Power supply build quality pictorial. part 2

[Wester547]

Re: Power supply build quality pictorial. part 2

I'm not familiar with the 2SC3320, but in Googling it I see its rated for 15A, pretty impressive compared to the 12A MJE13009 which could be adequate.

I wonder if that model was hard-wired for one voltage range or the other. Does the label say it's wide-range? I know STMicro produced an IC that sensed the I/P voltage and did the selection automatically, using a triac. That's another possibility.

The label does say it supports an AC input voltage of 100V-240V. As for the triac, there is one on the sister PCB, a glass passivated PNPN triac device by ST (BTA16-600B) - perhaps this is what senses the input voltage and does automatic selection.

[Wester547]

Re: Power supply build quality pictorial. part 2

Hmm, on second thought it looks like that Win-tact is actually an Enhance unit as the transformer markings bear great resemblance to another unit posted by goodpsusearch and the general layout is quite similar (it also uses an Enhance secondary supervisory chip). Seeing as how "RD-DW" transformer markings have been found in Deer PSUs as well, I'm wondering if this is actually a "high-end" Deer built for Enhance, rebadged by Win-tact... although those Deer units look quite different otherwise (the markings on the transformers are different as well aside from "RD-DW") so maybe Deer just tried to copy Enhance so as to look reputable... or they bought transformers from Enhance.

[momaka]

Re: Power supply build quality pictorial. part 2

^ Nice power supply momaka. Mitac is the company who produced the "Bluford" model motherboards for the Dell Dimension 2300/2350/2400 series PCs.

Interesting. Didn't know that. I have a ripped Dimension 2400 board (I think). (Someone ripped the entire back port area of the board ). The board has beautiful leaded soldering, though. I tried removing the (damaged) CPU socket on my gas stove/burner, and it lifted very very easily.

Never seen a Lelon with a honeycomb vent until now either

Well, just FYU, those Lelon capacitors were rated 85C, IIRC.

And now another retro PSU... a Win-Tact Electronics Corp. WP603A12 300W PSU

Now that's a great keeper PSU for old 5V-based socket A systems. Just add a Radeon 9700/9800 video card for maximum power draw from the 5V rail.

The primary switchers are biased by two Sanyo OS-CON SC 2.2uF 50V solid capacitors

That is the first time I see solid capacitors with such high voltage rating. Also the first time I see solid capacitors used in such an old PSU. Pretty unique PSU . I like the add-on board too. I wonder what functions it performs with that TRIAC on it

I forgot to mention that the PSU does not have a voltage selector switch, so I'd guess that it uses not only wide range flyback topology but wide range half bridge topology as well (or something like that?).

Actually, now that you said this, I think this is what the TRIAC might be for - automatic line voltage detection and selection.

[Wester547]

Re: Power supply build quality pictorial. part 2

Interesting. Didn't know that. I have a ripped Dimension 2400 board (I think). (Someone ripped the entire back port area of the board ). The board has beautiful leaded soldering, though. I tried removing the (damaged) CPU socket on my gas stove/burner, and it lifted very very easily.

Interesting because someone else commented that the Dell Dimension 2300 series boards (produced by Mitac as well around the same timeframe IIRC) at least used lead free solder in the VRM area.

Well, just FYU, those Lelon capacitors were rated 85C, IIRC.

I've seen Lelon REA 85*C series with cross-hatched vents before. I'd chalk up the alternate vent to inconsistency.

Now that's a great keeper PSU for old 5V-based socket A systems. Just add a Radeon 9700/9800 video card for maximum power draw from the 5V rail.

Or a Dell OEM Radeon 9700 TX? I have one of those which still works even with the original Nichicon HC laminated case (sleeveless) capacitors (the ADDA hyprobearing fan still works on it too.... for now). I say "even" because it was around the faulty Nichicon HM/HN era (2001-2005, the capacitors on that card were manufactured during the 3rd week of October, 2002, just like those on mockingbird's). I noticed that even some Nichicon HDs weren't safe from premature failure around that time period.

