V-POWER ATX-S500 rebuild

[Pentium4]

One of my beloved Sun Pro units! And this is a good one... Manufactured: January, 2007. I'm impressed that all of the caps are intact. They consist of only JEE and GoldLink. It's obviously not a 500W PSU, but I blame V-POWER, not Sun Pro. I'd just call this a reliable 300W PSU.

Uses a nice input plug, and a ceramic fuse on a holder. Input filter is good, consisting of 2 X caps, 3 Y caps, two coils. MOVs are missing. NTC is 15S050M. BR is 6A. Input caps are 680uF 200V JEE. They measured 668uF and 656uF. Switchers are D209L in a TO-3PN package. 5VSB switcher is a C3150.

I noticed that the 5VSB transformer is of good size, there is no critical cap, and they used a 51Ω minimum load resistor instead of the usual 22Ω for this platform that I've seen.

Uses MOSPEC S30D40C for 5V and 3.3V. Used a MOSPEC F20C20C Ultrafast on the 12V, which I already easily replaced with an ON-SEMI MBR30H100CTG. All I had to do was remove the main transformer instead of the whole heatsink (I wanted to see the markings anyways!) Since it's a schottky, the voltage loss is insanely lower! 0.76V @ 15A (25C) versus 1.3V @ 10A (25C) The increased efficiency will be nice because the fans likes to spin very slow. The silence is part of the reason I love these Sun Pro units so much. The secondary heatsink is very thick and has lots of fins so I don't think there will be any cooling issues. The toroid coils are good sized and have plenty of wire on them. Maybe Pete S can add more on that

So far I have only tested the caps on the 5VSB and 12V rail. To my amazement they all tested good. The only thing that required some work was redoing the soldering. It was pretty bad, but I've seen worse. There were a lot of blobs, and solder balls rolling around. Trimmed a few leads too.

The Te Bao fans both had their labels peeling off. There is one on the top for intake and the usual one for exhaust. They had some grease left in them but were getting dry due to the label pealing. I put oil in them and then glued the stickers back on there. All the wiring is 18AWG. It has a PCIE 6 pin but probably won't be using it. I'm building a HTPC for my brother, and I'm going to use this PSU. It will just be powering a Radeon HD 4670.

I'm not sure which cap I want to use for the 12V rail filter. There is only one spot. The original was a 3300uF 16V JEE IM 10x40mm. My options are Samxon RS 3300uF 16V 10x30mm, with 1675mA ripple capability, or Chemi-Con KYB 2200uF 16V 10x30mm, with 2230mA ripple capability. So basically it comes down to capacitance, or ripple. I know momaka would say go for the higher capacitance

Will post more pics of it when it's done

[PeteS in CA]

Re: V-POWER ATX-S500 rebuild

Nice copper on the O/P inductor. Wish the heatsinks were beefier, though having two fans might compensate. The % or water in the KYB is not as high as with "motherboard" caps, but using aqueous parts in PSUs still makes me nervous. It is becoming common, though.

[Wester547]

Re: V-POWER ATX-S500 rebuild

JEE, I never thought JEE and Goldink caps were still good to this day. I really like how those brands feel the need to advertise "LOW ESR" on the sleeve, as though they're trying to convince themselves. Advertising the temperature range on the sleeve is something characteristic of very old capacitors (as is "W.V", working voltage rather than just the voltage). Now, when you see "VENT" on the sleeve, you know that one day, you should probably run away.

So what you're saying is those Te Bao fans don't use a rubber plug but just a sticker for the seal. Doesn't sound good. Overall, looks like a good, upgrade-worthy PSU. I like the size of the main toroid.

Nice copper on the O/P inductor. Wish the heatsinks were beefier, though having two fans might compensate. The % or water in the KYB is not as high as with "motherboard" caps, but using aqueous parts in PSUs still makes me nervous. It is becoming common, though.

Well, I recall that some members here (mockingbird and kc8adu?) have seen 15-year old Rubycon water-base series (YXG and ZL, IIRC) test good. I'm pretty sure the Japanese majors use the right oxidizers, corrosion inhibitors, degassing agents, phosphate compounds, etc, in the electrolyte to stabilize the H2O's vehement tendencies (but high water content parts do have a lower pyrolysis temperature, E.G the boiling point), although nothing lasts forever, good caps included. I think of much greater concern than the % of water in the electrolyte is the pH balance. So long as the pH balance of the electrolyte has not drifted far into the alkaline range, it's theoretically possible for the electrolyte to correct any defects in the anodic oxide caused by the % of water (although that will consume more electrolyte). The oxide layer is resistant to chemical attacks as long as the pH balance is between 4.5 (acidic) and 8.5 (slightly basic).

If the pH balance does rise into the alkaline range (rather than remain balanced at a value of 7), the electrolyte will not be able to "reform" the dielectric and nothing will impede the formation of aluminum hydroxide and hydrogen gas rather than a stable alumina oxide (and the electrolyte will continue to attack the aluminum oxide layer, unimpeded, and the aluminum will dissolve into the electrolyte, drastically increasing capacitance and leakage current until the cap vents). So long as you are using good parts, though, none of that should happen. So I think KYB and RS are good choices.

