Power supply build quality pictorial. part 2

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

How much does it cost?

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Impressive.
HEC/Compucase build quality has definitely stepped it up a few notches from what I mostly saw from them a decade ago.

Yeah, after their first, rather shitty (too compact) try on an LLC Resonant Mode Converter, this one is way way better.
I should have posted it here, the GX V3 - wich had the Cap over the Cap ont he Secondary (a Teapo SC on top of a Polymer cap)...

Ah, I see. Thanks for the explanation. Didn't cross my mind to look for the (lack of) output toroid.

LLC Resonant Mode converter is the Default today
There aren't many new ones with other designs...
The good old Double Forward thingy is dying, the Designs you see right now are rather old, even the ones used in Cooler Master Master Watt and Cougar LX, that one was just improved with two Buck Converter Modules (with just one Dual MOSFET)...

Yes, LLC-resonant explains a lot how this PSU can deliver so much power out of so few components. I think that's also the same (or very similar) topology used in LCD monitor CCFL inverters as well as those small halogen light switching PSUs.

Well, its used more and more everywhere in the last 10 Years or so, it replaced Double Forward everyhwere where you need a bit more Wattage...

Even on Game Consoles like Playstation 3, 4 and of course Laptop PSU...

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

And another one...

I really don't know who made it. No Idea, really

[momaka]

Re: Power supply build quality pictorial. part 2

And another one...

I really don't know who made it. No Idea, really

Oh, what a mystery indeed!

Unlike those newer designs which make me wonder where to look to find "things", with this PSU, I felt just "at home". And needless to say, it took me less than 3 seconds of viewing the second shot before I figured who made it. Also, the output Fuhjyyu caps were another big giveaway. But I won't spoil the fun here and let other folks try to figure out who made it.

LLC Resonant Mode converter is the Default today
There aren't many new ones with other designs...
The good old Double Forward thingy is dying, the Designs you see right now are rather old, even the ones used in Cooler Master Master Watt and Cougar LX, that one was just improved with two Buck Converter Modules (with just one Dual MOSFET)...

But it's not important what's "modern"/fashionable today. Rather, the question is, which design will still be working 10-15 years down the road?
My single-transistor forward PSUs are still puttering after more than a decade of use - most still with their original crappy cap brands too.

Even on Game Consoles like Playstation 3, 4 and of course Laptop PSU...

Yeah, the PS3 I know. First time I took a few apart many years ago had me stumped at what was going on inside. I still have a small stack of failed Nichicon PS3 PSUs. The Sony ones are okay, though. Speaking of which, here are both...

First the Nichicon ZSSR539IA 400 Watt power supply.
https://www.badcaps.net/forum/attach...1&d=1526270481
https://www.badcaps.net/forum/attach...1&d=1526270481
https://www.badcaps.net/forum/attach...1&d=1526270481
https://www.badcaps.net/forum/attach...1&d=1526270481
This one is dead, IIRC. Many of these appeared to die due to overheating and/or bad design. On most, the output MOSFET rectifiers went bad. And on a few, the 5VSB also disappeared. One in my pile also possibly might have a bad APFC circuit too. I was going to fix all of these at some point... but never really got to that due to their compact design, which makes troubleshooting not so much fun.

Now compare that to the Sony APS-226 - it's the same 400W power supply, but much better designed, IMO:
https://www.badcaps.net/forum/attach...1&d=1526270481
https://www.badcaps.net/forum/attach...1&d=1526270481
https://www.badcaps.net/forum/attach...1&d=1526270481
Also runs much cooler under load.

Anyways, sorry for not posting a picture with the casing and info on it, but I had a hard time making my camera focus on the cases. But in any case, they are simple to tell apart. IIRC, the Sony had square vent holes, and the Nichicion PSU had round holes. There might have been another producer of these PSUs, but I don't recall who.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Oh, what a mystery indeed!

Unlike those newer designs which make me wonder where to look to find "things", with this PSU, I felt just "at home". And needless to say, it took me less than 3 seconds of viewing the second shot before I figured who made it. Also, the output Fuhjyyu caps were another big giveaway. But I won't spoil the fun here and let other folks try to figure out who made it.

^^
Got it with the A7M-266
The one that seems to kill TBred B for whatever reason...

