Are those two caps bad?

[goodpsusearch]

Re: Are those two caps bad?

If I were you I would replace the psu without buying a new unit. I would find a used psu for free and use that instead. All it takes is to know where to look for it.
It is hard to find worse psu than this!

But if I had to use that psu I would rebuild it. I would replace the 2 diodes on a bracket with a proper rectifier, I would replace some critical caps with Good Japanese caps and make sure the fan doesn't seize by lubricating it with oil.

You should completely remove the glue and make sure the caps are not touching sources of heat.

Better quality photos would help too.

[goodpsusearch]

Re: Are those two caps bad?

I wasn't sure if it would blow the switchers or not, but why would it do that? Excessive current pulses?

Those caps are part of the AC(mains) to DC converter circuit that is made of a bridge rectifier and the primary capacitors. The bridge rectifiers transforms the AC (sine wave) to positive pulses (that's why it's called rectifier!) and then the primary caps make it DC by filling the valleys between the pulses:

http://i.stack.imgur.com/LC3dA.png

Now imagine what happens if the primary caps dry and the transistors instead of DC voltage "see" the positive pulses.

[ben7]

Re: Are those two caps bad?

Those caps are part of the AC(mains) to DC converter circuit that is made of a bridge rectifier and the primary capacitors. The bridge rectifiers transforms the AC (sine wave) to positive pulses (that's why it's called rectifier!) and then the primary caps make it DC by filling the valleys between the pulses:

http://i.stack.imgur.com/LC3dA.png

Now imagine what happens if the primary caps dry and the transistors instead of DC voltage "see" the positive pulses.

Sorry, doesn't make sense, because you say they "see" the "positive" pulses. In order to fry them, you'd need to reverse bias them or exceed the SOA, or exceed die temp ratings.

[goodpsusearch]

Re: Are those two caps bad?

They see ripple and low voltage.

[ResistEnt]

Re: Are those two caps bad?

Hi again.

I got good news and bad news.

Good news:

• I completely removed the glue between components
• I managed to desolder the mains connector
• this way, I have been able to remove the PCB from its case
• I desoldered the bulged capacitor

The bulged capacitor is a 470uF 16V G-Luxon cap, and G-Luxon is the brand of most of the caps that are soldered on the PSU.
It's bulged either on the top and the bottom sides.

Now the bad news:

• while I was removing the glue, I slightly scraped a resistor
• while I was desoldering the mains connector, I didn't noticed I was also melting a capacitor of the line filter

The scraped resistor, that first was colored somewhat blue or light purple, has now a little black area from where the colored paint came off.
The capacitor is not electrolytic, it's a rectangular yellow cap of 0.33uF 250Vac but, to read more, I have to remove it.
It's not completely melted of course, I just touched it with the side of the iron tip. Damn...

Do you think I should consider also to replace both the melted cap and the scraped resistor? Paying a little more attention, of course...
I guess the damage is not so bad and the PSU should work anyway, but...

@goodpsusearch: can you tell me what capacitors you think are "critical"? I can't locate them... they're all the same from my point of view LOL
I'd like to try to repair the PSU anyway.
If I destroy it I'll search for a better one, but at least I'll have learnt a few more things.

I attached all relevant (at least, I think) sections of the PSU.
1,2,3 I'm moving from bottom to top, 4,5 I'm moving from left to right.
I'm sorry but I don't have a good camera, and my smartphone really sucks.

Thanks for your support, bye bye.

[SteveNielsen]

Re: Are those two caps bad?

I wouldn't worry about the cap. Can't see the resistor hardly at all. If it's just scraped and not broken in half it may be ok. Don't know. Look at the other side of it, the color bands go all way around. Measure it.

[goodpsusearch]

Re: Are those two caps bad?

I wouldn't worry about the resistor.

I am also quite confident that the yellow X series cap would be fine.

The 5vsb critical cap should be replaced with high quality Japanese capacitor. If that cap fails the 5vsb can go up to 10V or more and kill the motherboard!

It would be also a good move to change the 2 small capacitors that are used to pollarize the 2 NPN switching transistors.

