ISO-500 Rebuild

[Pentium4]

Classic CWT ISO series. Manufactured December, 2006. This one is pretty nice. I figured it was worth upgrading since it has a 39 transformer and decent heatsinks. And it's actually got a thermistor for the fan. Most just run at full speed all the time.

All the caps consist of Fuhjyyu, Koshin, and two Teapo SEK. The only caps that tested bad were the Fuhjjyu TM on the 5VSB, and a 1uF 50V Koshin cap

Original input caps were 820uF Fuhjyyu. The bridge rectifier was physically touching one of them! Oddly, the one being touched read 822uF, and the other one read 727uF. Nice that they installed MOVs. Switchers are 2SC2625 rated at 15A. Nice. NTC is SCK-058.

For the rectifiers, it used a 30A schottky for 3.3V, two 20A schottkys in parallel for the 5V (I prefer this over a single 30A) and a 16A ultra fast on the 12V.

The worst thing about it was the cables. 20 pin motherboard, 4 pin CPU, 6 molex, 2 berg.

This is how it looked before I got my hands on it.

[Pentium4]

Re: ISO-500 Rebuild

And now for the mods! First of all, I removed a lot of the glue. It had mostly that nasty tan stuff. I did a pretty good job removing just about all of it. I bent the bridge rectifier away from the input caps. Replaced the original input caps with 820uF 220V Elite caps pulled from a LiteON. They read 822uF and 818uF. Pretty impressive for 7 year old caps. As you can see I made it permanently 120V. The voltage selector switch was short so it was putting stress on some input components. I used that wire then on the bottom since it was 18AWG and well insulated.

Naturally, I replaced all the electrolytic caps. I did something a little different for one though. C25 on the secondary was a 0.22uF 50V Fuhjyyu TN cap. The board was originally silkscreened for a ceramic cap there. I don't have 0.22uF 50V caps so I put a 0.1uF 100V ceramic cap on the top of the board, and then one on the solder side, since they both wouldn't fit through the top. This makes me feel better, since I never want any Fuhjyyu in my equipment.

All of the rails have a PI coil, including the -5V and -12V.

I then replace the 16A ultra fast on the 12V with a 30A schottky. I didn't like that there was that dent in the heatsink right where the rectifier goes, but it should make enough contact with the heatsink. If there was a second slot for a 12V rectifier I would have added another one in parallel. 30A should be enough though. The 12V had a 250Ω minimum load resistor, I decided to leave it alone. The 5VSB had 100Ω, 3.3V had 15Ω, and 5V had 50Ω. I didn't have as good of a selection as I thought, so I ended up leaving the 5VSB one alone, and put 100Ω on both the 3.3V and 5V. Better than they were at least.

Since there was plenty of room, I added a salvaged 4 pin from another PSU to make this ones motherboard connector a detachable 20 + 4 pin connector. I also added a sleeved 3 SATA port power cable.

I also did some basic things such as making the fuse flush to the board, fixing some of the poorly done solder joints, and removing flux. I'm also going to cut out the punch grille and add a wired one + oil the Top Motor fan. It looks pretty good now, and ready for many years of service.

[Behemot]

Re: ISO-500 Rebuild

Seems like you are putting Samxons to a good use

[goodpsusearch]

Re: ISO-500 Rebuild

And now for the mods! First of all, I removed a lot of the glue. It had mostly that nasty tan stuff. I did a pretty good job removing just about all of it.

Me too... I have worked on 2 Thermaltake TR2-420 psus and now I am on the third that is blown. Than glue is a pain to remove...

Naturally, I replaced all the electrolytic caps. I did something a little different for one though. C25 on the secondary was a 0.22uF 50V Fuhjyyu TN cap. The board was originally silkscreened for a ceramic cap there. I don't have 0.22uF 50V caps so I put a 0.1uF 100V ceramic cap on the top of the board, and then one on the solder side, since they both wouldn't fit through the top. This makes me feel better, since I never want any Fuhjyyu in my equipment.

Nice!!

Since there was plenty of room, I added a salvaged 4 pin from another PSU to make this ones motherboard connector a detachable 20 + 4 pin connector. I also added a sleeved 3 SATA port power cable.