Though I'm not planning on using it with that Board...

But it's not important what's "modern"/fashionable today. Rather, the question is, which design will still be working 10-15 years down the road?

Nobody knows, that's the Problem
And you don't know what is the Problem...

This one is dead, IIRC. Many of these appeared to die due to overheating and/or bad design. On most, the output MOSFET rectifiers went bad. And on a few, the 5VSB also disappeared. One in my pile also possibly might have a bad APFC circuit too. I was going to fix all of these at some point... but never really got to that due to their compact design, which makes troubleshooting not so much fun.

Yeah, the Heatsink over half the PSU is kinda weird and makes working on it really bad it seems.
The PSU of the Slim and Super Slim is much much simpler...

And also seems like LLC-Resonant mode...


Are the Sony and Nichicon PSU interchangable??
Have you tested the efficiency of both??

Its possible that the efficiency of one of those might be a bit better and that's why it lasts longer...


So I'd say that that's a construction error...

[momaka]

Re: Power supply build quality pictorial. part 2

Are the Sony and Nichicon PSU interchangable??

Yes.
Both are rated for 32 Amps on their 12V rail and 3 Amps on the 5V standby.
In fact, here are the two label shots for the Nichicon and Sony PSUs above, respectively:



And here are the PSUs with their covers on side-by-side.

The Nichicon PSU is the one on the left with all of the round holes. The Sony has the square holes - that's actually one way to quickly tell them apart. I think there was another manufacturer for these PS3 PSUs that used the round holes, though they were arranged in a slightly different pattern, IIRC. So just keep that in mind if you do see a PS3 PSU with similar (but not exactly the same) circular holes.

And also seems like LLC-Resonant mode...

Actually, I'm not quite sure - at least when it comes to the Nichicon PSU. I got curious yesterday and took it apart to have a look at the main PS topology. While at it, I started drawing a circuit schematic... and before I knew it, I had a basic schematic of the main PS (i.e. the switch driving circuits, APFC section, and 5VSB are not drawn). So here is that:


As you can see, there IS an output toroid on the secondary side - L201. It is under the secondary heatsink and looks like a transformer. So I'm not sure if this changes anything or not, as I am not so well-versed in LLC topologies. But looking at the primary side, if we ignore Q101 and C101, the primary side looks pretty much like a single-transistor topology, minus a snubber circuit across the transformer. Instead of a snubber, we get Q101 and C101. So if I am understanding this correctly, after parallel MOSFETs Q102 and Q106 are switched On and the Off, Q101 engages and puts capacitor C101 across the primary side of the transformer, thus creating a resonant LC circuit.

That said, don't take the circuit above as 100% accurate. I may have missed something or not drawn it correctly. So consider this a preliminary schematic.

While on the subject matter of details about the Nichicon PSUs, here are a few other things I observed...
APFC MOSFETs (Q003, Q004): 2SK3683
APFC diode (D004): UHF10JT
APFC fuses (F003, F004): T4A 250V
Primary-side bulk capacitors: 2x Nichicon (series?), 450V, 140 uF
Output-side capacitors: all are Nichicon HV, with the 5VSB filtered by a 3900 uF cap.

Also, this was the PSU that had dead 5VSB (as some of you may have noticed by the text from the case shot above). Turns out fuse F002 was blown. This fuse not only connects to the main PS MOSFETs, but also the 5VSB offline switching IC. So removing F002 and putting a wire across it (note: I used an incandescent bulb across the PSU input) brought the 5VSB back to life. Of course, the main PS wasn't working. And in fact, before I replaced F002 with a wire, I removed MOSFETs Q101, Q102, and Q106, because Q101 and Q102 were shorted (yet Q106 appears to have survived ) . It is interesting to note here that while Q102 and Q106 are in parallel, they each have their own gate driver circuits - hence why I labeled them on the schematic above as "A1" and "A2". Perhaps that's why Q106 didn't die.

Meanwhile, on another dead Nichicon PSU that I have, it is MOSFETs Q202 and Q203 (both IRFB4310) on the output-side that have gone short-circuit. On that note, I can't quite figure out how the secondary-side rectification works... unless perhaps I made a tracing error? Even tough I double-checked everything several times, there are a few ambiguous traces that run under the secondary heatsink that I had to pretty much figure out where they ran to.