Finally, the 2 x 5vsb output filter caps are the caps that fail more often in ATX psus. I would replace them with 2x1000uF low esr GOOD caps.

[ResistEnt]

Re: Are those two caps bad?

Hi.

@goodpsusearch: I removed all the caps you suggested me.

I also removed the cap in attachment because it was the only one that wasn't "in touch" with the PCB. So I thought: if it's not in touch, even if it appears good to the top, maybe it started to bulge to the bottom since it has room to do that.
All other caps are in touch with the PCB, so I'm assuming if they look good they're also good (at least, for now).

@SteveNielsen: trivial question, to correctly measure a resistance I have to first desolder it, right? Otherwise I'd measure the equivalent resistance seen at the poles I attach the tester to.

Anyway I was not sure the leads of the resistor were not broken (I bent them to desolder the cap), so I preferred to desolder the resistor too and it looks fine, apart the scraping. I measured it and it's about 100-110 Ohm.
Color code should be brown-black-brown then, which I'm pretty sure match the colored bands on the resistor.

I also attached a picture with the "melted" cap.

I'll post again as soon as I have a list of caps to buy.
Thanks for your replies, bye bye.

[goodpsusearch]

Re: Are those two caps bad?

Hi.

@goodpsusearch: I removed all the caps you suggested me.

I also removed the cap in attachment because it was the only one that wasn't "in touch" with the PCB. So I thought: if it's not in touch, even if it appears good to the top, maybe it started to bulge to the bottom since it has room to do that.
All other caps are in touch with the PCB, so I'm assuming if they look good they're also good (at least, for now).

Please do yourself a favor and never buy caps off ebay.

The reason is this:
https://www.badcaps.net/forum/showthread.php?t=46208
More than 90% of the caps that are sold on ebay are fake. Using fake japanese caps is worse than using cheap taiwanese caps. As you can see in the link above, even when they are not bulging they are not at all in specs.

Make sure the replacement caps are 105C and low esr (not too low esr!).

You don't have to stick with the original values of the caps that are used as part of the output voltage filtering. I bet this psu has mostly 1000uF capacitors for 3.3V, 5v and 12V outputs. You can replace those with 2200uF without any problem. Actually the psu will be better in terms of ripple filtering and transient load response.

@SteveNielsen: trivial question, to correctly measure a resistance I have to first desolder it, right? Otherwise I'd measure the equivalent resistance seen at the poles I attach the tester to.

Not necessarily. Personally, I first check the color code on the resistor and then measure it in circuit. If the value I get is the nominal then I take it as good.

If I get too low resistance or open circuit then I unsolder it to test it again.

Usually, I just lift one leg of the part I want to test off-circuit without having to completely remove it from PCB. Then I solder it back.

[ResistEnt]

Re: Are those two caps bad?

Thanks all for the information,
and sorry for my late reply but I've been quite busy these days.

Needless to say the PSU is still open on my desk and caps haven't been ordered yet lol

Anyway, I managed to have a look at some Rubycon datasheets and I took into account these parameters for my choice:

• capacitance, I tried to stay to the original values because I really don't know which parts of the PCB can see a different capacitance value and which can't;
• working voltage, not necessarily the same as the original but, in general, not lower than the original;
• diameter, I chose the same diameter to be sure the new caps will fit when I'll go to solder them; same thing for the pitch;
• ripple current, a value equal or greater than the original one; most of the datasheets report the ripple current at 100kHz and give some coefficients to scale it down for lower frequencies (with the lowest being 120Hz); I tried to stay happy with both ripple values (100kHz and 120Hz);
• ESR, lower than or equal to the original value, but not too lower;
• life time, higher is better (all caps I desoldered had a lifetime of barely 2000 to 3000 hours).

The parameter I don't know if I should take into account or not is the leakage current (lower should be better). If I choose capacitance and/or voltage higher than the original ones, the leakage current increases and sometimes it gets 3 times higher... Should I ignore it or not?