I do those things too!

I also add a second -12V cap since there is an empty spot on pcb.

There are also 2 spots for 12V filtering caps, 1x3300uF and 1x100uF(!).
I replace the 100uF with 1000uF 16V entry level esr withjout any oscillation problems.

Congrats! This psu is perfect now! Please put it in a 12V demanding PC and tell us how it does with voltage regulation.

[Pentium4]

Re: ISO-500 Rebuild

Seems like you are putting Samxons to a good use

Works great in this PSU. Original Fuhjyyu cap only had ~950mA ripple current capability, that's way too low! I bought 130 from you, and I think I have about 10 left. If I keep up this pace I will be buying some of those Chemi-Con caps from you.

Me too... I have worked on 2 Thermaltake TR2-420 psus and now I am on the third that is blown. Than glue is a pain to remove...

Yeah, it's such a pain! I recently saw one killed by glue as well, it was worst on the secondary by the supervisory IC resistors/zeners.

I do those things too!

I also add a second -12V cap since there is an empty spot on pcb.

There are also 2 spots for 12V filtering caps, 1x3300uF and 1x100uF(!).
I replace the 100uF with 1000uF 16V entry level esr withjout any oscillation problems.

Congrats! This psu is perfect now! Please put it in a 12V demanding PC and tell us how it does with voltage regulation.

Yeah, it makes the PSU so much better adding modern cables to them! I left the -12V alone on mine since it has a coil and a 220uF cap. The original second filtering cap for the 12V was Fuhjyyu 330uF 16V TN. I replaced it with a 8x15mm 470uF 16V. It has about 3 times the ripple current capability. I will load it up with a Pentium 4 machine and tell you how it does

[momaka]

Re: ISO-500 Rebuild

Classic CWT ISO series. Manufactured December, 2006. This one is pretty nice. I figured it was worth upgrading since it has a 39 transformer and decent heatsinks. And it's actually got a thermistor for the fan. Most just run at full speed all the time.

Indeed worth saving.
It's like my CyberLink and Turbolink, but actually with all of the missing and proper components installed.

Original input caps were 820uF Fuhjyyu. The bridge rectifier was physically touching one of them! Oddly, the one being touched read 822uF, and the other one read 727uF.

I doubt the bridge rectifier ran that hot to matter. And even if it did, I'd still take them over crappy MK that read HALF (!) of their 330 uF capacitance

... two 20A schottkys in parallel for the 5V (I prefer this over a single 30A)

Why is that?

The worst thing about it was the cables. 20 pin motherboard, 4 pin CPU, 6 molex, 2 berg.

I guess yours is the "delux" version, so you get 2 extra molex connectors over my TurboLink. Everything else is the same, though .
SATA connector? Who needs THOSE?

The 12V had a 250Ω minimum load resistor, I decided to leave it alone. The 5VSB had 100Ω, 3.3V had 15Ω, and 5V had 50Ω. I didn't have as good of a selection as I thought, so I ended up leaving the 5VSB one alone, and put 100Ω on both the 3.3V and 5V. Better than they were at least.

If you will be using this in a 12V-based system, then you can take out the load resistor completely or just have something "symbolic" in there (like 1 KOhm or so ). Definitely helps that you removed that 15 Ohm resistor from the 3.3V rail though. It was dissipating 0.667 Watts. My threshold is 0.5 Watts for load resistors, and they must be rated for at least 2W if they would be dissipating that much power.

Original Fuhjyyu cap only had ~950mA ripple current capability, that's way too low!

For PC PSUs (i.e. not flyback topologies), capacitance is the most important item. Then ESR.
That said, CWT PSUs do seem to like killing Fuhjyyu caps more than any other PSU brands. I wonder
The Fuhjyyu caps I pulled from my Macron MPT-301 were still showing good capacitance and ESR last I checked them about 2 years ago. I think I even put one of them on the 5VSB of one HiPro I have.

I bought 130 from you, and I think I have about 10 left.

