Power supply build quality pictorial. part 2

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Looking at your pix, I was tempted to post, "Where's the beef?!" But for a 250W rating, that actually isn't awful. As long as the input lytics are at least 330uF, it could do 200W reasonably OK-ish.

[momaka]

FSP Group Inc. AX400-PN

Got this one a while back on eBay with a bunch of other old PC parts. Namely, it was with the ASUS K8V-SE Delux motherboard shown in this build, along with the accompanying HDD and RAM. Everything worked from that parts lot. The PSU was just unstable, because it had bad caps, as you will see a little further below.

Anyways, let's show it… as usual, starting with case shots:




As can seen, pretty standard gray-box PSU with 120 mm overhead fan and non-sleeved wires – simple! (Just the way I like them.)

Label shot next:


The label shows a pretty honestly-rated PSU – 350 Watts continuous and 400W peak, with 300W for the “dual” 12V rails combined (i.e. 25 Amps.)

So let's see what the internals hold and if they can support the label's claims.




Well, for a 350W continuous PSU, this looks really good – in fact better than what the pictures show here. In particular, the main transformer is ERL39 size (39 mm wide), which should definitely be enough. Also, the main output toroid is a T130-26 part with plenty of copper on it, so it too should be more than adequate. The design is a half-bridge with BJTs, so the Micrometals -26 core of the main output toroid should be fine here. Both heathsinks are quite hefty too. All in all, I can't think of anything to complain about. Perhaps I can try to pick on the lack of a NTC thermistor on the input? Seems like that might actually be a valid concern… but it may not be, as there is also a large multi-turn single-mode choke on the input. So I imagine that could attenuate inrush currents perhaps. OK, I know what I can complain about – the tan/brown conductive glue. But it hasn't turned brown yet, suggesting the PSU has been running pretty cool.

Anyways, here is a shot of the primary side in more detail:


No complaints here either. Primary caps are Teapo LXK @ 200V and 680 uF – good enough for the 350W continuous rating. Main BJTs are a pair of D209L in TO-3P case. Bridge rectifier is an 8-Amp part (GBU8J.) There's even space behind the primary heatsink for a PPFC coil, which wasn't installed on this unit. On the plus side, this should allow the overhead 120 mm fan to push air easier through the PSU – especially on the primary side.

Moving onto the secondary side…




Ah, now we see what the problem is: bulged caps. In particular, these two are the output caps on the 12V rail – 2x Teapo SC, 16V, 2200 uF. I think that's probably the reason why the seller on eBat decided to “scrap” the system and sell it for parts. Shame, because this is a good PSU and was an easy fix overall. The motherboard is quite nice too and probably getting harder to find now.

Next, here is our “noise killer”:


There is an LM358 dual op-amp IC on the other side of that board, along with the power transistor. The NTC for the fan controller is located next to the secondary heatsink on the main PCB, right below the 3.3V rail's output toroid.

Speaking of the fan, this is it: a Yate Loon D12SH-12, rated for 0.3 Amps @ 12V.


I just opened the PSU for cleaning this week after using it for close to 3-4 years now (mostly in my gaming PC - the Optiplex 790 from the Post Your System thread.) Although I didn't put it through any 24/7 use, I did use it for about 1-2 hours a day on average… and that's doing mostly gaming or CPU-intensive tasks. So after a few years of this, the fan was starting to sound a little dry. For whatever reason, I didn't lube it when I recapped the PSU back then. The factory grease was pretty much completely dry now. After a good cleaning with IPA and a quick resurface of the sleeve bearing, followed by more cleaning and addition of fresh oil, the fan runs like new again.

Lastly, here is a shot of the underside (the solder side):


Overall… pretty good!

That completes the image gallery for this PSU. A more detailed component breakdown can be found below.

Primary Side:
* EMI/RFI filtering: three 472M Y2-class caps; two X2-class caps (1x 0.47 uF? and 1x 0.1-0.22 uF?); two CM chokes and one SM choke
* Protection: F8AL or F10AL fuse; 2x MOVs across each primary cap; NO NTC thermistor
* Other: GBU8J bridge rectifier (not heatsinked)
* Caps:
*** 2x Teapo LXK, 200V, 680 uF, 22x45 mm
*** 1x Teapo SEK, 50V, 22 uF, 5x11 mm (5VSB startup / run cap)
*** 2x Teapo SC 50V, 2.2 uF, 5x11 mm (part of BJT driver circuit)
*** 1x 2.0 uF 250V P.P. for main PS transformer coupling
* Main PS (H-bridge topology): 2x D209L NPN BJTs (TO-3P); 51-Ohm 2-Watt resistor + 102K 1KV ceramic cap for snubber ciruit
* Transformers: ERL39 (39 mm) for main PS; ERL/EE16 (16 mm) for BJT drive; ERL/EE16 (16 mm) for 5VSB

ICs:
* DM311 (P-DIP8) for 5VSB
* FSP3528 (P-DIP20) custom PWM chip with protections from FSP
* LM358 (P-DIP8) for fan controller

Secondary Side:
*Output Inductors: T130-26 common-mode toroid for 5V/12/-12V; T106-26 (25 mm dia.) for 3.3V rail mag-amp filter

* 5VSB
*** 2x Teapo SC, 10V, 1000 uF, 8x20 mm caps with PI coil between
*** 3Amp or 5Amp diode rectifier
*** PI coil: 32-turn, 22 AWG, 3.5x25 mm rod core

* 3.3V Rail
*** 1x Teapo SEK, 10V, 3300 uF(?), 10x30 mm cap before PI coil
*** 1x OST RLP, 10V, 4700 uF, 10x30 mm cap after PI coil
*** 1x 30 Amp or 40 Amp (TO-247) Schottky rectifier
*** PI coil: 5.5-turn(?), 14-16 AWG, 5x15 mm rod core
*** Load Resistor: 10-Ohm, 2 Watt

