell that tiny fuse is gonna do anywhere past 200W or so without going nuclear.

[RukyCon]

Well, since my last project isn't working out how i hoped it would, I thought i would take a look at another of my PSUs that needs some repairs. In this case, an Ultra V-Series 500w PSU i found for $10 at a Savers thrift store.

Now unlike every other used PSU i find, this one still had its retail box despite being used, and came with just about everything minus the mains power cable.
IMG_20220420_172901[1].jpg
Now another thing about this PSU was, the case had been opened up, and the fan was loose inside the box, but it did not appear the last owner had done anything to the board itself, because if they had, then this thing probably wouldn't have been worth trying to repair.

Anyway, here is the PSU label:
IMG_20220416_185842[1].jpg
Not much to say here, but judging by the UL number, this PSU appears to have been made by Wintech.

Board Overview:
IMG_20220415_195822[1].jpg
IMG_20220415_195848[1].jpg
So a few things are clearly off here, the most obvious thing being the secondary caps are nearly all dead, but also, it appears something had been spilled onto the board, and had been allowed to pool up around the lower-right corner of the board. The heatsinks appear to be decently sized. And on the bottom of the board, there is a model number on it of WIN-450WP. Now it seems like this PSU board may have originally been designed for a 450w PSU, so it seems a bit sketchy to push the board to 500w, but that might just be me.

Primary Side:
IMG_20220415_195913[1].jpg
So for input filtering, we have three .22uf X2 capacitors split between the main PS board and a separate filtering board, two common mode chokes, and two sets of 2 4.7nf Y2 capacitors. The rectifier is a RS805 (8A, 600V), and the smoothing capacitors are a pair of JEE 820uf 200v caps, both of which test at roughly 760uf. The main switchers are a pair of 2SC3320 transistors in what looks like a dual-foward configuration, and the 5VS/Aux switcher is a C3866 transistor in a two-transistor configuration. The main transformer is roughly 40mm x 40mm.

Secondary Side:
IMG_20220415_200315[1].jpg
IMG_20220415_200223[1].jpg
The filtering for each rail is as follows:
5VSB - SB540 (5A, 40v) Schottky Diode, 2x Canicon 680uf 16v with a PI coil between them.
12V - 2x F12C20C (12A, 20V) Fast Recovery Rectifiers, Su'scon SD 1000uf 16v, PI coil (4mm Core, 10 turns, 14AWG), and a Su'scon SD 2200uf 16v.
5V - S40D45C (40A, 45V) Schottky Rectifier, Canicon 3300uf 10V, PI Coil (5mm Core, 7 turns, 14AWG), JEE 2200uf 10V.
3.3V - SBL3040PT (30A, 40V) Schottky Rectifier, Canicon 3300uf 10v, PI Coil (4mm Core, 10 turns, 14AWG), JEE 2200uf 10v.
-12V - FR154?, PI Coil, Canicon 680uf 16v.
-5V - 2x FR154?, PI Coil, Canicon 680uf, 16v.

Of all of the main secondary capacitors, only three appear to have not failed yet, those being the Canicon 680uf 16v cap on the -5V rail, and the two Su'scon caps on the 12v rail, all of the smaller capacitors (except for one) are all Canicon, and a few appear to have been baked, judging by the fact their sleeves have turned a yellowish brown, though next to a couple of these baked capacitors is the only non-Canicon tiny capacitor (Su'scon SK 0.22uf 50V), and its sleeve is still bright yellow, which might say that the Canicon sleeves are just crap (just like the capacitors )
The controller chip is a KA7500C PWM IC, paired with a SG6510D1 supervisor chip.

And lastly, the fan.
IMG_20220416_185902[1].jpg
Not much to say about this fan, other than the fact it's totally seized up, but i didn't notice that until i was trying to test the PSU and noticed that the fan wouldn't move, despite getting the PSU to turn on, the PSU itself was also very whiny when i powered it up, and that was with no load on the output, i assume the failed capacitors have something to do with it, but i guess i'll have to recap it and see.
Also, this happens to be my 666th post O-o.

[PeteS in CA]

How often have I posted this: those heatsinks do look reasonably substantial. The main transformer looks reasonable as well. 820uF for the input lytics seems a bit marginal for 500W (more OK for 400W-450W), and I don't love that they are 85C rated. ASSuming they current-share perfectly, those paralleled 12V rectifiers are not good for 28A ... even 16A rectifiers paralleled would be marginal for 28A. In contrast and oddly, the 5V and 3.3V rectifiers are reasonable for the rated current. Another oddity is pairing a -52 material output inductor with a TL494 PWM and bipolar transistor switches (both of which keep the switch frequency low, probably around 25-35 KHz).

But for the under-rated 12V rectifiers, that looks to be a fairly decent PSU crapped up by low quality capacitors. And it did last for 10-16 years ("06" date code on one of the input lytics). Given that vintage, replacing the crap caps on the secondary side with PWs or LXZs should yield a pretty decent 400W-450W that could deliver 500W peak power for more than a femtosecond or two.

[Dan81]

Had this PSU planned for a KT133A build with a recapped ECS K7VZA rev 3, and a 1100MHz Duron (wanted a ceramic chip specifically, and the only ceramic Athlon I had is dead as a doornail ) and figured I should post it here since I don't think it looks to shabby.