That said, it follows a list of the caps I have to replace, their characteristics (as per manufacturer datasheet) and the caps I'm going to replace them with.

1. the only bulged cap was a G-Luxon, 470uF 16V 8x15, ripple current 730 or 800mA, ESR 93 or 85mOhm (I don't know them exactly since I measured the length of the cap and it's not listed in the datasheet, so I took the params of the nearest formats). Replacement: Rubycon ZLJ 470uF 16V 8x11.5, ripple current 1200mA, ESR 75mOhm;
2. +5VSB critical cap, it was a G-Luxon 22uF 50V 5x11, ripple current 75mA (@120Hz), ESR unreported on datasheet. Replacement: Rubycon YXH 22uF 50V 5x11, ripple current 180mA (75.6mA @ 120Hz), ESR 700mOhm;
3. +5VSB filter caps, they were G-Luxon 1000uF 10V 8x20, ripple current 1080 or 1100mA, ESR 65 or 50mOhm. Replacement: Rubycon ZLJ 1000uF 10V 8x20, ripple current 1960mA, ESR 41mOhm;
4. switching transistor caps, they were CapXon 2.2uF 50V 5x11, ripple current 20mA (@120Hz), ESR unreported. Replacement: Rubycon YXF or YXJ 2.2uF 50V 5x11, ripple current 43mA (18.06mA @ 120Hz), ESR 2500mOhm. Ok it's very large, but in order to reduce it I'll have to go up with capacitance and/or working voltange (not sure if it can be done);
5. "extra" cap I removed, it was a G-Luxon 1000uF 10V 10x16, ripple current 990 or 1100mA, ESR 85 or 50mOhm. Replacement: Rubycon YXJ 1000uF 25V 10x20, ripple current 1400mA, ESR 46mOhm. Another possible replacement would be YXJ 2200uF 10V, same format and specs but lesser lifetime, involves changing capacitance value and I don't know if it can be done. Also, both these replacemens have a leakage current almost doubled with respect to the one of the original cap (220, 250uA versus 100uA). Reducing it would mean to reduce the ESR;
6. graphics card caps, they were Sacon 1000uF 16V 10x12.5, ripple current 1800mA, ESR 22mOhm. Replacement: Rubycon ZLH 1000uF 25V 10x23, ripple current 2250mA, ESR 18mOhm. Good replacements would also be ZLH 1000uF 16V 10x16 ESR 28mOhm, and ZL 1000uF 16V 10x20 ESR 23mOhm (this last one with a life time lower than the other's, 4000h versus 10000h).

That's all.
I thought to buy from RS, now I just gotta check if they have the above capacitors in stock.
If you, while reading, noticed something I'd better to know before buying, plz let me now.

Bye bye.

[goodpsusearch]

Re: Are those two caps bad?

"I tried to stay happy with both ripple values (100kHz and 120Hz);"

No! Only 100KHZ matter for psus.

" The parameter I don't know if I should take into account or not is the leakage current (lower should be better). If I choose capacitance and/or voltage higher than the original ones, the leakage current increases and sometimes it gets 3 times higher... Should I ignore it or not?"

Completely ignore it.

Ok, replacing caps on motherboards and graphic cards is harder because of the multilayer pcb and large copper paths that absorb heat quickly. Just wanted to let you know. Find a dead GPU or mobo and train first.

" "extra" cap I removed, it was a G-Luxon 1000uF 10V 10x16, ripple current 990 or 1100mA, ESR 85 or 50mOhm. Replacement: Rubycon YXJ 1000uF 25V 10x20, ripple current 1400mA, ESR 46mOhm. Another possible replacement would be YXJ 2200uF 10V, same format and specs but lesser lifetime, involves changing capacitance value and I don't know if it can be done. Also, both these replacemens have a leakage current almost doubled with respect to the one of the original cap (220, 250uA versus 100uA). Reducing it would mean to reduce the ESR;"

That psu would be happy with capacitance change on its Main Voltage output caps (3.3V, 5V, 12V).