Have you tested the efficiency of both??

Unfortunately, no. It didn't cross my mind to do that back when I still had access to many of these units in working condition (that is, back when I worked in a console/PC repair shop). But the Nichicon PSUs always seemed to run much hotter than the Sony's, especially when we tested PS3s with their top plastic cover off (which I later found out is required to keep the PS3 PSU cooled).

Its possible that the efficiency of one of those might be a bit better and that's why it lasts longer...

Yup, that's what I suspect too. Actually, in the case of the Nichicon PSUs, it seems that the main issue is hot-spots developing near the MOSFET driving circuits. So perhaps when things get hot enough, that's what causes these PSUs to fail.

Nobody knows, that's the Problem
And you don't know what is the Problem...

Oh, I know actually.
Classic half-bridge topologies have proven themselves for so many years now (at least from the well-built PSUs anyways - not the crappy cheapo PSUs, even though those can last a very long time too, with good caps). The only problem with them is pretty low efficiency. Of course, if you live in a place with cooler climate, that shouldn't matter too much.
Single-transistor forward topologies have also proven themselves, if there is no APFC to complicate things. Two-transistor forward also seems to be okay... but overall, I don't consider it as reliable due to slightly higher complexity.

P.S. and since no one seems to want to guess the PSU you posted in post #2903 above... well, I'll be the party-pooper and spoil the fun : Channel Well Technology (CWT), as obviously written on all of the transformers.
Was this re-labeled by Antec, by the way? Looks to under-built to be an old Antec, IMO. And the case ventilation sucks, judging by the lack of vent holes.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Both are rated for 32 Amps on their 12V rail and 3 Amps on the 5V standby.

400W PSU??
that's really high...

Actually, I'm not quite sure - at least when it comes to the Nichicon PSU. I got curious yesterday and took it apart to have a look at the main PS topology. While at it, I started drawing a circuit schematic... and before I knew it, I had a basic schematic of the main PS (i.e. the switch driving circuits, APFC section, and 5VSB are not drawn). So here is that:

That doesn't look like LLC Resonant mode...
If you want to know more about those, Texas Instruments should have some Datasheets

But also the usual suspects:
CM6901
ICE2HS01G
NCP1910 (doesn't seem to have many diagramms)


But anyway, do you have any information about the big chip on the Primary side (lower right side of the PCB), IC101.
That one could explain what Topology is used...

As you can see, there IS an output toroid on the secondary side - L201. It is under the secondary heatsink and looks like a transformer. So I'm not sure if this changes anything or not, as I am not so well-versed in LLC topologies.

I haven't seen LLC-Resonant mode Converters with Output toroids. But the Original PS3 is from 2006 or so, so its possible that its anormal PWM one...
LLC-Resonant mode was rather uncommon at the time.
Though the PS3 Slim and SUper Slim PSU seems to be LLC-Resonant mode as well as the PS4 one.

That said, don't take the circuit above as 100% accurate. I may have missed something or not drawn it correctly. So consider this a preliminary schematic.

Yeah, it might be a normal PWM one, maybe a standard Double Forward, maybe something like Active Clamb. Though we are talking about 2006...

Unfortunately, no. It didn't cross my mind to do that back when I still had access to many of these units in working condition

That would have been really interesting and could help.
But I doubt both of those reach 90% Efficiency, 85% maybe.

P.S. and since no one seems to want to guess the PSU you posted in post #2903 above... well, I'll be the party-pooper and spoil the fun : Channel Well Technology (CWT), as obviously written on all of the transformers.
Was this re-labeled by Antec, by the way? Looks to under-built to be an old Antec, IMO. And the case ventilation sucks, judging by the lack of vent holes.

Yeah, its hard to miss the Markings

Though its an Everpower and onls rated 200W:
10A 3,3V, 20A +5V and 6A +12V.

[RukyCon]

Re: Power supply build quality pictorial. part 2

Delta also made PS3 power supplies.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The quasi-resonant designs I've seen were full bridge and phase modulated (e.g. the UCC3875). The "output inductor" // resonant inductor was the transformer leakage inductance plus an additional inductor on the primary side. The topology also has zero-voltage switching, so power is not dissipated during turn-on, improving efficiency.

