Power supply build quality pictorial. part 2

[370forlife]

Originally Posted by Pentium4

Came across an interesting PSU so I decided to post it. I think it's made by XHY Power? http://www.xhy-power.com.cn/ But it looks so similar to a KME I have. It uses PFC (kinda) for full range, I recapped that part first. That bigger caps' value is 4.7µF 400V originally Canicon, replaced with nichicon VC. No input filtering was installed except for those lousy "Y" caps that look damaged from the factory. I assume normal input filtering won't mess with that PFC circuit?
As you can see, it was designed for a normal PFC circuit

10A bridge, the two FET's on the primary are HFA9N90 rated for 9A each @ 25C and 5.7A @ 100C, combined with a TL3843P. The input caps are 1000µF 200V Fuhjyyu, they read 904µF and 909µF. 5VSB transistor is a Toshiba C5353 rated at 3A

The secondary looks pretty decent, with a huge heatsink, 30A (MOSPEC) schottky for 3.3V, two 40A (ST) schottky in parallel for 5V, and 30A (MOSPEC) schottky for 12V (All TO-247) Each rail has CLC, and the toroids look big. There are actually two 12V rails with OCP shunts, only the 12V CPU connector is on the first rail, and towards the end of the 4 pin plug, one of the wires is split into two wires going into the same pin. Controller is a PS222S http://www.siti.com.tw/product/spec/Power/PS222S.pdf

Soldering looks pretty good except for some big blobs.


Let me know what you guys think.
EDIT: Cap brands used: Canicon, Fuhjyyu, Jun Fu, Forever, JUNZL, Asia'X, and EU CAP

This is most likely a KMEX, as this is a copy of an old (but good) ATNG platform that KMEX is known to make a copy of:

http://badcaps.net/forum/showthread....highlight=ATNG

[Pentium4]

Doesn't KMEX always have "Ever Power" on their transformers though? Cool ATNG PSU, thanks for sharing! I also have a Sun Pro with the same PCB that I will post later.

[c_hegge]

Originally Posted by Pentium4

c_hegge will love this one. Found this beauty in the garbage. 5VSB only gave 4.11V. The person who put it in the garbage was verbally scolded by yours truly What a well built unit. Too bad the glue person at the factory went buck wild on this thing, most of it is still white though. The caps that went bad are Ltec LZG and LTG (The brown one) The Taicon on the 12V appear to be fine. This thing did get over 6 years of 24/7 use, and I don't think a single top mounted 92mm fan is adequate enough for this PSU, so the Ltec did good. There are some CapXon in there too, they might be out of spec. The bulk ones are good sized at least. The switchers are two W9NK90Z rated at 9A. Seems plenty for 460W although I'm not a fan of that heatsink, especially with how slow the fan spins. The fan is a death Adda but it still spins very well even though it's pretty dry. It has two bridge rectifiers that share a heatsink, and each is rated for 8A. It has a nice input filter, with huge MOV's. Is there any benefit to having two NTC thermistors in parallel? They're both SCK 058. There's actually three 12V rails with OCP. Rail A is for CPU, Rail B is for mainboard connector + peripheral. Rail C is for the PCI-E connector. They all share two MBR20H100CT's. 5V uses two 20A schottky's as well. 3.3V uses one 20A schottky. 5VSB and -12V both have rectifiers on the secondary heatsink as well. 2A for 5VSB and 1.5A for -12V. Soldering looks excellent, oh yeah, cause it's Delta! Also love the mostly 16AWG wiring. Gonna recap it.

That's a great looking PSU! And yes, you should really let people know what a mistake it is to toss a Delta PSU like that

As far as I know, that white glue doesn't have issues at all, and is perfectly safe to use. It's only that brown glue that is a problem. Fix up the caps on it and it should be a nice reliable PSU.

[momaka]

Originally Posted by c_hegge

As far as I know, that white glue doesn't have issues at all, and is perfectly safe to use. It's only that brown glue that is a problem.

+1
That white glue is, I believe, proper electronics-safe silicone glue. Delta has been using it for a while.

Nice PSU indeed. In addition to yelling at whoever threw out this PSU, you should also them.

I hate it when Bestec does the same thing (drowning everything) with that crappy tan conductive glue, though.