For those curious:

- transformers are from an discarded Allied unit (it's one of those PSUs I've done over the years) and are their real advertised sizes (ERL35, and 2x EEL-19 from what it reads on their top sides)
- BJTs are 2x TT2194, as well as a standard CET CEP2N6 FET for 5vSB.
- secondary sillicon: 2x SBL3040PT for 3.3v and 5v, STPS20S100CT for 12v.
- Sanyo WG 1500uF 6.3v for 5vSB.
- fan is a thermally controlled Xinruilian LP

[PeteS in CA]

400W? No. The heatsinks might be good for 300W, but that main transformer and 470uF input caps put the practical continuous power in the 250W range, IMO.

[Dan81]

Originally Posted by PeteS in CA

400W? No. The heatsinks might be good for 300W, but that main transformer and 470uF input caps put the practical continuous power in the 250W range, IMO.

That's a 35 sized transformer though (not the puny 33 that most Deer/L&C usually ship with), so apart from the main caps, I'd put it at 280W continuous and 300 peak. Maybe 300 continuous if I could bump up the main caps (and if I had an ESR meter to check if some 680uF primaries I found would suit it)

But considering that the machine itself barely hits 250W probably (1x HDD, 1x DVD drive, FX5200 AGP, LAN, TV Tuner and a 1.6GHz Duron), I don't really think I have to tinker with it further than installing it back in the case and using it.

And 400W... yeah, that's a no from me too.

300-350W with upgraded primaries, since the sillicon seems up to it, and I trust the 35 size transformer to do so too.

(double checked and it indeed IS a real 35 sized trafo, unlike Allied's stupid practice of masking a rounded 28 sized trafo as 35 size - yes, some of the newer units they were making (some Rosewills being good examples) had a rounded 28 transformer passing as 35. Only upon noticing there was too much free space between the main transformer and the one next to it, (and comparing to a real 35 I have from another Deer, marked LT535-something + ERL-35ASH) I found out it's a 28. Talk about saving a penny, as the sillicon and othrr parts were way overkill for that 28.)

Anyways, I might post up a Sun Pro/Leadman next. Doesn't look too bad to me IMO, but the driving transformers are small compared to the main transformer (which for whatever reason, Sun Pro decided to slap an actual EI-40 transformer out of the blue ). Gotta see if it passes the startup blow test, though last time I took it out of the case, the only things wrong with it were bulging 5vsB and secondary caps, nothing else looked shorted.

[Dan81]

Welp, the Sun Pro got scrapped. 5vSB had an issue where it would constantly tick, despite being recapped.

Instead, I've taken my time to open up an unit I wanted to work on for quite some time. Rhis is a "HKC" SZ-430PDR unit.

I'd guesstimate 300W continuous and 350 peak? Main switchers are D304X, main transformer is ERL35 and the drivers look beefy enough. A recap is in order of course (I have saved some Rubycon and Panasonics) although the primaries will stay (IIRC I got those out of a beefy Rasurbo unit that had bigger issues with the secondary side)

Also has PFC coil (and it's real!) and a Shenzen Xin Wang Electronic fan (which I'll probably replace with a Young Lin/ Recom branded fan). Mains filtering was redone by me (unsurprising since the original positions are done by a bunch of monkeys) and the 2200uF Jun Fu on the secondary is also my work (the only 2200uF 16v cap I had on my hands..) but otherwise the rest is as bone stock as possible (except the fan which I had it from a scrapped RPC unit that was so gutless I wouldn't have trusted it near anything ATX.)

[leotron]

hello badcaps i was selling for a scrap a few psu and noticed one that was heavy, hmm i said, i opened up and i found a surprise, looks like a second brand from a psu manufacturer maybe? what do you think?

[pc7fan]

Originally Posted by leotron

hello badcaps i was selling for a scrap a few psu and noticed one that was heavy, hmm i said, i opened up and i found a surprise, looks like a second brand from a psu manufacturer maybe? what do you think?

Looks like a filled up Deer (full input filter and nothing seems to be omitted). If you fully recap this L&C/Deer (upgrade the primary caps to at least 470u for 250W if you have any laying around). Even if this has a 35 size transformer, the somewhat scrawny heatsinks themselves, the classic half-bridge design, and the single 80mm fan will limit this supply at 250W continuous.

[Dan81]

200W.

I'll say this as long as necessary - don't be fooled about the ERL-35-2005 labeling and the transformer shape - that is a rounded ERL-28 transformer which Deer uses to fool people into thinking they have an actual 35 sized transformer. If you have trained eyes, you'll easily notice the rather big free space inbetween the main transformer and the one close to it. An actual 35 size trafo WILL FILL that space.

Also it doesn't surprise me it's Allied who does that. They've been doing this absolute BS of using rounded 28 size trafos for almost as long as I could encounter. The only Allieds I've seen that actually had real 35 sized transformers were usually either Modecom FEEL-III units and some of Thermaltake's Litepower units that were subcontracted to Solytech (not all Litepowers were Deers, by the way.)

[leotron]

Originally Posted by pc7fan

Looks like a filled up Deer (full input filter and nothing seems to be omitted). If you fully recap this L&C/Deer (upgrade the primary caps to at least 470u for 250W if you have any laying around). Even if this has a 35 size transformer, the somewhat scrawny heatsinks themselves, the classic half-bridge design, and the single 80mm fan will limit this supply at 250W continuous.

Thanks for the answers, so its a realiable PSU so long 200~250w power consumption? Or it it's better to buy a generic 500w psu anyways? I was very curious because the main filter, and the other horizontal circuit, also kinda heavy.

thx for the patience

[Dan81]

It will be good for 200W continuous. Again, that's a 28 size transformer you have there, not a 35 and not a 33 either. I've actually desoldered one of those fake 35 labeled size 28 trafos from an Allied rated a bit higher than yours by quite a lot ("400W", my ass!) and it did indeed read "ERL-28-2005 XF". Not sure what the XF meant.