[momaka]

Astec AA21470 - 256 Watt PSU

You know what? Forget all this modern LLC resonant stuff. Let’s post another classic reliable workhorse PSU: an Astec AA21470. I doubt many of you have seen this one, as it’s not a unit commonly found in regular PCs. It’s actually the PSU that came with my dumpster-picked HP NetServer E-800 some years ago and still is its PSU. I almost trashed the thing after I found the PC had roaches in it before (but not when I got it). Eventually decided to clean everything after giving the whole PC a RAID treatment – no, not the kind with multiple hard drives. Like our fellow member Junk Parts (RIP) used to say, the only RAID I’ve used came from a spray can.

I’ll start with a fan shot. Why? - I don’t know… because I never do? Hopefully this might clue you in on the build quality, though.
https://www.badcaps.net/forum/attach...1&d=1526878844
… a Nidec BETA SL D08T-12PG. Good stuffs.

Next, a shot of the back of the PSU:
https://www.badcaps.net/forum/attach...1&d=1526878844
It looks almost like a standard ATX PSU, but I think it’s a little different. For one thing, I don’t think the screw holes match to fit in a standard ATX case.

And here is the front side with vents:
https://www.badcaps.net/forum/attach...1&d=1526878844

Finally, the last of the case shot: the label.
https://www.badcaps.net/forum/attach...1&d=1526878844
Yup, definitely a “classic” unit with a 5V-heavy design, rated for 30 Amps and 180 Watts total on the 5V + 3.3V rails. Meanwhile, the 12V rail gets only a puny 6 Amp rating… which is odd, because this PSU has, if I recall, 6 or 7 Molex drive connectors. If each of these is connected to an HDD or an optical drive, I’m not sure how the 12V rail will manage all of that. But currently, I do have it loaded with two optical drives and two HDDs (7k and 10k RPM in staggered spin-up) and it’s not sagging too much. Also note another interesting thing from that label shot above: double-fusing - great for those ambiguous-Live Schuko plugs.

Not let’s undress her
https://www.badcaps.net/forum/attach...1&d=1526878844
Don’t know about you, but I like what I am seeing here – nice heatsinks, good cap brands, and in general a very well-laid out PCB. Also great input filtering. Speaking of which… some of you may have noticed from the case shots before that there is no input voltage selector switch, and given the big red toroid in the upper left corner of the above picture, you might conclude that this PSU has an APFC circuit. However, this would be an incorrect assumption. Instead, this PSU has two TRIACs that automatically sense the input line and switch to 115/230V operation – neat!

Moving onto the primary side…
https://www.badcaps.net/forum/attach...1&d=1526878844
… we can see the main transformer, kinds of hidden under the primary heatsink. It is a proper 35 mm core, of course. To the right of that, we also get two Rubycon MXR caps, rated for 200V and 1000 uF – all of that for a 256 Watt PSU!!! Impressive? I think so!

Another angle, showing some of the secondary-side caps…
https://www.badcaps.net/forum/attach...1&d=1526878844
… tall, proud Nichicon PM’s with a lone United Chemicon LXV (or two) for diversity. You can tell this PSU was made at a time when lower-ESR PSU-grade caps like Chemicon KY, Nichicon HE, and Rubycon ZL weren’t available or popular just yet, because most of those Nichicon PM’s have a much greater voltage rating than needed – a move probably done to decrease the ESR of the caps while maintaining a certain capacity.

You can also see the bundle of wires coming out of the PSU, which are all 18 AWG and UL-approved (E117873) rated for 300V and 90°C.

Primary-side 5VSB area:
https://www.badcaps.net/forum/attach...1&d=1526878844
As can be seen from this shot, there are a lot of small square polypropylene and polyester caps used throughout rather than cheaper ceramics (or worse: electrolytics). Oh, and the 5VSB has it's own PWM-FET IC - no 2-transistor unprotected circuits in this PS!