* 5V Rail
*** 2x Teapo SC, 10V, 3300 uF, 10x30 mm caps with PI coil between
*** 1x 30 Amp or 40 Amp (TO-247) Schottky rectifier
*** PI coil: 5.5-turn(?), 14-16 AWG, 5x15 mm rod core
*** Load Resistor: 15-Ohm, 3 Watt

* 12V Rail
*** 2x Teapo SC, 16V, 2200 uF, 10x30 mm caps with PI coil between
*** 2x SF1603g (16 Amp, 150V, TO-220) ultra-fast rectifiers in parallel
*** PI coil: 10-12-turn(?), 14-16 AWG, 5x30 mm rod core
*** Load Resistor: 1-KOhm, 0.5 Watt

* -12V Rail
*** 1x OST RLP, 16V, 220 uF, 6.3x11 mm cap after PI coil
*** 2x 2-3 Amp (?) diodes
*** Load Resistor: 1-KOhm, 0.5 Watt

* Auxiliary Sec. Side
*** 1x Teapo SEK, 50V, 47 uF, 6.3x11 mm cap

*Other caps:
*** 1x Teapo SEK, 50V, 4.7 uF, 5x11 mm
*** 5-6 mm height: 1x 16V 16 uF; 1x 35V 10 uF; 1x 16V 47 uF; 1x 50V 1 uF

* output wires: 20+4 pin ATX, 1x 4-pin 12V CPU, 1x 6-pin PCI-E, 6x Molex, 1x floppy, 2x SATA… all rated for 300V and 18 AWG, except floppy wires.

And that's pretty much it. Up next follows a cap diagram and my partial recap. Stay tuned!

[momaka]

FSP Group Inc. AX400-PN – cap diagram, partial recap, & resistor mods

As far as recapping this PSU goes, it’s very simple. Each main rail has just 2 caps, as shown above (except for the -12V rail.) The only difficult part is getting to those caps due to the tan glue and also due to how cramped the secondary side is. (Those output wires really do get in the way!) But in any case, here is the cap diagram… or rather 2 of them, due to the wires blocking view for some of the caps.




Note that I didn’t mark the two 5VSB caps on there, just because of how the images came out. But I think those should be pretty easy to figure out, as there are only 2x 10V, 1000 uF caps in this PSU, and they are right next to the two smaller transformers.

As usual, I was short on capacitors, so I just replaced the “first” cap in each rail (that is, “cap #1” in the diagram above), except the 12V rail, since it had both of its caps failed. I also replaced a lot of the smaller caps that are of importance – namely, the 5VSB start/run cap, the BJT drive circuit caps, and the auxiliary secondary side rail. Here are how the results turned out:






For the 5VSB rail cap #1:
1x Teapo SC, 10V, 1000 uF ---> 1x United Chemicon KZE, 10V, 1000 uF

For the 3.3V rail cap #1:
1x Teapo SEK, 10V, 3300 uF(?) ---> 1x United Chemicon KYB, 6.3V, 3300 uF

For the 5V rail cap #1:
1x Teapo SC, 10V, 3300 uF ---> 1x United Chemicon KYB, 6.3V, 3300 uF

For the 12V rail caps:
2x Teapo SC, 16V, 2200 uF ---> 2x United Chemicon KYB, 16V, 2200 uF

For the -12V rail cap:
1x OST RLP, 16V, 220 uF ---> 1x Rubycon YXJ, 25V, 220 uF

For the 5VSB startup/run cap:
1x Teapo SEK, 50V, 22 uF ---> 1x Nichicon PW, 50V, 22 uF

For the 2x BJT drive circuit caps:
2x Teapo SC 50V, 2.2 uF ---> 2x Panasonic FC, 50V, 2.2 uF

For the auxiliary secondary side cap:
1x Teapo SEK, 50V, 47 uF ---> 1x Nichicon PM, 50V, 47 uF

I also swapped out/around some of the minimum load resistors on the outputs.

3.3V rail minimum load resistor mod:
1x 10 Ohm, 2 Watt ---> 1x 100 Ohm, 0.5 Watt

5V rail minimum load resistor mod:
1x 15 Ohm, 3 Watt ---> 1x 15 Ohm, 3 Watt in series with 1x 10 Ohm, 2 Watt (taken from 3.3V rail)… so total is 25 Ohms, which corresponds to about 1 Watt of power dissipation split between a 3-Watt and a 2-Watt resistor. Needless to say, they both run much cooler now, so this PSU no longer has any hot-running minimum load resistors.

And that’s all for the recapping and modifications I did on this PSU. As mentioned, I’ve had it running this way in my primary “gaming rig” for the last 3 years or so now. When I did the recent check-up, cleaning, and fan lube job a few days ago, all the old Teapo and OST caps I had left from the first time still appeared good. I didn’t pull them to check them, but I didn’t have a reason to either – the PC is still running happy.

Looking at your pix, I was tempted to post, "Where's the beef?!" But for a 250W rating, that actually isn't awful. As long as the input lytics are at least 330uF, it could do 200W reasonably OK-ish.

They are 470 uF, actually... according to the label, anyways. Being JEE brand, who knows. Usually the cheap HV caps regularly show 1 notch lower... but not always.
In any case, I probably won't ever push that PSU past 100W, so even those JEE caps should be good.

[RukyCon]

Re: Power supply build quality pictorial. part 2

So i was cleaning some stuff up, and dug up this busted PC power supply i found in the trash some years ago, and thought i would get some photos of it before i part it out, so here it is, i don't know the brand or wattage, and i'm not really bothered to look it up, due to what i stated above.