Holy Moly, what do you do? Eat caps for breakfast? I bought some caps way back in 2009 and I still a few left . I know... keeping them on the shelf this long is not good. But that is as fast as I work.

BTW, curious question for you, Pentium4: I see some "Weller" packaging. Is that from the new soldering station you got?

[Pentium4]

Re: ISO-500 Rebuild

I doubt the bridge rectifier ran that hot to matter.

But still, I don't like the idea of a Fuhjyyu cap being used as a "heatsink"

And even if it did, I'd still take them over crappy MK that read HALF (!) of their 330 uF capacitance

As if 330uF wasn't bad enough already!

Why is that?

Even at low loads, wouldn't splitting the load between the two 20A schottky packs create higher efficiency than a single 30A schottky?

If you will be using this in a 12V-based system, then you can take out the load resistor completely or just have something "symbolic" in there

It would have been silly to double the 12V rail capability and then use it in a 5V based machine It will definitely be going into a 12V based machine, so I decided to remove it completely. The 12V hasn't done anything funky without it.

capacitance is the most important item. Then ESR.

Really? This makes me feel like I've done some projects in the past improperly now. So, you're saying a PSU would do better with a 2200uF 16V Fuhjyyu TN with 960 RC, compared to a 1500uF 16V nichicon PW with 1815 RC? That doesn't make sense

The Fuhjyyu caps I pulled from my Macron MPT-301 were still showing good capacitance and ESR last I checked them about 2 years ago. I think I even put one of them on the 5VSB of one HiPro I have.

Are you some sort of adrenaline junkie, or thrill seeker? You must like to live dangerously

Holy Moly, what do you do? Eat caps for breakfast? I bought some caps way back in 2009 and I still a few left

Haha, not even close. That's only 120 in almost 2 years. SIDMX has gone through a lot more than I have!

BTW, curious question for you, Pentium4: I see some "Weller" packaging. Is that from the new soldering station you got?

It sure is I posted it here: https://www.badcaps.net/forum/showthread.php?t=46638

[Wester547]

Re: ISO-500 Rebuild

Really? This makes me feel like I've done some projects in the past improperly now. So, you're saying a PSU would do better with a 2200uF 16V Fuhjyyu TN with 960 RC, compared to a 1500uF 16V nichicon PW with 1815 RC? That doesn't make sense

There's no such thing as too high a ripple rating. But some control loops on the secondary of some PSUs simply won't tolerate capacitors that are lower ESR than the originals by a significant amount. And of course it won't do well with Fuhjyyu.

[goodpsusearch]

Re: ISO-500 Rebuild

Really? This makes me feel like I've done some projects in the past improperly now. So, you're saying a PSU would do better with a 2200uF 16V Fuhjyyu TN with 960 RC, compared to a 1500uF 16V nichicon PW with 1815 RC? That doesn't make sense

They both make sense for power supplies. The esr is important in ripple filtering along with the pi coil in series and the capacitance is important for transient response.

[momaka]

Re: ISO-500 Rebuild

Even at low loads, wouldn't splitting the load between the two 20A schottky packs create higher efficiency than a single 30A schottky?

Not always. It depends on the device and manufacturer. Probably best to look at the datasheet and see the Forward Voltage vs. Forward Current graphs and compare them for a certain Current value that interests you.

Let's say the 12V rail is to handle 20A nominal. So you have the choice of either two 20A schottky diodes or a single 30A diode. If you choose the first option, then there will be (theoretically) 10A going through each rectifier. If you choose 30A rectifier, the 20A will be going all through it. So you go in the datasheets for these rectifiers and you look up the voltage drops at their respective currents (i.e. 10A for the 20A rectifier and 20A for the 30A rectifier.).
Let's say the 20A rectifier has 0.5V Vf at 10A and the 30A rectifier has 0.6V Vf/ Then the two 20A rectifiers will be dissipating 0.5V x 10A x two = 10W. And the 30A rectifier will be dissipating 0.6V x 20A x one = 12W. In that case, the 30A part will make the PSU less efficient.

But that is assuming the load is equally split between the two 20A rectifiers. And to have that, you need to have very good thermal coupling between the two rectifiers. Any small difference in temperature between the two will make the hotter rectifier Vf drop more, thus making it conduct more current... which in turn will heat it up more and make it conduct even more current.