No PSU post is complete without solder-side shots, so here goes that:
https://www.badcaps.net/forum/attach...1&d=1526878844
Nice, clean solder work, Astec! Definitely gives Delta and HiPro a run for their money. The little side/daughter-board looks just as good:
https://www.badcaps.net/forum/attach...1&d=1526878844

Final thoughts: it’s one well-designed and well-built PSU. I would certainly trust it to power an older (5V-heavy) PC for many more years.
______________________________________________
And now for a more detailed component breakdown…

Primary Side
5x 2.2 nF safety Y-class caps, 2x 0.22 uF safety X-class caps, 1x CM choke + 2x SM chokes
2x 10 Amp ceramic fuses (for L and N), 3x MOVs (1x between L-N, and 1x across each primary cap)
GBU8j bridge rectivier, 2x BTB12 800B TRIACs for 115/230V auto-switching
2x Rubycon MXR, 200 V, 1000 uF, 25 x 50 mm (ø x h)
1x ST W9NB90 MOSFET (TO-247) for main PS (single-transistor forward)
35 mm main transformer + 19 mm 5VSB transformer

Secondary Side
5V Rail:
- 2x STPS3045CP (TO-247) rectifier
- 1x Nichicon PM, 35V, 1000 uF, 12.5 x 30 mm  before PI coil
1x Nichicon PM, 25V, 2200 uF, 12.5 x 40 mm  after PI coil
- 6-turn, 5 mm core PI coil

12V Rail:
- 1x ??? 10 Amp??? (TO-220) rectifier
- 1x Nichicon PM, 25V, 2200 uF, 12.5 x 40 mm  no PI coil

3.3V Rail:
- 1x MBR2045CT (TO-220) rectifier
- 1x Nichicon PM, 35V, 1000 uF, 12.5 x 30 mm  high side of buck VRM?
1x Chemicon LXV, 6.3V, 3900 uF, 12.5 x 40 mm  low side, before PI coil
1x Nichicon PM, 6.3V, 2200 uF, 10 x 30 mm  low side, before PI coil
- 18-turn, 10 mm core PI coil

-12V Rail:
- 1x FE3D(?) 3 Amp(?) diode
- 1x Chemicon __, __V, ____ uF, 10 x 25 mm  on 7912 linear regulator input
1x Nichicon PM, 25V, 220 uF, 6.3 x 13 mm  on 7912 linear regulator output and input to -5V rail’s 7905 regulator

-5V Rail:
- 1x FE3D(?) 3 Amp(?) diode
- 1x Nichicon PM, 25V, 220 uF, 6.3 x 13 mm  on 7905 linear regulator output and input to -5V rail’s 7905 regulator

5VSB Rail:
- 1x SB36(?) 3A diode
- 2x Nichicon PM, 16V, 470 uF, 8 x 20 mm  with PI coil in between
- 15-turn, 4 mm PI coil

ICs, Primary Side
LM2903 (input-voltage sensing), AS2214F PWM IC (main PS), MC33365) (5VSB PWM-FET IC), 3x 617A opto-isolators

ICs, Secondary Side
AS2333E (supervisor IC?), LM2901 (OCP?), LM2902 (UVP/OVP/OCP?), AS3843 (PWM-IC for 5V  3.3V rail buck converter… equivalent to UC3843)

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Well, since the PSU was made in mid-2000, and the fan has what appears to be a mid-2005 datecode, I'm going to guess that is not the original fan.

That .75A rating on the 5VSB was pretty wimpy even for 2000. That's going to limit how you use that PSU, sadly.

1000uF for a 256W PSU, HP must've wanted crazy hold-up time (Rubycon, very nice). All those box film caps are nice. PW series output caps - double nice: good brand; excellent series for 2000. The PCB design year seems to have been 1999.

I wouldn't call the heatsinks spectacular, but definitely adequate. Astec made good stuff. That wimpy 5VSB output is unfortunate.

[Wester547]

Re: Power supply build quality pictorial. part 2

I think the datecode on that fan (“0525E”) actually means May of 2000 or so. Looking at this same fan model from an old Hipro PSU yanz posted nearly nine years ago, the datecode on the main transformer in that Hipro is “0137” (week 37, 2001), but Nidec’s datecode format seems to differ (“1913K” probably means September 2001).

OTOH, the “0028” datecode on that Astec label uses the more conventional YYWW datecode format (week 28, 2000). The Nichicon series seems to be PM rather than PW. I’m not even sure if a PSU this old would tolerate PW grade capacitors on the output (what with the series on the output filter being PM and NCC LXV). And IIRC, the STPS3045CP rectifier is of the SOT-93 package rather than TO-247 (in accordance to the datasheet).