Anyway, here is the board overview, and you can probably see i won't be bothering repairing it:

Yep, it's in two halves, i could try to bodge them back together, but it probably wouldn't be very safe. The main transformer is a ERL38 or roughly thereabout.

Onto the bottom side of the board:

The soldering doesn't look bad, though i think that's the least of its concerns given the shape that it's in.

Now let's dive into the primary side of things.

So lest start with the input filtering, which consists of two .22uf X2 capacitors, two 2.2nf Y2 capacitors, a single coil, and a big looking common mode choke. In spite of what it has, it seems a bit minimalistic to me, but who am I to judge. After that, we have a GBU4J 4A amp bridge rectifier, and a pair of Ltek SM 390uf 200v capacitors.

The main switchers are a pair of IRFBC20's, and the 5vsb appears to be switched by a chip marked 1004BL, which i can't seem to find much about, but i assume it's a somewhat standard all-in-one switch mode controller.

Now onto the secondary side of this thing.

So to start off with, the 5vsb is fed into a SB560 diode, before being filtered by a Nichicon PJ 1500uf 10v capacitor, a small pi coil, and a Nichicon PJ 470uf 10v capacitor. The 12v output is fed though a STPR10 diode bridge, and gets the bare minimum when it comes to filtering, since it only has a single Ltek LZG 1000uf 16v capacitor with not further filtering. The 5v output goes through a STPS20 diode bridge, and is filtered through a Ltek LZG 1000uf 10v capacitor, a pi coil, and a 680uf 10v capacitor. The 3.3v output is fed though another STPS20 diode bridge, and is filtered though a 1500uf 10v capacitor, a pi coil, and a 1000uf 10v capacitor (all Ltek of course). And the -12v rail (which appears to be derived from the 3.3v rail if i'm not mistaken), is filtered through a 680uf 25v capacitor, before being fed through a KA7912 negative voltage regulator, before being filtered though a 100uf 25v capacitor.

The 16-pin chip on the output is yet another chip i can not find info about, but its number is 1002DS, and is likely the supervisor/protection IC for this thing. But that's pretty much it in regard to this thing, it looks like it may have been an okay PSU at one point (assuming it was rated for at most 250w), but the physical damage to the PCB makes it nothing more than a donor for other power supplies. Besides, i have many more PC power supplies that i've found in various places (mostly at thrift stores and in the trash), so not really a major loss to me anyway.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Ruky, my WAG is that that is a ~200W Newton Power (that "N" logo near the toroidal input inductor suggests this) or Delta, possibly mid 1990s vintage. I'm guessing the power based on the 390uF input caps, and the 1002 might be a Delta ASIC supervisor IC (Newton is owned by Delta).

The 10A 12V rectifier and 20A rectifiers for the 5V and 3.3V are consistent with that time frame and power level. Likewise the PJ series output caps. The 5A standby rectifier seems over-rated for the common .5A-1A standby current of that time-frame, though.

[lti]

Re: Power supply build quality pictorial. part 2

That is a Newton Power, and from the board size and shape, I'm guessing that it was one of Gateway's proprietary power supplies (standard ATX connectors, but an unusual small form factor that was mostly used in cases that could have held a full-size ATX power supply). Every one of those I remember seeing was 200W.

[RukyCon]

Re: Power supply build quality pictorial. part 2

Okay, thanks for letting me know, because I didn't have a case or model number for this PSU, i could only guess what its power rating was (and i do think i was in the ballpark when it came to the wattage, only off by 50w). Not that it really matters too much due to the shape that the PSU is in.

The 5A standby rectifier seems over-rated for the common .5A-1A standby current of that time-frame, though.

Correct me if i'm wrong, but did some PSU manufacturers use overrated diodes because they would dissipate less heat when used at a lower current? (compared to their lower rated counterparts) I feel like i remember hearing that here before.

Anyway, i have many more PSUs that i plan on posting here someday.

[momaka]

Re: Power supply build quality pictorial. part 2

I concur with PeteS - this is a Newton Power PSU made by Delta. Probably late 90's or very early 2000's - think Pentium II and 3 era / 1st "gen" slot A or socket 462 AMD Athlon. So output-wise, it was probably rated for 200W or less, most of which was for the 3.3V and 5V rails... though not too much, because after all, the 20 Amp rectifiers on the 5V and 3.3V rails probably limit(ed) the output current to 15-16 Amps, since this is a forward-converter design. Thus, the combined 3.3V and 5V output was likely rated around 110-120 Watts. The 10 Amp rectifier on the 12V rail probably limited the output to 6-7 Amps... which comes out to another 70-ish Watts... and that's your 200W.

That said, I don't think that primary transformer is ERL38. Measure the top part in mm and see. It's probably 28 mm across (ERL28)... which for 200W is adequate. I've seen early Delta/Newton and Bestec use these smaller transformers back in the day, and it was never an issue.

Regarding the "1004BL" chip on the primary side - that's very likely an equivalent to a UC384x series (e.g. UC3842, 3843, and etc.) current-mode PWM controller and it's *not* for the 5VSB. It's for the main PS and drives the MOSFET closer to the main transformer. The other MOSFET is for the 5VSB circuit, which is a 2-transistor self-oscillating design with feedback. You can actually see some darkening of the PCB near the 5VSB circuit and its "startup" cap, suggesting this cap has started to go (or has gone) bad. However, the actual reason was probably diode D901 behind it. Not sure why, but Delta liked to use standard 1N4002 diodes for the primary-side auxiliary rail (which feeds the 5VSB's driver transistor and also the PWM controller for the main PS) instead of proper fast-recovery diodes. I made a detailed post about it here (a little further down in that post, after the recapping of the PSU):
https://www.badcaps.net/forum/showpo...postcount=3136

And very likely, the 5VSB circuit on your Newton PSU above probably looks very similar to this:
https://www.badcaps.net/forum/attach...4&d=1621301619
In the case of the above schematic, since it's from a newer Delta PSU (made around 2007 or so), the diode in question is D951.