Really? This makes me feel like I've done some projects in the past improperly now. So, you're saying a PSU would do better with a 2200uF 16V Fuhjyyu TN with 960 RC, compared to a 1500uF 16V nichicon PW with 1815 RC? That doesn't make sense

The ripple current is more of a rating that tells us how much ripple (i.e. abuse) the cap can take before it starts overheating.

The capacitance is what keeps the PSU voltages stable between output pulses from the transformer. So the more capacitance you have, the smoother and more stable the voltage will be.

Of course, the capacitance won't matter too much if the ESR of the caps is too high.
If the Fuhjyyu you mentioned still has its ESR in spec, then likely it will give better filtering and transient response. But it also depends on what the PSU is designed for too and how much of a variation on ESR it can take. Ideally, you want to keep the ESR close to what the PSU had originally. If the caps ESR is too high, then the PSU will filter less noise. But if the ESR is too low, then the noise will be filtered, but the caps may create "new" noise due to ringing effects (the caps and inductor/PI coils and ESR of the caps form a tuned RLC filter).
So in essence, there is a sweet spot for the ESR of the caps. Meanwhile, for capacitance: bigger is usually better.

Are you some sort of adrenaline junkie, or thrill seeker? You must like to live dangerously

I just have too much old computer stuff, so sometimes I do these things for experimentation.
I actually checked that HiPro PSU two days ago. I remember what happened now: I put that Fuhjyyu cap temporarily on the 5VSB but then took it out shortly afterwards and put back the original Asiacon cap after I confirmed it was okay with an ESR meter. Then again, IIRC, Fuhjyyu and Asiacon are like cousins or something (if not twins ).

[Wester547]

Re: ISO-500 Rebuild

Then again, IIRC, Fuhjyyu and Asiacon are like cousins or something (if not twins ).

Fuhjyyu has no relation to Asiacon whatsoever, in all seriousness. Fuhjyyu = JunFu = AsiaX. Asiacon = Hermei, whose parent company, Kaimei, is also Jamicon's. There was a very awful brand called "Arcon" which Hipro used back in 2003 for a short while (worse than CapXon), and I honestly can't remember opening up a Hipro in which there wasn't a blown Arcon if an Arcon (if not that, they were all blown) was present in the unit.

The other bad brands they use are hit or miss (G-Luxon, Sus'con, Teapo, Asiacon, OST, LTEC, etc), but usually Teapo and OST go bad without showing it. And Teapo seem to be way too fragile for their own good so I'm more inclined to distrust them now adays. Another thing I noticed about the HP-P2507F3P is that the ones made in Thailand had Panasonic on the input and Asiacon on the output and the ones made in China had Teapo on the input and a mingling of Teapo, Asiacon, and G-Luxon on the output.

And that ISO-500 has a nicely sized +5V/+12V toroid.

[Behemot]

Re: ISO-500 Rebuild

The ripple current is more of a rating that tells us how much ripple (i.e. abuse) the cap can take before it starts overheating.

The capacitance is what keeps the PSU voltages stable between output pulses from the transformer. So the more capacitance you have, the smoother and more stable the voltage will be.

Of course, the capacitance won't matter too much if the ESR of the caps is too high.
If the Fuhjyyu you mentioned still has its ESR in spec, then likely it will give better filtering and transient response. But it also depends on what the PSU is designed for too and how much of a variation on ESR it can take. Ideally, you want to keep the ESR close to what the PSU had originally. If the caps ESR is too high, then the PSU will filter less noise. But if the ESR is too low, then the noise will be filtered, but the caps may create "new" noise due to ringing effects (the caps and inductor/PI coils and ESR of the caps form a tuned RLC filter).
So in essence, there is a sweet spot for the ESR of the caps. Meanwhile, for capacitance: bigger is usually better.