That is the first time I see solid capacitors with such high voltage rating. Also the first time I see solid capacitors used in such an old PSU. Pretty unique PSU . I like the add-on board too. I wonder what functions it performs with that TRIAC on it

I later clarified that those Sanyo OS-CON SCs are rated for 25V and not 50V.

Actually, now that you said this, I think this is what the TRIAC might be for - automatic line voltage detection and selection.

Yeah, Pete clarified that and that was the conclusion I also came to earlier in the thread. The TRIAC automatically detects the line voltage. Hence the lack of a voltage selector switch.

[momaka]

Re: Power supply build quality pictorial. part 2

Interesting because someone else commented that the Dell Dimension 2300 series boards (produced by Mitac as well around the same timeframe IIRC) at least used lead free solder in the VRM area.

Okay, I just checked a bit more on that; turns out, what I have is an OptiPlex GX260 motherboard (SDRAM slots). IIRC, that is either Intel or Foxconn... I'm leaning towards Foxconn, though.

Or a Dell OEM Radeon 9700 TX? I have one of those which still works even with the original Nichicon HC laminated case (sleeveless) capacitors

The sleevless Nichicon HC capacitors are pretty reliable. I have not seen a bulged one up to date and I've browsed through a lot of ATI hardware on eBay and elsewhere. As for HD, yes there are cases where they bulged / went bad. But they are nowhere near as bad as the faulty HM and HN series. Just a tiny bit more sensitive to heat than usual, it seems.

[dmill89]

Re: Power supply build quality pictorial. part 2

Now that's a great keeper PSU for old 5V-based socket A systems. Just add a Radeon 9700/9800 video card for maximum power draw from the 5V rail.

Win-Tact built the "Turbo-Cool" series for PC Power & Cooling as well, ridiculously overbuilt is pretty standard for them. Their older units often didn't have the best caps, but still lasted since the caps weren't stressed, the more recent stuff is usually 100% Japanese caps.

I posted the Turbo-Cool 425W a while back, insanely overbuilt for a 425W unit: https://www.badcaps.net/forum/showpo...&postcount=409

[Wester547]

Re: Power supply build quality pictorial. part 2

Okay, I just checked a bit more on that; turns out, what I have is an OptiPlex GX260 motherboard (SDRAM slots). IIRC, that is either Intel or Foxconn... I'm leaning towards Foxconn, though.

As I understand it, Foxconn was the OEM for Intel from 2001 until a few years back. However, Foxconn is a HUGE ODM (original design manufacturer). They own a number of motherboard assets but don't actually have enough resources to manufacture all their own motherboards so they may contract some OEM manufacturers like Flextronics (the actual OEM for some of the old PCI slot expansion boards) to produce the boards despite that the boards may be of Foxconn's design (and they may also oversee the final assembly) and may have Foxconn components on them.

The sleevless Nichicon HC capacitors are pretty reliable. I have not seen a bulged one up to date and I've browsed through a lot of ATI hardware on eBay and elsewhere. As for HD, yes there are cases where they bulged / went bad. But they are nowhere near as bad as the faulty HM and HN series. Just a tiny bit more sensitive to heat than usual, it seems.

The lack of a sleeve does pose a potential heat hazard for both the HC and HD series. The PVC or PET sleeve actually plays a big part in keeping the capacitors cool and acts as a "mini heatsink" of sorts for the aluminum cans (see the attached file here). I can confirm this, because the sleeveless 10x12.5mm 470uF 10V Nichicon HC capacitors do get significantly hotter than the PVC sleeved 470uF 10V Rubycon ZA capacitors on my Radeon 9500 Pro. mockingbird did have some Nichicon HCs bulge on his 9700TX, but the GPU fan had seized as well as the CPU fan and it came out of a poorly cooled machine.

I conveyed concern for those old HDs and HCs because there have been some reports of unused (unsoldered) and genuine HDs that bulged in storage and read very high capacitance (high VLOSS and leakage current), just like the faulty HMs and HNs. The older HDs are indeed more sensitive to heat than KZE and ZL (but the same could be said of Sanyo/Suncon WX).