Best scenario would be: 3.3V: 2x2200uF
5V: 2x2200uF
12V: 1x3300uF (if the pcb has only one spot for 12V) or 2x1500uF (if there are 2 spots on pcb)

On what voltage rail is the cap that you removed? You can easily tell from the soldering side of the psb. Just follow the traces from orange (3.3V), red (5V) and yellow (12V) wires.

[ResistEnt]

Re: Are those two caps bad?

Ok for 120Hz/100kHz.
Ok for leakage current.
Thanks for suggestion about the difficulty of soldering on mobos and gfx cards, I figured it out. I've an old MB I'd be happy to destroy lol

The extra cap looks directly connected to orange wires.
The bulged cap is directly connected to the blue wire (-12V?).

To orange wires seems also connected the cap below the extra cap (see pic 3 from the above post), but not directly.

Looking at the same picture, the capacitor partially hidden by orange wires, just above the extra cap, is directly connected to yellow wires.
Caps on the right are a real mess!

[goodpsusearch]

Re: Are those two caps bad?

Ok for 120Hz/100kHz.
Ok for leakage current.
Thanks for suggestion about the difficulty of soldering on mobos and gfx cards, I figured it out. I've an old MB I'd be happy to destroy lol

The extra cap looks directly connected to orange wires.
The bulged cap is directly connected to the blue wire (-12V?).

To orange wires seems also connected the cap below the extra cap (see pic 3 from the above post), but not directly.

Looking at the same picture, the capacitor partially hidden by orange wires, just above the extra cap, is directly connected to yellow wires.
Caps on the right are a real mess!

Now that I am not posting from a mobile phone I can write more about multilayer pcb soldering. You need a 40W or more capable soldering iron, lead solder and a needle. The stainless steel needle is necessary to remove the solder from holes. There is a guide somewhere in forum about that.


The output filtering of power supplies usually consists of:
cap in parallel
coil in series
and cap in parallel.
So, it is expected that 1 cap will be directly connected to the cables and the other would be right after the coil.

I attach a picture. This is my guess about the caps and the voltages. Can you confirm?

If it is correct, then this is my suggestion:

3.3V: 2x2200uF
5V: 2x2200uF
12V: 1x3300uF 16V

-5V: 470uF
-12V: 470uF 16V

I have recapped this platform before with the above values without any problem.

[ResistEnt]

Re: Are those two caps bad?

Okie thanks for your post.

I'll check it as soon as I can, don't know if I'll do it tomorrow.

But my guess about your guess is you're right, what you marked as "coils" are actually coils and, for the C-L-C filter, I remember to have seen something like that this morning, about the cap below the desoldered cap.

I'll let you know, bye.

[ResistEnt]

Re: Are those two caps bad?

P.S.: yeah, I read and downloaded the desoldering FAQ by Willawake
Now it's all about practicing!

[ResistEnt]

Re: Are those two caps bad?

I can definitely confirm.
I reposted your pic with the voltage rail of each coil.

So you're suggesting me to desolder all output filter caps and replace them with the values you said above, right?

Last question before I do that:
as you said before, it looks like there's no cap greater than 1000uF.
If I replace them with 2200, 3300, ..., but the PSU doesn't accept the new values, what might happen?
Caps may explode, or the entire PSU, or the entire motherboard, or simply nothing powers up, or what else?

[goodpsusearch]

Re: Are those two caps bad?

In worst case scenario the output voltages wont be stable and fluctuate and/or you will get a high pitched sound because of the lowered switching frequency.

I believe it will be fine though. Just don't go too low with esr, lower than 0.03 ohms.

[ben7]

Re: Are those two caps bad?

Yes, replacing with higher capacitance caps (about 2 to 3 times higher in this case) is very unlikely to result in an explosion. Worst case, as goodpsusearch said, it will cause the power supply to not regulate the voltage well.

[ResistEnt]

Re: Are those two caps bad?

Hi.

Here is the updated list.