Yeah and that's why folks here suggest everybody very low ESR 2200uF caps (which, in this case, would have like 1/4 the ESR, definitelly not the "sweet spot" IMO) over 3300uF Samxons? Makes sense

[momaka]

Re: ISO-500 Rebuild

Fuhjyyu has no relation to Asiacon whatsoever, in all seriousness. Fuhjyyu = JunFu = AsiaX. Asiacon = Hermei, whose parent company, Kaimei, is also Jamicon's.

Ah okay. I always get them mixed. I think you've probably explained this to me (and others?) at least 3 times now, but I keep forgetting. I guess I remember the bottom line, though - they all can't be trusted.
That said, I'll still take Fuhjyyu over Hermei. Really, aside from CWT PSUs, I haven't seen Fuhjyyu fail in other PSUs ever. Not saying they are good (heh, not even by a long shot), but it seems that Channel Well definitely knows how to punish them.

And that ISO-500 has a nicely sized +5V/+12V toroid.

I see what you did there
Trying to appear on-topic with a single sentence

Yeah and that's why folks here suggest everybody very low ESR 2200uF caps (which, in this case, would have like 1/4 the ESR, definitelly not the "sweet spot" IMO) over 3300uF Samxons? Makes sense

You mean Panasonic FR and FM?
Yeah, they are not ideal for many PC PSUs, but they will work. Again, as long as you get the capacitance reasonably close to the original (or higher), a good deal of the PSUs will still work - even with very low ESR caps. Of course, they may end up outputting more ripple that before. But it is still unlikely to be way out of ATX spec. If you start getting PSU oscillations, that's when you know the ESR is too low.

Now, if you are dealing with flyback topologies, go for low ESR all you want. You might get annoying audible whine, but in many cases, the voltage regulation will actually improve. Try it. Put some polymer caps on the 5VSB of a PSU.

[Behemot]

Re: ISO-500 Rebuild

Yep. But I have Chemi-Con KYB on the way which are mostly on par with them and of course cheaper as always from me, so I expect you all stop whining and start making orders

Stand-by is often also a question of available power. With cheap underrated PWM ICs and one-way rectification with only one diode, you just have too low power pumped there to keep the votlage. Thats the case of Silverstone Strider Gold S/Gold V2.0 (by High Power) which drops like 0.2 V under load. I don't think any cap on the world would help that.

[Wester547]

Re: ISO-500 Rebuild

Ah okay. I always get them mixed. I think you've probably explained this to me (and others?) at least 3 times now, but I keep forgetting. I guess I remember the bottom line, though - they all can't be trusted.

Only twice on my behalf. AsiaX and Asiacon are easy to confuse, so it's not a big deal. I don't recall explaining it to others.

That said, I'll still take Fuhjyyu over Hermei. Really, aside from CWT PSUs, I haven't seen Fuhjyyu fail in other PSUs ever. Not saying they are good (heh, not even by a long shot), but it seems that Channel Well definitely knows how to punish them.

Well, other people's experiences are different with respect to that. But none of the brands are perfect anyway - the "good" brands' foul ups (I'm sure people know what I mean here) are an austere example as to that. And Teapo are definitely not to be trusted except maybe on the voltage doubler of a PSU that hasn't overheated. They "trick" people into thinking they're good because they fail silently very often. Also, you haven't seen another Fuhjyyu blown, or haven't found one that measured out of spec otherwise? Or both?

I see what you did there
Trying to appear on-topic with a single sentence

Son, don't roll your eyes at me.

Because let's face it, no one knows how to stay on-topic and off-topic at the same time like me.

I originally posted to this thread so as to answer P4's question about ripple currents, ESR, and the feedback loop on the output. But I guess I have more to say about this thread - that "nice" main toroid uses -26 material, so that's one sign of cheapness but quite common to be seen in Chinese half-bridges units anyway (though CWT is a Taiwanese company). The unit is "sort of" purported to be 500W and has a 39 size transformer, though, and 820uF primaries, so it better have good output toroids. And I also hate conductive glue. Those heatsinks aren't very impressive for a 500W unit either, but since it has strong rectifiers in a half-bridge configuration, it may not be an issue. I don't think those 2SC2625s are good enough for 500W, either, yet I know Unique has pulled 650W out of 13009s before.