And the -12v rail (which appears to be derived from the 3.3v rail if i'm not mistaken)...

I'm pretty sure that's not it. -12V rail should be generated from it's own tap on the main transformer... and from what I can see from the underside / PCB shot, it is. It appears to be that first pin on the bottom of the transformer, as seen in this picture:
https://www.badcaps.net/forum/attach...9&d=1646129626
Then it goes up and through a diode, and into the main output toroid.

I know probably none of this matters to you, since you said you're not going to fix this PSU... and I don't blame you! With a broken PCB like that and a "hottie" 5VSB circuit, it indeed isn't worth fixing, unless you really wanted an era-appropriate PSU for a Pentium II / 3 PC. In fact, I'm glad to hear you're going to salvage it for parts - there should be plenty of useful components from this PSU. If you're going to discard any parts from it, though, let me know - I'm mostly interested in the 5VSB transformer. I have a PSU that needs one. The 2-transistor circuit burned it up pretty good. I was actually going to rewind it (and probably still may do that)... but perhaps as a backup? ... well again, only if you're going to discard it. Otherwise, I probably can salvage one from the 10's of PSUs I already have. I tried one from a 5VSB circuit of a single-transistor PSU without feedback, but the pri. aux. winding voltage is lower. I made it work after some modifications, but the low pri. aux. voltage is inadequate for the PWM controller of the main PS.

Correct me if i'm wrong, but did some PSU manufacturers use overrated diodes because they would dissipate less heat when used at a lower current? (compared to their lower rated counterparts) I feel like i remember hearing that here before.

Yes, this is exactly it.
The PSU was likely rated for either 1 or 1.5 Amps (or maybe 2 Amps if it was a really early 2000's model ). So by using a 5 Amp diode, that makes the diode run much cooler. SB560 is also Schottky type. If they used a regular fast-recovery rectifier, the dissipation would have been a lot worse. I kind of showed what happens when you do that with this PSU's fix here:
https://www.badcaps.net/forum/showpo...07&postcount=2

Anyway, i have many more PSUs that i plan on posting here someday.

You and me both.

[RukyCon]

Re: Power supply build quality pictorial. part 2

That said, I don't think that primary transformer is ERL38...

Yeah, that was a typo i failed to noticed before posting (meant to put in ERL28), either that, or my brain was running close to empty, which given the fact i posted that post at around 3AM, might explain it.

Regarding the "1004BL" chip on the primary side - that's very likely an equivalent to a UC384x series (e.g. UC3842, 3843, and etc.) current-mode PWM controller and it's *not* for the 5VSB. It's for the main PS and drives the MOSFET closer to the main transformer. The other MOSFET is for the 5VSB circuit, which is a 2-transistor self-oscillating design with feedback...

I really should've realized that, given the lack of a 3rd transformer (which i believe would've been used to drive the primary side mosfets from the secondary/stand-by side, if i'm not mistaken (again)). Of course if i could find a datasheet on it, i probably would've realized that, but i guess that goes to show why you should never assume things.

I'm pretty sure that's not it. -12V rail should be generated from it's own tap on the main transformer... and from what I can see from the underside / PCB shot, it is. It appears to be that first pin on the bottom of the transformer...

Okay, i realized that, somehow, i got stuck around the where it appears the 3.3v diode bridge is partially tied to the 7912 via a jumper, but i definitely should've spent more time looking at it

I know probably none of this matters to you, since you said you're not going to fix this PSU... and I don't blame you!

While it is true that i won't be repairing this PSU, the info that you provide me here helped me realize the mistakes i made when reverse engineering it, so that (hopefully) i won't make the same mistakes again, so in a way, it does matter to me.

...I'm mostly interested in the 5VSB transformer...

I can send it your way, since i'm not fully sure what i'd do with it anyway, i also don't remove all the smaller components from the PCB (resistors, ceramic/film/tiny electrolytic capacitors, multi-pin chips, etc.), so if you also want the mostly stripped PCB from this, then i could send it along to (i also have some older computer stuff that i'm looking to get rid of) so hmu if you're interested.

[goodpsusearch]

Re: FSP Group Inc. AX400-PN – cap diagram, partial recap, & resistor mods

all the old Teapo and OST caps I had left from the first time still appeared good. I didn't pull them to check them, but I didn't have a reason to either – the PC is still running happy.

You even managed to save those Teapos and Osts from failing or getting pulled out. Excellent

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

The review is still not out, but I think I can share the pics

Corsair VS450 - CWT
Interesting thing is this CAP004DG on one X cap, providing savings on it's safe resistors. GBU406 rectifier. All crappy caps, 180 uF/400 V AiSHi(t) on input; it hovever has CM6805+CM03X which claims to have advanced PFC/PWM requiring cheaper cap and silicon.

Some cheap chinese copy of 13N50 on PFC with PFCD86G diode. Switchers are some 10N50. This all has high Rds(ON), that's the low efficiency of it hitting 80 %, but average is under 80 % thus it has no certification.

TNY177PN for +5 V SB, other crappy caps - small JunFus arround which may be baked by the chip, C(r)apXon2200+1000 uF combo on output. However it has nice beefy diode with nothing arround, so it may stay cool.