On the other hand, I've seen a ton of blown 3300uF 6.3V Nichicon HMs and KZGs (in parallel with polymers) on old RDRAM based Intel D850EMV/MV/MD/EMD boards, and the CPU VRM output is of a very similar design to that of the Dell Dimension 8250 motherboard, which uses 1500uF 6.3V KZE or sleeveless HD in their place (again in parallel with polymers). Never seen a blown HD or KZE in that position. Haven't seen a blown Nichicon HD on the original Xbox motherboards either (but many blown HMs in the VRM area) in the VRM area. Haven't seen a blown HD on the old Dell Dimension 4300 or Optiplex GX240 series motherboards either, SFF or otherwise (but MANY blown HMs on those in the VRM output) so I suppose they are okay capacitors from that time period so long as you keep them cool and don't let them sit on the self for too long (this is not counting the 8mm 1500uF 6.3V HDs which seem to be prone to failure in my experience).

Win-Tact built the "Turbo-Cool" series for PC Power & Cooling as well, ridiculously overbuilt is pretty standard for them. Their older units often didn't have the best caps, but still lasted since the caps weren't stressed, the more recent stuff is usually 100% Japanese caps.

I posted the Turbo-Cool 425W a while back, insanely overbuilt for a 425W unit: https://www.badcaps.net/forum/showpo...&postcount=409

It appears the PSU I posted is actually made by Enhance Electronics and not Win-Tact, although it's branded by them.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Something new from me:



Interesting thing:
It has TWO OTP circuits...

[Behemot]

Re: Power supply build quality pictorial. part 2

The PVC or PET sleeve actually plays a big part in keeping the capacitors cool and acts as a "mini heatsink" of sorts for the aluminum cans (see the attached file here).

Srsly? Without the sleeving the temperature is slightly over 1 °C higher. With some airflow it is further 1 °C lower. That is hardly close to "significant" difference as that Vishay guy says. I mean yeah, in relative values (self-heating over ambient) it is tens of %. But in absolute values it makes no difference whatsoever if the cap has 64 or 66 °C. Absolutely beyond compare larger difference in the capacitor life makes if you design the appliance such a way it will not reach 60 °C ambient inside in the first place. I have not read the whole material but if it is not said in there it is just BS paper. If all their people are such experts it explains all the crap they produce.

Everybody who listened during high school physics lecture knows radiation plays only minuscule role (in single percents at most) of the heat transfer at normal temperatures up to small hundres of °C. Later about the point things start radiating visible light it shifts and it is slightly more, but still low. It starts to be signifficant at temperatures where high-energy radiation starts emitting, like with ultra-violet.

That is also why there are such problems with heat in space, though the temperature is close to absolute zero, there is actually nothing to transfer the heat to. And since radiation is very ineffective you have to use huge surface area…

[Per Hansson]

Re: Power supply build quality pictorial. part 2

Axpertec AXP-750P14A
Manufacturers specifications: https://web.archive.org/web/20090302...power_750W.asp
Review: http://extreme.pcgameshardware.de/an...-750-watt.html

Didn't see this brand of PSU here before.
Has been in a pretty highend PC for 5 years.
User asked me to upgrade it and first thing out is this PSU, getting replaced by a Corsair HX750i
Caps are:
2x 3300uF 16v CapXon GF 10x35mm
4x 3300uF 10v CapXon GF 10x25mm
1x 2200uF 16v CapXon GF 10x25mm (5vsb)
I desoldered them and their ESR is not terrible but getting a bit high after these years of use...

Curiously absent is U10, we don't need no steenking OPP or OCP on our "750w" PSU!