[japlytic]

Lite-On PS5251-5

560uF 200V OST RLG primary pair with 2.2uF 450V LTEC TY.
220uF 35V LTEC LZG for primary side converter control B+.
2200uF 16V Taicon PW on +12V rail.
1000uF 10V Taicon HD before and after +5V inductor.
2200uF Taicon HH before +3.3V output inductor with 680uF 10V Taicon PW after it.
2200uF Taicon PW on -12V rail.
1500uF 16V OST RLS before +5VSB inductor with 680uF 16V OST RLS after it.
330uF 25V OST RLS on speed controlled fan supply.

Other capacitors are Taicon PW (x2), Taicon HH (??? x1), LTEC LZG (x1) and OST RLP (x2)

KHB 900N90N primary MOSFET with GBU608 primary rectifier.
MBRB20100 rectifier for +12V rail.
MBR30L45 x2 rectifiers for +5V rail.
MBRB2045 rectifier for +3.3V rail.
TIP117 fan speed control output driver.

Has Lite-On gone downhill a bit? It seems so.

[Wester547]

Pictures don't show....but what do you mean by downhill? They've been using crappy capacitors for ages now...

[Pentium4]

The Delta came back up at exactly 5.00V on the 5VSB The 2200µF 16V Taicon PW on the 12V were perfectly in spec.

[Pentium4]

Here's a pretty new-ish Bestec that I recapped. 100% CapXon in there originally. But the bright side, no fucking 0.1µF lytics! I know it's a low value but I still don't trust them when they're cheap brands, and I can't really measure them. The CapXon KF read high capacitance out of spec on the 5VSB, 5V, and 3.3V but the 12V ones were fine I really like that there's 4 filtering caps for both 12V rails (Which uses two 20A schottky's). Originally 4 CapXon KF 2200µF 16V. Replaced with 3 Samxon RS 3300µF 16V and 1000µF 16V Panasonic FM for the cap adjacent to the CPU power wires. The 820µF bulk caps read 816µF and 813µF. Redid a few soldering joints, such as that one on the bridge rectifier which is pretty poor if you ask me. Nice PSU overall though!

[c_hegge]

HP/Hipro D2567F3P. Being a Hipro, I'm expecting a decent unit, of course.


Two 12V rails, with 18A total 12V capacity. Not too bad, at least for a lower powered PC


Let's adda fan shot to this post


Nice beefy secondary heatsink there. Not that we expect any less from Hipro, though.


We have PPFC, and 560uF primary caps. Good enough for a 250 Watter if you ask me. I can't tell what the switcher is, though.


Secondary caps are mostly LTEC, with the exception of a couple of teapos. I'm not sure what the silicon is there, as I would have to remove the secondary heatsink to figure it out. The 3.3V rail does use linear DC-DC conversion from the 5V, though. I guess that's why we need such a chunky heatsink.


It may not be immediately obvious from that picture, but that Teapo cap on the 5vsb there is bulging.


The soldering is pretty good, as we would expect from Hipro.

This one looks like it should definitely be a keeper after a recap.

[Wester547]

Originally Posted by Pentium4

Here's a pretty new-ish Bestec that I recapped. 100% CapXon in there originally. But the bright side, no fucking 0.1µF lytics! I know it's a low value but I still don't trust them when they're cheap brands, and I can't really measure them. The CapXon KF read high capacitance out of spec on the 5VSB, 5V, and 3.3V but the 12V ones were fine

It's because the higher voltage prolongs the electrolytic dissolution or at least rebuilds or keeps the oxide layer from thinning faster than lower voltages do. Not that CapXon wouldn't eventually fail anyway....

We have PPFC, and 560uF primary caps. Good enough for a 250 Watter if you ask me. I can't tell what the switcher is, though.

Look at how much airflow that passive PFC blocks to the primary heatsink. Guess it doesn't matter in a PSU this overbuilt. The main switcher looks to be in TO-3PN so it should be good for 300W-350W.

Secondary caps are mostly LTEC, with the exception of a couple of teapos. I'm not sure what the silicon is there, as I would have to remove the secondary heatsink to figure it out. The 3.3V rail does use linear DC-DC conversion from the 5V, though. I guess that's why we need such a chunky heatsink.