The measly 256W rating given such beefy silicon and overspec’d components must be because of the aforementioned heatsinks. But I still think this PSU could easily do its rating at an ambient temperature of 50ºC.

[momaka]

Re: Power supply build quality pictorial. part 2

Well, since the PSU was made in mid-2000, and the fan has what appears to be a mid-2005 datecode, I'm going to guess that is not the original fan.

PSU had never been opened before, so most likely the date code on the fan has a different format, as Wester suggested.

That .75A rating on the 5VSB was pretty wimpy even for 2000. That's going to limit how you use that PSU, sadly.

Not really.
It's not until you go into the late socket 478 Pentium 4 era and socket 754 AMD that you need more than 1 Amp of power on the 5VSB. It mostly has to do with powered USB ports and sleep function (socket 462 AMD does not support CPU S3 state, so PSU keeps running in sleep/stand-by mode, and socket 370 and earlier boards with SDRAM need very little power to preserve RAM in system S3 state). In particular, power for USB on most early boards came from the 5V rail, not 5VSB, so the 5VSB only needed to provide power to the mobo for sleep function (RAM Vtt rail) and possibly WOL/WOR (Wake-on-LAN and Wake-on-Ring/modem). Power-On by Mouse/Keyboard also doesn't require much power (~100 mA max for most keyboards). Thus, 0.75 Amps is enough for most older boards made up until the early 2000's.

Also, don't forget this 0.75A rating is an Astec rating, not cheapo PSU rating. Means the PSU will do 0.75A sustained and probably 1 Amps peak. Same goes for the 6A rating on the 12V rail: even with two 10k HDDs, it never dropped out of spec, despite the high current draw at start-up (over 5A with all drives).

And lastly, this PSU will be staying in the HP Netserver that it came with. So if HP determined that this PSU was enough for the system, it's probably okay to continue using it this way.

1000uF for a 256W PSU, HP must've wanted crazy hold-up time (Rubycon, very nice).

When I unplug the PSU, the LAN activity LED on the motherboard stays lit for almost a minute before going out. I don't recall what the hold up time is when the power goes out with the PC turned on (I rarely run that PC, so I actually don't recall the power ever going out when the PC was on). It's also amusing that this 256W PSU has more primary capacity than my Corsair CX750M.

Astec made good stuff.

Yup, I can't find a PSU from them that had any sort of issue. It's too bad I don't see them making PSUs anymore - at least not in consumer-grade stuff.

I’m not even sure if a PSU this old would tolerate PW grade capacitors on the output (what with the series on the output filter being PM and NCC LXV).

I don't think there's that much difference between PM and PW. Also, this being a more "modern" topology (STF vs. the commonly used h-bridge at the time), it probably would work fine with PW.
..............................
Forgot to answer this one regarding the PS3 PSU:

But anyway, do you have any information about the big chip on the Primary side (lower right side of the PCB), IC101.
That one could explain what Topology is used...

IC101: UPC1909G by NEC

... so looks like you guessed it right: it's probably an active clamp -type PS.

meanwhile, IC100 is an IR1150 by Infineon, responsible for PFC control:


And IC202 on the secondary side is a 10358 dual op-amp.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

... The Nichicon series seems to be PM rather than PW. I'm not even sure if a PSU this old would tolerate PW grade capacitors on the output (what with the series on the output filter being PM and NCC LXV). ...

The measly 256W rating given such beefy silicon and overspec'd components must be because of the aforementioned heatsinks. But I still think this PSU could easily do its rating at an ambient temperature of 50ºC.

Yeah, that was a misread on my part. I partially corrected it yesterday, but didn't change the "W" to an "M". The PM series and LXV series both seem to be in production. As old as they are (I think the pre-RoHS PL series came out 30 years ago), I guess they still sell too well to discontinue the series.

momaka mentioned that the PSU was in an HP server, so HP probably went for a PSU that would stay operating in the field for a long time.