[Behemot]

Re: Power supply build quality pictorial. part 2

That's pretty normal for cheapest units to have only the basic 8pin supervisor. If the primary is not underrated and has HW limit, under overload the unit usually turns off as soon as the secondary voltages drop low enough. For real PC that usually means it gets unstable long before that and freezes or restarts, effectivelly removing the overload

[Wester547]

Re: Power supply build quality pictorial. part 2

Srsly? Without the sleeving the temperature is slightly over 1 °C higher. With some airflow it is further 1 °C lower. That is hardly close to "significant" difference as that Vishay guy says. I mean yeah, in relative values (self-heating over ambient) it is tens of %. But in absolute values it makes no difference whatsoever if the cap has 64 or 66 °C. Absolutely beyond compare larger difference in the capacitor life makes if you design the appliance such a way it will not reach 60 °C ambient inside in the first place. I have not read the whole material but if it is not said in there it is just BS paper. If all their people are such experts it explains all the crap they produce.

64˚C vs. 66˚C may seem tiny but for the life of the cap, by way of the arrhenius rule, it would probably mean roughly a +10% increase in lifespan for the cap. The article argues that the thermal resistance increases quite a bit (+48%) without the sleeve, so the cap's ability to dissipate heat is limited without it. It's true that plastic cannot be used as a heatsink for silicon despite the fact that it radiates heat better than aluminum in this case. I suppose the difference would disappear once one adds air cooling. And the article does stem from the product marketing department... I suppose that makes it a tad suspect. Still, the difference in temperature between the sleeved Rubycon ZAs and sleeveless, laminated case Nichicon HCs was larger than that on my old card... more like 4˚C.

Everybody who listened during high school physics lecture knows radiation plays only minuscule role (in single percents at most) of the heat transfer at normal temperatures up to small hundres of °C. Later about the point things start radiating visible light it shifts and it is slightly more, but still low. It starts to be signifficant at temperatures where high-energy radiation starts emitting, like with ultra-violet.

It's true that convection (the transfer of thermal energy by the movement of fluids... or air or liquid cooling) and conduction (transferring heat through particles and electrons, or in many cases a thermal compound and heatsink) play bigger roles in heat transfer than radiation. Thermal conductivity would highly favor metal (or copper) over plastic. Thermal convection would favor ambient temperature over all else (heat is also conducted through the PCB traces and substrate). Thermal radiation has to do with electromagnetic wave emissions and how those waves remove heat (thermal energy) from whatever object (in this case the cap).

That is also why there are such problems with heat in space, though the temperature is close to absolute zero, there is actually nothing to transfer the heat to. And since radiation is very ineffective you have to use huge surface area…

Yes, the surface area of the top of the can plays a larger part in dissipating heat. Larger diameter caps will run cooler than smaller diameter caps.

[ChaosLegionnaire]

Re: Power supply build quality pictorial. part 2

what about sticking something like a small mosfet heatsink on top of the cap to help cool it even further? will that make a difference or is it another one of my terribly bird-brained ideas? since the heatsink may block the cap from bloating the vent to relieve the pressure and the cap blows up explosively as a result of that?

[Behemot]

Re: Power supply build quality pictorial. part 2

What's 10 % if designing it right means double or quadruple the time?

Plus as has been even noted in the paper, if there are hotter objects around the cap, it actually warms it instead of cooling. And there usually are, the losses in capacitors are usually much smaller than the losses in semiconductors or transformers, if we talk about e. g. PSU. In such case the cap is heated instead of cooled.

So in reality it plays a role in case there is heavily stressed cap in a midle of nowhere where there is nothing else to heat it. Like say a cap around the PCI/PCIe slots on a motherboard. But that's never the case, if a cap is heavily stressed, it means it is moving vast amounts of power. So in case of the cap near the slots, it is not stressed thus it does not heat internally.

To cut the losses in such cases where the capacitors are stressed, than they are placed very close to the source (or demand side) of the power. That is…wait for it…to hot objects like transistors in CPU VRM, which themselves are very close to the CPU (as with the ~1 V voltages, the currents reach 80 A easily). Than the caps are heated instead of cooled! And they die anyway, because the whole thing is just hot. Nobody cares if they bloat in, say, 60 or 62 months, that's no difference. But hey, if you add a strip of aluminium heatsink to the transistors, guess what. The temperature drops by 20, sometimes 30 °C (some boards get discolored by the FETs how hot they are) and you get 10+ years from the caps instead of 5!