As I recall, the 300W Hipro you reviewed that used the magamp circuit for +3.3V had the same heatsink so I don't think that is why. And the +3.3V rail in this unit actually has its own transformer tap, hence its own rectifier and its own winding in the main toroid and its own set of filtering capacitors before and after the regulator.

It may not be immediately obvious from that picture, but that Teapo cap on the 5vsb there is bulging.

Probably because the one before the coil failed so that poor 8mm Teapo had to handle all that ripple....

[c_hegge]

Originally Posted by Wester547

Look at how much airflow that passive PFC blocks to the primary heatsink. Guess it doesn't matter in a PSU this overbuilt. The main switcher looks to be in TO-3PN so it should be good for 300W-350W.

Yeah, I know, but I loaded it up to 250W on the tester, and the secondary heatsink still got much hotter. The primary heatsink was still cool enough to touch comfortably.

Originally Posted by Wester547

As I recall, the 300W Hipro you reviewed that used the magamp circuit for +3.3V had the same heatsink so I don't think that is why. And the +3.3V rail in this unit actually has its own transformer tap, hence its own rectifier and its own winding in the main toroid and its own set of filtering capacitors before and after the regulator.

Yeah, I know. It was more of a payout on how inefficient linear regulation is

Originally Posted by Wester547

Probably because the one before the coil failed so that poor 8mm Teapo had to handle all that ripple....

The 5vsb ripple was only 25mV with 2A load, in spite of the bad cap. So, on the contrary, the Teapo is probably the only one that failed and the cap before it on the other side of the heatsink is now doing all of the work. If the other cap had also failed, the ripple would be way out of spec. It's not in a hot zone either - there are no load resistors or anything next to it. I think it failed just because it's a junk Teapo.

[Wester547]

Originally Posted by c_hegge

The 5vsb ripple was only 25mV with 2A load, in spite of the bad cap. So, on the contrary, the Teapo is probably the only one that failed and the cap before it on the other side of the heatsink is now doing all of the work. If the other cap had also failed, the ripple would be way out of spec. It's not in a hot zone either - there are no load resistors or anything next to it. I think it failed just because it's a junk Teapo.

How well is the ADDA fan working? Yes, though, it could easily be the fact that it's an 8mm Teapo. I think a 10mm Teapo would have survived, though... it would be interesting to see how high the ripple would be with only 2A on the +5VSB without the fan spinning (standby mode only). I thought that the flyback topology puts quite a bit more ripple current through the capacitors without the aid of the magamp.

Originally Posted by c_hegge

Yeah, I know, but I loaded it up to 250W on the tester, and the secondary heatsink still got much hotter. The primary heatsink was still cool enough to touch comfortably.

I noticed this too with the Newton 250W I posted a few months ago. I was taken aback by how warm the secondary heatsink was after a while of 100% CPU usage (Pentium 4 Northwood 2.66GHz). It was much warmer than anything else in the unit and it is similarly massive, and that's without a linear regulated +3.3V rail. No wonder the thermistor is on the secondary heatsink in these units.

[Behemot]

I don't think the PFC coil is blocking anything in there. On the contrary, it is in the air to cool itself. I have just reviewed a Thermaltake Litepower 400 PP (W0161) based on 300-350W Delta board. The PFC coil was clearly the hottest part inside, outcoming air from (by the outtake grill) peaked at temperature over 50 °C.

[Pentium4]

Wow, nice unit. 560µF bulk capacitance is huge for 250W! I also don't think the PPFC coil will cause too many issues, since it's a "low wattage" PSU. Once you recap it, it won't really matter too much how good the airflow is. If these can last up to 5 or 6 years with low quality caps, good caps in there will probably be good till ATX is no longer the default form factor

[goodpsusearch]

560uF is ok for 300 peak watts depending on the design! This psu is an ordinary example of Hipro build quality.

I am disappointed with the bulging Teapo...

[Pentium4]

Originally Posted by goodpsusearch

I am disappointed with the bulging Teapo...

It's on the 5VSB though.