Other secondary: double STPS20L60CT/+12 V, STPS30L45CT/+5 V and finally magamped MBRF2045CTG for +3,3 V. Sitronix ST9S429 aka Unisonic Technologies S3515 secondary monitor. Mostly 2200+1000 uF C(r)apXon/AiSHi(t) combo.

It has actually very good voltage control on crossloading, voltages varied even less than on combined control (aprox. 2 % vs 3 %). Ripple is within 60 % of spec (very low on +12 V). BUT. The -12 V rail may be very problematic. Loading it with 0,36-0,38 A of mine Papst fan between -12 and +5 V made it oscillate like hell. I haven't checked at 0,3 A precisely but I think it may have strong problems as well. It's feedback is connected to +3,3Vs voltage-wise and to +12Vs current-wise.

Built quality is not that good providing it's the cheapest Corsair has, but it's not actually dangerous or something.

Where I work, we had a Corsair VS450 that had its fan almost seized.

The fan was barely moving and the PC was buzzing due to vibrations from the dying fan.

We have bought ~ 15 Corsair VS power supplies (CWT platform) and none of them had its fan fail until now. Some of them failed due to dried primary caps (Aishi 400V).

I knew that Corsair/CWT put Yate Loon fans on VS series, so I was shocked to find a Chinese crap fan inside this unit:
https://www.badcaps.net/forum/attach...1&d=1646525302
https://www.badcaps.net/forum/attach...1&d=1646525302

Nobody cracked open this psu before, the warranty sticker was intact.

Lessons learnt from this story:

Brand/model matters a lot! It is not a random incident that this fan died while the Yate Loon fans on the other units keep going.

Corsair and other companies should never be trusted. It is acceptable to change cap manufacturer during production of a model, but sending a psu for reviews with good quality fan and then years later randomly switching to terrible fan brands found on crap power supplies that cost 15-20 euros is plain fraud. If, for some reason, they had trouble sourcing Yate Loon fans they could switch to similar grade fans.

Please correct me if I am wrong and this is not what it looks to me: a typical crap Chinese fan

[lti]

Re: Power supply build quality pictorial. part 2

That's one of those Hong Hua fake-FDB fans. They were popular for a while, but they're just cheap crap.

[RukyCon]

Thermaltake Purepower HPC-420-102 DF

The main switchers are a pair of IRFBC20's...

Nope, it's only one IRFBC20, the other one is a P6NB80.
-------------------------------------------------------------------------
Anyway, here is another PSU that i have, this one (unlike the last one I posted) is going to be fixed up, since it appears to be a well-built PSU (and it wasn't in two halves like the last one).

Label;

Nothing too noteworthy here, not sure what the 'ATX 12V' sticker is supposed to mean (don't all ATX supplies have 12v?), but i'll probably be told if it does have any meaning. :p

Overview;


Not much to say here, it's a decently packed power supply, the soldering mostly looks good, overall just very good looking. Some of you may have noticed that something is bodged onto the primary side, and that is a 220pf 1kv ceramic capacitor, and it's hooked right across the drain and source pins of the Mosfet there, i may be wrong, but i'm guessing someone forgot to put a proper snubber in the PCB design, so this is supposed to be a quick fix.

Primary Side;


Starting off, we have a pair of Teapo LXK 820uf 200v capacitors, both of which test around 748uf in circuit, feeding them is a GBU805 bridge rectifier, and for filtering, we have two 2x capacitors (1x0.22uf, 1x0.47uf) plus a third one located on the IEC connector/EMI filter (0.47uf), the EMI filter contains another x2 cap (0.1uf), 2x 3.3nf Y capacitors, and a common mode choke. Back to what's on the power supply PCB, we also have a NTC thermistor, a rather big looking common mode choke, and another pair of 3.3nf Y2 capacitors. There's also a pair of varistors located after the bridge rectifier.

The main switchers are a pair of 2SC3320 transistors, with the 5VSB/aux switcher being a SSP2N60B Mosfet, which is set up in what appears to be a classic two transistor design. The main transformer is 40mm wide by 45mm tall, which seems oversized for 420w, but better that than undersized.

Secondary side;

The filtering for each real is as follows;
5VSB: SB5100 (5A, 100V) diode, Teapo SH 1000uf 16v, PI coil, Teapo SEK 470uf 16v.
12V: MBR30150PT (30A, 150V) rectifier, Teapo SEK 10uf 50v, PI Coil, Teapo SC 2200uf 16v.
5v: 2x D83-004 (30A, 40V) rectifier, 2x Teapo SC 3300uf 10v (10mm) capacitors with a PI coil between them.
3.3v: SBL4045PT (40A, 45V) rectifier, 2x Teapo SC 3300uf 6.3v capacitors with a PI coil between them.
-12v: 2x PR3001 diodes, PI coil, PR1001 diode, Teapo SEK 470uf 16v.
-5V (fed off of -12V): LM7905CT, Teapo SEK 10uf 50v.

The main controller is a SG6105D supervisor, with a LM339N quad op-amp.
The output wires on this thing were cut off (this was found in the trash around the same time as the newton PSU, so they were likely taken for the copper), but they appeared to have mostly been 16AWG with some 18AWG wires, the only wire that was 20AWG or smaller was the 3.3v sense wire.

The Fans:

Don't know much about this brand, but i took both apart to clean them, and they're both ball bearing fans, so from that, they'll be good i hope.

Anyway, thats this power supply covered, hopefully with fewer mistakes than the last one, as i mentioned at the beginning of the post, i'm going to be fixing this one up, this will hopefully include new wires, and maybe a full recap (i have most of the parts for a recap, just lacking some 3300uf and 2.2uf caps). Judging by the datecodes on this thing, i'm guessing this was made mid 2005.