[Behemot]

Re: ISO-500 Rebuild

Wasn't fuhjuu the Antec killer back in those days?

[goodpsusearch]

Re: ISO-500 Rebuild

Teapo SC are hit and miss. Sometimes I see them doing fine in psus for years, but other times they fail, especially if they are next to heatsinks / minimum load resistors / toroids etc. The worse place for them though is in the 5vsb output.

Oddly Teapo SEK that are GP caps and should not be used in psu outputs seem to do better.

[momaka]

Re: ISO-500 Rebuild

Only twice on my behalf. AsiaX and Asiacon are easy to confuse, so it's not a big deal.

And you remember even that.

Also, you haven't seen another Fuhjyyu blown, or haven't found one that measured out of spec otherwise? Or both?

Both.
I've seen Fuhjyyu in numerous non-CWT PSUs, and none appeared bad. Only was able to test a few, though... two of them being my own: a Linkworld with a good amount of hours (judging by the dust in it) and old Leadman/Powmax PSUs. In fact, the 16V 1000 uF Fuhjyyu caps from the PowMax showed good ESR and capacitance the last time I was able to borrow a meter and check them, which was after many years in storage.
As for my Inno Power (Macron) MPT301: that one had its fan seize and board discolor - thats how hot it got. And yet, it only had 2 or 3 failed Fuhjyyus on the output, just closest to the hot spots on the board. The rest were all fine and in spec. And still are.

Son, don't roll your eyes at me.

I don't think those 2SC2625s are good enough for 500W, either, yet I know Unique has pulled 650W out of 13009s before.

Yes, those 2SC2625 transistors certainly won't make it to 500W.
But IIRC, didn't the Unique pull that much power with a nearly scratch-made PSU (on which he rewound the original transformer)? AFAIK, he was able to get that because he really put some thought behind that transformer and the PSU voltage output was like 24 or 35V.

Anyways, I do have bad memory, so don't take my word for what I said there and/or feel free to correct anything.

Wasn't fuhjuu the Antec killer back in those days?

It was. And Antec was made by who back then?
See what I mean now?

[Wester547]

Re: ISO-500 Rebuild

As for my Inno Power (Macron) MPT301: that one had its fan seize and board discolor - thats how hot it got. And yet, it only had 2 or 3 failed Fuhjyyus on the output, just closest to the hot spots on the board. The rest were all fine and in spec. And still are.

I had something similar happen to a Hipro 250W. The ADDA fan seized for a while and the PSU was as hot as an oven (had roughly 17,000 hours on it) but none of the capacitors vented and most of them (except one tiny 2.2uF 50V Teapo SS) measured in spec, even the G-Luxon and Teapo. As a striking contrast, I managed to get my hands on another Hipro that was never used before. The ADDA fan in that was lubed well and most of the capacitors were fine except one 2200uF 10V G-Luxon HM on the +3.3V rail which bulged on its own and stunk badly. "Unpredictable" for these reasons is probably the best way to give definition to bad capacitors. Other's experiences with Fuhjyyu are testaments as to the same sort of "inconsistency".

Also, I don't think the amount of dust in a power supply is directly pertinent to running hours. Some people just don't do a very good job of cleaning the dust in their home, and it shows relatively quickly, IMO.

But IIRC, didn't the Unique pull that much power with a nearly scratch-made PSU (on which he rewound the original transformer)? AFAIK, he was able to get that because he really put some thought behind that transformer and the PSU voltage output was like 24 or 35V.

He managed to pull that off with an ambient temperature of 35*C, with a massively overspec'd secondary in terms of rectifiers. I don't know if the same is manageable at 115V (lower efficiency on the primary). And CWTs aren't the only Fuhjyyu killers. Older FSPs kill Fuhjyyus as well, though those PSUs aren't designed very well either IMO.

Truth be told, my own memory isn't very good, at least my short term memory. I think badcaps.net has "maxed" it out.

[Behemot]

Re: ISO-500 Rebuild

It was. And Antec was made by who back then?
See what I mean now?

Weren't some AcBel too? Those which did not even handle labeled power?

You don't have yur own ESR meter?