[Pentium4]

Re: V-POWER ATX-S500 rebuild

JEE, I never thought JEE and Goldink caps were still good to this day.

Haha. I spoke too soon. The two 2200uF 10V GL caps on the 5V rail failed in the classic "Bad HN/M" way, and so did the one after the coil on the 3.3V rail. The JEE 1000uF 10V before the coil on the 3.3V and 3300uF 16V cap on the 12V rail tested good. GoldLink should be extremely embarrassed that JEE held up better than them! I also find it strange that the caps on the 5VSB tested fine.

So what you're saying is those Te Bao fans don't use a rubber plug but just a sticker for the seal. Doesn't sound good.

They did have a rubber plug, but for whatever reason, they bulged upwards causing the sticker to peel up and the grease to get out. They still spun okay but "rocked" noticeably better after I put oil in them and let them run for a while.

Overall, looks like a good, upgrade-worthy PSU. I like the size of the main toroid.

I thought so too Sun Pro usually did use good sized toroids. You saw the one on that other one I posted? It has a MASSIVE toroid!

So I think KYB and RS are good choices.

I decided to use the RS here. Like momaka always said, capacitance is usually the best route to go (within reason) so even though there is only one filtering cap + PI coil for the 12V rail, 3300uF should keep it happy, and I don't think it will need more than 1675 of ripple filtering. The machine it's powering is:
Pentium 4 641 (3.20GHz)
Recapped Intel D101GGC
2GB DDR2 533MHz (1GB x2)
Recapped HIS Radeon HD 4670 1GB (IceQ version)
500GB Seagate 7200.10
TP-Link TG-3269 PCI Gigabit card


It's going to be for my brother. It will be hooked up to the server I built him through a Gigabit switch to stream HD mkv files and Netflix

[Pentium4]

Re: V-POWER ATX-S500 rebuild

Here is after I finished working on it. It looks good! I made it 120V only, replaced 100% of the caps, even the bulk caps. Upgraded the 12V rectifier. Redid a lot of the soldering. Replaced the 51Ω minimum load resistor on the 5VSB rail with 100Ω and heatshrinked it. I also heatshrinked the 1200uF 6.3V KY cap on the 5VSB since it was so close to the heatsink and toroid. Oiled the fans and super-glued the stickers back on them. This PSU and computer should be pretty happy

The CPU should run really cool. It's a 65nm Cedar Mill P4 (D0 revision, 65W TDP) with this heatsink on it:

[Wester547]

Re: V-POWER ATX-S500 rebuild

Haha. I spoke too soon. The two 2200uF 10V GL caps on the 5V rail failed in the classic "Bad HN/M" way, and so did the one after the coil on the 3.3V rail.

Well, the bad HM/HN way references high leakage current, or a dramatically thinned or non-existent oxide (resulting in very high capacitance readings), usually for the aforementioned reason regarding water-base electrolyte or electrolyte whose pH has fallen out of balance (but it can also happen if a bulged capacitor has deformed enough to damage the dielectric, or if the electrolyte has broken down and the cap is no longer capable of reforming). It's not a failure mode that's unique to the faulty HM and HN caps, though, as it more or less began in droves with the "P50/P51" capacitor plague of the very early 2000s. This study from 2004 observes that the failing low impedance motherboard caps from Taiwan at that point were failing in a very different way than conventional electrolytics (capacitance rising rather than dropping over time and leakage current rising).

The JEE 1000uF 10V before the coil on the 3.3V and 3300uF 16V cap on the 12V rail tested good. GoldLink should be extremely embarrassed that JEE held up better than them! I also find it strange that the caps on the 5VSB tested fine.

Maybe because the PSU was used 24/7 so the fan was keeping the +5VSB output caps cooler? But it just goes to show that all the "naughty" brands are inconsistent in quality.

They did have a rubber plug, but for whatever reason, they bulged upwards causing the sticker to peel up and the grease to get out. They still spun okay but "rocked" noticeably better after I put oil in them and let them run for a while.

It sounds like the seal wasn't very tight to begin with which gave leeway for the oil to leak from the bearing.

I decided to use the RS here. Like momaka always said, capacitance is usually the best route to go (within reason) so even though there is only one filtering cap + PI coil for the 12V rail, 3300uF should keep it happy, and I don't think it will need more than 1675 of ripple filtering.

That depends on the output frequency of the PSU. Half-bridge PSUs usually are of a lower frequency on the output which means larger inductors are a necessity and that bulk capacitance matters more. But as the switching frequency goes up (like in forward converters), the ESR becomes more important than the bulk capacitance and you can get away with smaller inductors (although the impedance / ripple current handling ability of the output caps will matter more).

All that said, nice work! The "after" shots look great.