[Per Hansson]

Re: Thermaltake Purepower HPC-420-102 DF

Nothing too noteworthy here, not sure what the 'ATX 12V' sticker is supposed to mean (don't all ATX supplies have 12v?), but i'll probably be told if it does have any meaning. :p

This is sort of marketing speak to tell you that the PSU is of a more modern design and should be able to accept a power hungry CPU on the 12v rail with little load on the 5v and 3.3v without going out of regulation.
This is because in the olden days CPU's where fed by 5v and even 3.3v in ancient days, so 12v saw very little load.
This changed with the introduction of the Pentium 4.
Another way is to call it "P4 ready" or something along those lines.
Now whether the label actually means anything or is just bullshit is a completely different topic
(Well, based on its label it is bullshit, because it has more power available on the 3.3v and 5v rails vs the 12v rails).

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The SG6105D is a voltage mode PWM and supervisor combo. The LM339 is a quad comparator, and the LM324 a quad op-amp.

I'm not loving the output side heatsink, needs a bit more "beef" I think. All in all, though it's at least a credible 300W-350W PSU, and with good cooling, maybe its full rated power/

Re that 220pF snubber, I've seen designs where a crappy .1uF cap was a snubber, in which its high ESR was the "snubber resistor" (this was at Boschert in the early 1980s). Maybe Thermaltake is doing the same.

This is sort of marketing speak to tell you that the PSU is of a more modern design and should be able to accept a power hungry CPU on the 12v rail with little load on the 5v and 3.3v without going out of regulation.
This is because in the olden days CPU's where fed by 5v and even 3.3v in ancient days, so 12v saw very little load.
....

That's my thinking as well. A few years earlier, around the turn of the century, 12V currents were lower, while 5V and 3.3V currents were similar to this PSU, except with 250W-300W ratings.

[RukyCon]

Re: Thermaltake Purepower HPC-420-102 DF

This is sort of marketing speak to tell you that the PSU is of a more modern design and should be able to accept a power hungry CPU on the 12v rail with little load on the 5v and 3.3v without going out of regulation.

I did think it had something to do with being a more modern PSU, but when i saw the 5v rail was more powerful than the 12v rail, i started to think it meant something else, but i guess my initial suspicion was correct.

I'm not loving the output side heatsink, needs a bit more "beef" I think. All in all, though it's at least a credible 300W-350W PSU, and with good cooling, maybe its full rated power.

This unit does have two fans, one mounted in the usual location on the back, and the other mounted over the primary heatsink with some overlap with the secondary, so maybe that's why they used a smaller heatsink, but maybe i gave this PSU too much credit, as i am not a expert at these things and just do these posts so fun and also to maybe learn a thing or two in the process.

[dmill89]

Re: Thermaltake Purepower HPC-420-102 DF

This is sort of marketing speak to tell you that the PSU is of a more modern design and should be able to accept a power hungry CPU on the 12v rail with little load on the 5v and 3.3v without going out of regulation.
This is because in the olden days CPU's where fed by 5v and even 3.3v in ancient days, so 12v saw very little load.
This changed with the introduction of the Pentium 4.
Another way is to call it "P4 ready" or something along those lines.
Now whether the label actually means anything or is just bullshit is a completely different topic
(Well, based on its label it is bullshit, because it has more power available on the 3.3v and 5v rails vs the 12v rails).

I did think it had something to do with being a more modern PSU, but when i saw the 5v rail was more powerful than the 12v rail, i started to think it meant something else, but i guess my initial suspicion was correct.



This unit does have two fans, one mounted in the usual location on the back, and the other mounted over the primary heatsink with some overlap with the secondary, so maybe that's why they used a smaller heatsink, but maybe i gave this PSU too much credit, as i am not a expert at these things and just do these posts so fun and also to maybe learn a thing or two in the process.

Yes it was common back in the early days of 12V heavy CPUs (Pentium 4/D, AMD Athlon 64, early Core2Duo/Quad, etc.) to mean the PSU could handle these CPUs (at least theoretically) and had the 4-pin 12V CPU connector.
It may or many not be more "modern", early on some PSU manufactures had some "interesting" solutions to "shoehorn" a larger 12V rail into an older design (in some cases even just shoving in a separate board with just a 12V rail on it). Prior to this many PSUs had a sub-10A 12V rail since most load was on the 3.3/5V rails. Here's a PSU (a Compaq branded Hi-Pro) from an AMD K6-2 system with only a 8A 12V rail.



This is pretty heavily built for a 250W unit, hard see the markings on the silicon without disassembling it though.





Note no 4-pin CPU connector:

[momaka]

Re: Thermaltake Purepower HPC-420-102 DF

Nope, it's only one IRFBC20, the other one is a P6NB80.

Good catch!
Sounds about right - IRFBC20 is for the 5VSB (since these are generally low-current parts) and the P6NB80 should be the main PS switch.

Anyway, here is another PSU that i have, this one (unlike the last one I posted) is going to be fixed up, since it appears to be a well-built PSU (and it wasn't in two halves like the last one).

I concur.
Looks OK for a fix-up.

not sure what the 'ATX 12V' sticker is supposed to mean (don't all ATX supplies have 12v?)