1. -5/-12V output filter caps, they were 2x 470uF 16V 8x15mm, ripple 800mA, ESR 85 or 93mOhm, replacement: ZLJ 470uF 16V 8x11.5mm, ripple 1200mA, ESR 75mOhm;
2. +5VSB critical cap, it was 22uF 50V 5x11mm, ripple 75mA, ESR unreported, replacement: YXH 22uF 50V 5x11mm, ripple 180mA, ESR 700mOhm;
3. +5VSB output filter caps, they were 2x 1000uF 10V 8x20mm, ripple 1100mA, ESR 50 or 65mOhm, replacement: ZLJ 1000uF 10V 8x20mm, ripple 1960mA, ESR 41mOhm;
4. transistor's polarizing caps, they were 2x 2.2uF 50V 5x11mm, ripple 20mA, ESR unreported, replacement: YXF 2.2uF 50V 5x11mm, ripple 43mA, ESR 2500mOhm;
5. +3.3/+5V output filter caps, they were 2x 1000uF 10V 10x16mm, ripple 1100mA, ESR 50 or 85mOhm, replacement: YXJ 2200uF 10V 10x20mm, ripple 1400mA, ESR 46mOhm.

Is it ok till now? Mostly with ESR values, I mean.

Now the crucial things.

About the graphics card, I don't know if it's worth to change capacitance value. My guess is that the exploded cap is part of the circuit that takes power from the +12V rail and gives it to the board (hence, the message I was receiving, "the board is not receiving sufficient power").
I have no idea about what the other cap does, but since it's a Sacon I'll replace it anyway.
Assuming not to change capacitance value, we have thus:

• 2x 1000uF 16V 10x12.5mm, ripple 1800mA, ESR 22mOhm, replacement: ZLH 1000uF 25V 10x23mm, ripple 2250mA, ESR 18mOhm.

Finally, +12V output filter cap.
It was a 1000uF 16V 10x20mm, ripple 1250mA, ESR 39mOhm.
Its replacement would have been ZLJ 1000uF 16V 10x16mm, ripple 2000mA, ESR 38mOhm, but you suggested me to look for 2200 or 3300uF.
One that seems good is YXJ 3300uF 16V 12.5x25mm, ripple 2230mA, ESR 32mOhm, or there's a very similar one of 2200uF 25V.
Or another one, 2200uF 16V with a closer ESR (41mOhm).
But the problem is, all of them have a different diameter, 12.5mm instead of 10mm.
I checked the PCB and there're a yellow wire and a short circuit just at the sides of the place where the cap has to fit in, in other words it's impossible to put a 12.5mm-diameter cap there unless it's raised from the PCB plane (I mean, unless its bottom is not in direct contact with the board). And if it's not in contact, its length becomes a problem because it's already tall (25mm), plus I'd have to add some millimeters of the legs (and it's a real mess where it has to fit in).
That said, the largest cap (>1000uF) with a diameter of 10mm is not 3300uF but it falls down to:

• 1800uF (ZLJ, ESR too low, 24mOhm);
• 1500uF (ZLJ, ESR 28mOhm, or YXH, ESR 31mOhm);
• 1200uF (YXH, ESR 42mOhm).

What would you do?

[ResistEnt]

Re: Are those two caps bad?

Ok guys, I've had some problems to desolder the caps, I bet there's something wrong I make while doing it.

I use a 40W soldering iron with a quite thin tip, and the first thing I noticed is that when I touch the solder on the PCB, it does not melt, no matter how long I stay with the tip on it.

So I put a small amount of solder directly on the tip, and this way I manage to melt the solder but only for certain points.

For other ones, I have to put fresh solder also on the joint itself, to make the old solder melt, but most of the times I do it, the solder prefers the iron tip to the components' legs or old solder! Uff! I pay attention to touch the leg, the copper (or old solder) and the fresh solder with the iron tip, but no luck...
What's wrong with this?

Second, to clean the tip while working, I use a wet cloth. Is this ok or a wet sponge is necessary?

And last, when I finished to use the iron tip, I let it cool down and put it away. Is this ok or should I wet it with solder before cooling it down?


Sorry for asking these basic questions, but anyone who's taught me "how" to solder simply forgot to mention these apparently useless things.

Thanks, bb.