So what I mean is, this paper is a nice theoretical exercise. With close to no impact in real world. If they instead tought them manufacturers (=their customers) to design the stuff to not bake the caps in the first place, it will be much better. On the other hand, do the manufacturers want the HW to last longer? I doubt so, warranty engineering is todays norm.

ChaosLegionnaire: as stated, the losses in the capacitors are quite small compared to anything else. So in reality the point is to achieve a state where the caps are heated as little as possible from everything else around them.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Pretty neat Super Flower made PSU. The U10 could possibly be an LM339.
The sad thing is that they advertized the PSU as a quad rail while it's just a Single Rail...

By the way: I have a similar thing here, though mine has other caps than yours: JP CE TUL...
So I'd say yours has better caps overall than mine does...

€dit:
Yip, that U10 is absolutely for an LM339, 95% sure it is!!

PS: and it's from 30th week 09...

[momaka]

Re: Power supply build quality pictorial. part 2

Axpertec AXP-750P14A
...
Didn't see this brand of PSU here before.
Has been in a pretty highend PC for 5 years.
User asked me to upgrade it and first thing out is this PSU, getting replaced by a Corsair HX750i
Caps are:
2x 3300uF 16v CapXon GF 10x35mm
4x 3300uF 10v CapXon GF 10x25mm
1x 2200uF 16v CapXon GF 10x25mm (5vsb)
I desoldered them and their ESR is not terrible but getting a bit high after these years of use...

That still looks like a very decent PSU. Definitely worth recapping. Are you going to?

The only thing I am really worried about is the CapXon cap on the primary side, as they are known to go open-circuit or high-ESR a lot more often than other brands and cause the APFC to malfunction rather violently.

It's also nice to see a high power PSU that doesn't use SMD components. Not that I have anything against SMD (in fact, I am so used to SMD/SMT stuff now, that I sometimes prefer reworking it over through-hole), but I still like to see my PSUs SMD-less.

By the way: I have a similar thing here, though mine has other caps than yours: JP CE TUL...
So I'd say yours has better caps overall than mine does...

I don't know... between CapXon and JPCE TUL/TUR (CEC holdings), I think I would rather go with JPCE.

[dmill89]

Re: Power supply build quality pictorial. part 2

I don't know... between CapXon and JPCE TUL/TUR (CEC holdings), I think I would rather go with JPCE.

X2, neither is spectacular but CEC (JPCE) seems to be a lot better than CapXon, I've got some older (10+ years) Enermax PSUs with CEC caps that a still have good CEC caps (and one that I had to recap, the PSU was 8-10 years old at the time of recapping, and only the 5vsb was bad, but I replaced them all to be safe) while I've yet to see a PSU more than 5 years old (that was actually used regularly) with CapXons that didn't have at least one failed cap.

[Per Hansson]

Re: Power supply build quality pictorial. part 2

Stefan Payne: Ok interesting, maybe would be a fun project to try and add a LM339 to it then
P.S: I'd take your JPCE caps over my CapXon's anyday!

momaka: Yea I agree on the primary cap, I have seen several dead CapXon primaries!
This one tested ok though, 400uF 0.25ESR and 0.5% leakage.
I added a PCB pic after desoldering the primary, now we see the revision:

 HXW REV:D
SF-08/03/22
SF-APFC2.33PCB-A

Not sure if I'll recap it yet, need to buy some caps from Behemot then, because I don't have any D10mm 3300uF 16v caps...

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Mine has the Same HW Version and a CapXon Primary...

And also a 420V/390uF one...

[Behemot]

Re: Power supply build quality pictorial. part 2

Think you'll need to also add all the feedback components. If you don't know the right values in the best case it will do nothing even when it is supposed to trigger come protection. In worst case it will trigger them even when everything will be fine.