Well, looks like Per, PeteS, and Dmill already answered that.
But indeed it was just to denote that the PSU had (supposedly ) a strong enough 12V rail for a 12V-based system. In the early 2000's and before Pentium 4 came out, most motherboards used the 5V rail to power the CPU, and this is where most of the load from the PSU was going. Then around 2002-2003, the 4-pin 12V CPU connector started to appear on motherboards - mostly on Pentium 4 motherboards at first, but eventually also on latter-made Athlon XP boards... and it caught on, because it allowed for more power to be transferred to the motherboard without the need for very thick cables.
That being said, PSU manufacturers still kept the strong 5V rail for a few more years afterwards in order to be backwards-compatible with older motherboards. And yet to show that their PSU was also capable of handing new systems, such "ATX 12V" stickers were quite common. Other ones commonly seen were "Intel Pentium 4 - ready", "Approved by Intel and AMD", "Complies with Intel Pentium 4 and AMD motherboards", and similar. In reality, though, only usually the cheaper PSUs had these labels. That's not to say that the PSU you posted is necessarily a cheap one. Some manufacturers just continued to do that well into the mid-late 2000's for some reason (too lazy to re-design their labels, I guess. )

Some of you may have noticed that something is bodged onto the primary side, and that is a 220pf 1kv ceramic capacitor, and it's hooked right across the drain and source pins of the Mosfet there, i may be wrong, but i'm guessing someone forgot to put a proper snubber in the PCB design, so this is supposed to be a quick fix.

Yes, kind of.
Actually, it's probably more to do with slowing down the turn-On time of the switching transistor on the 5VSB. I've seen this on a few PSU's 2-transistor 5VSB circuits, though it's rare. Indeed it was probably a last-second production floor "fix". I think I even tried removing it on one PSU just to see what happens... and the 5VSB circuit was definitely not happy - regulation all over the place and harsh squealing noises. I did it with an incandescent light bulb in series with the line input, of course... so no damage to the PSU. Just put back the cap in and all was good.

with the 5VSB/aux switcher being a SSP2N60B Mosfet, which is set up in what appears to be a classic two transistor design.

Yup... 2-transistor self-oscillator it is.

The main transformer is 40mm wide by 45mm tall, which seems oversized for 420w, but better that than undersized.

Nice!
It's probably an "ERL39" (i.e. 39 mm). But with the tape insulation an all, it does usually come out to 40 mm measured.
Good to see a large transformer like this. Half-bridge PSUs tend to have slow switching speed of the main PS, so they need more turns for their windings... and thus bigger transformer size if thick windings are to be put inside. This along with the 2SC3320 BJTs makes is a pretty capable PSU - at least on the primary and transformer side of things.

Secondary side; ...

Right off the bat, I see a green toroid core for the 3.3V rail's output inductor, and a pretty good sized one at that. This is a Micrometals -52 core and definitely good to see (lower losses.) The main output toroid appears standard -26 core, but at least it's still pretty large in size and appears to have a good amount of copper on it.

12V: MBR30150PT (30A, 150V) rectifier, Teapo SEK 10uf 50v, PI Coil, Teapo SC 2200uf 16v.

A-ha!
I was wondering why this PSU looked somewhat familiar.
The SG6105d PWM IC was already running a few ideas up in my head.
But that cap combo arrangement on the 12V rail is the giveaway here: this is a Sirtec / Sirfa -made PSU. I believe it is a "High Power" series/design.

Here's a similar Jeantech PSU I posted about awhile back:
https://www.badcaps.net/forum/showpo...postcount=2727

And to some extent, you can also see the similarities in this Task TK-930TX PSU, also made by Sirtec (but slightly older, dating back to 2003):
https://www.badcaps.net/forum/showpo...postcount=2735

That said and regarding the filter caps on the 12V rail - don't change the values too much. For the 10 uF cap before the PI coil, you could probably increase that up to 330-470 uF and not very low ESR (GP 105C is OK)... but I wouldn't go further. Seems the feedback loop doesn't like it otherwise. Actually, IIRC, someone (the Unique??) fixed this issue on one of these PSUs before by moving the 12V rail's feedback point somewhere else... but I'll have to look through my bookmarks to see if I save this or not.

The Fans:
...
Don't know much about this brand, but i took both apart to clean them, and they're both ball bearing fans, so from that, they'll be good i hope.

SuperRed are pretty good fans - about the same level as Yate Loon and such... well, at least the older SuperRed fans, like in this PSU. On ATI/AMD Radeon HD video cards, they transitioned to a "sealed" design, where the fan can't be opened (easily, without some destruction.) Same with PowerLogic.
I'm not a big fan of those newer fans... pun intended.

Judging by the datecodes on this thing, i'm guessing this was made mid 2005.

Yup, you can see that even on the label where it says "0525" - made in 2005, week 25.

I did think it had something to do with being a more modern PSU, but when i saw the 5v rail was more powerful than the 12v rail, i started to think it meant something else, but i guess my initial suspicion was correct.

It's an old 5V-heavy design adapted/tweaked to work with newer PCs.
Actually, in the case of my TASK TK-930TX PSU, I eventually found out that 12V-heavy (but not too heavy, lol ) PCs work OK on it. If I try any 5V-based motherboard, the 12V rail easily goes over 12.6-12.7V, which is above 5% spec... and it also tends to swing a lot as the load on the 5V rail changes. As for the Jeantech PSU, I never tried that in a 5V-based PC. I put it in an old Athlon 64 X2 rig with GeForce 8500 GT video card, and it's been powering that rig OK with decent voltage regulation.

but maybe i gave this PSU too much credit, as i am not a expert at these things and just do these posts so fun and also to maybe learn a thing or two in the process.

Well, still give it a try at re-building, if you have the time or would enjoy doing it. Probably put some wires on the output first and see how the PSU runs with the original caps. If all is well, then recap it. This would also give you a baseline of whether the PSU was actually working or not. Otherwise you might start thinking you did something wrong with the recap and other 2nd thoughts.