Here's another Topower, that actually says Topower on it. It was locking up the system, and the top fan was grinding very loudly. 23,958 running hours on it powering an X-ray machine at a vet clinic, the system was Core 2 Duo with mostly low power components and the special X-ray PCI cards. All the caps looked good but I don't trust those Jun Fu. Turns out the caps on the 5VSB, 3.3V, and 5V failed without bulging. Can't really blame the one on the 3.3V after the coil, it was touching the 15Ω dummy resistor for the 3.3V (And you can see it changed the color of the glue). The one on the 12V was 215Ω, none on the 5V. All the caps are Jun Fool by the way. Hey at least it uses a Viper22A for 5VSB. Soldering looks pretty good for a Topower. Both Globe Fans were really dry.

[Wester547]

Originally Posted by Pentium4

It's on the 5VSB though.

But it's after the ferrite coil, so the capacitor before it would be taking most of the ripple current but it hadn't yet failed as evidenced by c_hegge's results on the oscilloscope. I doubt that small 470uF 10V 8mmx11mm Teapo SC was getting hot at all so I think it failed because it's a crappy 8mm Teapo and nothing else. 560uF should be fine for 300W continuous as far as bulk storage goes, anything lower would be cutting it a might bit close though. And 24,000 hours is impressive for JunFu even failed as JunFu = Fuhjyyu. Then again, I'm surprised those CapXon didn't explode after 50,000 hours of use (6 years of 24/7 use) in the Delta you posted. In fact, only the LTEC bulged, that was disappointing, though I doubt the CapXon are in spec anyway except maybe those in the voltage doubler.

[Agent24]

Originally Posted by goodpsusearch

560uF is ok for 300 peak watts depending on the design! This psu is an ordinary example of Hipro build quality.

I am disappointed with the bulging Teapo...

Hipro, I am disappoint.

[Pentium4]

But it's after the ferrite coil, so the capacitor before it would be taking most of the ripple current but it hadn't yet failed as evidenced by c_hegge's results on the oscilloscope. I doubt that small 470uF 10V 8mmx11mm Teapo SC was getting hot at all so I think it failed because it's a crappy 8mm Teapo and nothing else.

I agree. I don't think the rest of them would be healthy either.

560uF should be fine for 300W continuous as far as bulk storage goes, anything lower would be cutting it a might bit close though.

I know this is ideal for the wattage, but in some of c_hegge's PSU reviews in the cheap PSU roundup proves that some can go 300W+ with 330µF or 470µF. This Huntkey did ~375W with 470µF: http://hardwareinsights.com/wp/the-2...ply-roundup/3/

350W with 470µF: http://hardwareinsights.com/wp/the-e...-round-up-2/8/

And their capacitance might be lower, cause we know how shady the ratings are on chinese high voltage caps

And 24,000 hours is impressive for JunFu even failed as JunFu = Fuhjyyu.

True. Plus the fan designs in these Topower's are pretty poor, they have to get very hot to rev up. The sides of the caps touching the resistor were charred black, permanently fading the paint.

Then again, I'm surprised those CapXon didn't explode after 50,000 hours of use (6 years of 24/7 use) in the Delta you posted. In fact, only the LTEC bulged, that was disappointing, though I doubt the CapXon are in spec anyway except maybe those in the voltage doubler.

I was surprised too to see how in spec they were. But they had easy roles like PSU startup, -12V rail filtering, and voltage doubler. The bulk ones read 989µF and 992µF. Pretty good. I've never personally seen CapXon fail on the voltage doubler, and I check them every time I see them. Seen one fail on APFC duty in an OCZ ModXstreme 700W though.

[Wester547]

Originally Posted by Pentium4

I know this is ideal for the wattage, but in some of c_hegge's PSU reviews in the cheap PSU roundup proves that some can go 300W+ with 330µF or 470µF. This Huntkey did ~375W with 470µF: http://hardwareinsights.com/wp/the-2...ply-roundup/3/

I found that to be bewildering as here is another review of a PSU that looks almost identical from Huntkey (with both fans in both PSUs directly wired to +12V, or at least no thermistor is evident) and yet the one in that review exploded at 350W instead of 420W. (in other words, it exploded at its rated load as is characteristic of most older Huntkeys) Both were tested at 230V and the efficiency and power factor of the other one is quite a bit worse too.