[Wester547]

Re: Power supply build quality pictorial. part 2

When I unplug the PSU, the LAN activity LED on the motherboard stays lit for almost a minute before going out. I don't recall what the hold up time is when the power goes out with the PC turned on (I rarely run that PC, so I actually don't recall the power ever going out when the PC was on). It's also amusing that this 256W PSU has more primary capacity than my Corsair CX750M.

I have some Newton Power 250Ws with only ~280uF total bulk capacitance on the input (560uF 200V x2 in series), and after unplugging the PSU, motherboard LED stays lit for about 30 seconds so that sounds about right. Modern PSUs use synchronous-buck converters and very low VF schottkys, (and very low RDS(on)) MOSFETs), so I guess that helps somewhat, but PSUs with APFC don't have always the greatest hold-up time since less bulk capacitance is required in order to store the same amount of energy with a single 400/450V capacitor than with two 200V capacitors in the voltage doubler.

I don't think there's that much difference between PM and PW. Also, this being a more "modern" topology (STF vs. the commonly used h-bridge at the time), it probably would work fine with PW.

Yup, that was a flub on my part. You could use 3900uF 6.3V and 1000uF 25V Nichicon PWs (which are only slightly shorter in terms of case size) to replace the PMs with near identical specs (and a 3900uF 6.3V LXZ to replace the LXV as well, again, only slightly shorter but almost the same specs). Same goes for 470uF 16V Nichicon PWs replacing the 470uF 16V Nichicon PMs. A 220uF 25V Nichicon PW could replace the PM (only slightly shorter) with pretty close specs (but I don't think it would matter much, being on the input and output of a -5V linear regulator). That said, it may be simpler to just replace the PMs with newer PMs since the series is still in production, and same goes for LXV since that series hasn't been EOLed yet.

Then again, both series are non-quaternary salt non-aqueous capacitors, so they may still be good for a couple years or so if the rubber bungs have held up.

[kevin!]

Re: Power supply build quality pictorial. part 2

A friend gave me an old dell optiplex, I took out the PSU of 200w that was originally LITEON, and I put it on my computer for backup, it has a core 2 duo e8400, 3 HDD, and keeps the 12v lane 11.90v, a marvel. Like the lane from 5vsb to 4.98v, 3.3va to 3.34v, and 5v to 5,098v.
A very stable PSU.
I leave these photos, I love seeing this PSU so well built.
A pity that there is no space to see the IC, switch, diodes, switch, I would like to show it, but I do not want to destroy the PSU.
It only takes 4 LTEC of poor quality, which will be exchanged for one rubycon, or Nippon chemicon or similar of high quality.

Bye

[momaka]

Re: Power supply build quality pictorial. part 2

A friend gave me an old dell optiplex, I took out the PSU of 200w that was originally LITEON, and I put it on my computer for backup, it has a core 2 duo e8400, 3 HDD, and keeps the 12v lane 11.90v, a marvel. Like the lane from 5vsb to 4.98v, 3.3va to 3.34v, and 5v to 5,098v.
A very stable PSU.

Nice PSU Kevin!
The only thing to beware about many LiteON PSUs is that they use the tan (brown when heated over time) glue that goes conductive. Otherwise, they are solid PSs and very stable, like you noted. Just have to remove the tan glue (as much of it as possible - especially from the primary side) to make them even more reliable.

[kevin!]

Re: Power supply build quality pictorial. part 2

Nice PSU Kevin!
The only thing to beware about many LiteON PSUs is that they use the tan (brown when heated over time) glue that goes conductive. Otherwise, they are solid PSs and very stable, like you noted. Just have to remove the tan glue (as much of it as possible - especially from the primary side) to make them even more reliable.

I did not know, I will remove the glue with urgency.
thank you very much momaka for your advice. .

[Sparkey55]

Re: Power supply build quality pictorial. part 2

I did not know, I will remove the glue with urgency.
thank you very much momaka for your advice. .

Since I have bought lots of PSU both brand new and well used I have seen all stages of this killer glue. Brand new it is a yellow color changing to a light tan then a light brown then a dark brown or black. It is the later stages of brown to black that is most conductive. This glue also shrinks a lot when drying up. It goes from being semi-soft to extreme brittle in its life cycle. best to remove it while it is yellow or tan in color.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

That Optiplex PSU is interesting, still 5V/3.3V heavy, but with a healthy 5VSB current capability (2A).