Here's a PSU (a Compaq branded Hi-Pro) from an AMD K6-2 system with only a 8A 12V rail.
https://www.badcaps.net/forum/attach...1&d=1647041084


This is pretty heavily built for a 250W unit, hard see the markings on the silicon without disassembling it though.
https://www.badcaps.net/forum/attach...1&d=1647041084
https://www.badcaps.net/forum/attach...1&d=1647041084
https://www.badcaps.net/forum/attach...1&d=1647041084

Nice classic PSU there!
Those are pretty much bullet-proof with good caps.

We have bought ~ 15 Corsair VS power supplies (CWT platform) and none of them had its fan fail until now. Some of them failed due to dried primary caps (Aishi 400V).

I knew that Corsair/CWT put Yate Loon fans on VS series, so I was shocked to find a Chinese crap fan inside this unit:
https://www.badcaps.net/forum/attach...1&d=1646525302
https://www.badcaps.net/forum/attach...1&d=1646525302

Nobody cracked open this psu before, the warranty sticker was intact.

Lessons learnt from this story:

Brand/model matters a lot! It is not a random incident that this fan died while the Yate Loon fans on the other units keep going.

And on the sub level, it probably comes down to what kind of grease/oil these fans use.
I think the main problem is that the cheap/cheaper fans use either very little lubricant or just some low-quality stuff that gunks up / dries away quickly, leaving the bearing to run dry.

Please correct me if I am wrong and this is not what it looks to me: a typical crap Chinese fan

Well, TBH, I could think of many more "crappier" fans.
Te Bao Metallic Plastic / Raidmax fans come to mind here... and not only.
Globe fans are also very model-dependent - some are complete crap, others are quite OK.
And Rulian Science? Remember those from early 2000's Deer and L&C PSUs? - Those seize up very easily too.

That said, I still don't mind even the cheap "crap" bottom-of-the-barrel fans... most of the time.
Usually, they just need a good cleaning / "grooving" of the sleeve bearing and some high quality machine oil. After that, they'll work for many more years without problems.
It is a shame though, that Corsair switched brands like that mid-production.
Then again, it's Corsair - they literally put their name on stuff made by someone else, just like EVGA and many others like that. I tend not to trust such companies too much.

Now if it was a Delta-made PSU with a Delta fan, you know we are talking serious business.

[RukyCon]

Re: Thermaltake Purepower HPC-420-102 DF

Well, looks like Per, PeteS, and Dmill already answered that...

Yeah, i guess thats what i get for saying 'i'll probably be told if it does have any meaning'. Though i find it a bit funny to have that one question answered 4 times by 4 people.

A-ha!
I was wondering why this PSU looked somewhat familiar. /snip/ this is a Sirtec / Sirfa -made PSU. I believe it is a "High Power" series/design.

I was wondering who made it, but the UL number is no longer associated with anything on the UL product IQ page, however, googling it took me to this old Badcaps post, which then led me to this JonnyGURU post which has a bunch of common PSU UL numbers with what manufacturers they were associated with, which might be helpful since UL doesn't seem to keep a permanent record of these things.

...don't change the values too much. For the 10 uF cap before the PI coil, you could probably increase that up to 330-470 uF and not very low ESR...

There isn't really a lot of space to mount any capacitor wider than 6 or 7 mm, since it's in a bit of a tight spot.

Well, still give it a try at re-building, if you have the time or would enjoy doing it. Probably put some wires on the output first and see how the PSU runs with the original caps...

I haven't done a full test yet, but i did do a no-load test and it seemed to have passed that (5VSB came on, PSU fired up when PSON was pulled low). Now one thing i have to point out though is that whatever solder they used for this thing, seems to be highly prone to cracking under any mechanical stress, i'm guessing it must be a cheap or early lead free solder since I've head that stuff had similar issues with cracked joints. I also swapped out some of the tiny capacitors on the primary side before testing, because they were testing with really high ESR (compared to new/used parts), which was a "risky" move, since i didn't know if the PSU worked beforehand. But it appears to have worked out in the end.

[momaka]

Re: Thermaltake Purepower HPC-420-102 DF

Yeah, i guess thats what i get for saying 'i'll probably be told if it does have any meaning'.

Ask a question... get some answers - that's how BCN usually works.

Though i find it a bit funny to have that one question answered 4 times by 4 people.

We like to be redundant.
We like to be redundant.

I was wondering who made it, but the UL number is no longer associated with anything on the UL product IQ page, however, googling it took me to this old Badcaps post, which then led me to this JonnyGURU post which has a bunch of common PSU UL numbers with what manufacturers they were associated with, which might be helpful since UL doesn't seem to keep a permanent record of these things.

Nice find!
I have something similar saved as an image/file on my PC so I can refer to it when I encounter older PSUs. Seems like not many people/places/sites are keeping track of UL numbers anymore... if at all. In fact, even UL's website kind of went dead (or at least it's not search-able by us mere mortals.)

I suppose I don't care much anymore either - I just open PSUs and use my gut feeling to tell if it looks alright or not.

There isn't really a lot of space to mount any capacitor wider than 6 or 7 mm, since it's in a bit of a tight spot.

Probably a good thing.
I got baited with my TASK PSU and put in a full 10 mm dia. cap in there. PSU didn't like it.

Now one thing i have to point out though is that whatever solder they used for this thing, seems to be highly prone to cracking under any mechanical stress, i'm guessing it must be a cheap or early lead free solder since I've head that stuff had similar issues with cracked joints.

Yeah, it probably is.
It's even more sad when you see a PSU use leaded solder and the joints are still bad.

I also swapped out some of the tiny capacitors on the primary side before testing, because they were testing with really high ESR (compared to new/used parts)

I don't blame you.
Small Teapo caps always tend to be problematic after a number of years.