keropi wrote:

I also used silicone to glue together the caps on the ATX psu like the manufacturer had , just in case the cables start pushing them.

That actually is not usually necessary.
They only do that at the factory to hold everything in place so they can solder the whole board at once.
On rare occasion it's used to keep something from grounding/shorting against other components but that's not usually the case.
Most of the time you are better off without it because it restricts airflow (cooling) - as someone else mentioned.
.

TELVM wrote:

In fact as the higher the capacitance and the bigger the cap can => the lower the ESR

I've read a bunch of your posts and you clearly know your stuff but this statement simply isn't true.
.
There is no correlation between uF and ESR.
Within a series of cap the ESR varies with the can size.
The uF is irrelevant to ESR unless you change to an entirely different series of cap - or change the can size within that series.
.
You can see what I'm saying in about any datasheet from a reputable cap manufacturer by looking at all the caps with the same can size.
The ESR and can size are pretty much locked together, while the uF's associated with that can size (and ESR) will vary greatly.
It can vary if you go way outside the voltages used on motherboard (like 50v caps)
.
For example:
Panasonic datasheets generally have a table before the main datasheet that lists the can sizes, ESR and ripple for that series and that table excludes uF numbers entirely.
I've attached the Ruby MCZ datasheet (because it's not huge) and you can see -
- that regardless of uF or voltage all the 8x16mm have the same ESR/Ripple ratings, all the 10x20mm have the same ESR/Ripple - and so-on for every can size.
.

You're right I should just have said that within the same series, the larger the lytic cap can size => the lower the ESR (without mentioning capacitance at all).

I will need some help too, but not with an ancient psu.
A good friend of me has a first generation I7 build with an Asus P6T deluxe V2 mainboard that fried it's third psu in 4 years. 😵 😢
One thing that is for sure not guilty is it's graphics card since he had three different ones in the system.. each time another when a psu blew.

They are all decent psu's from what I can tell so I am curious if recapping brings them alive again.
The last blown PSU in question is a Chieftec CTF-850g-df, I opened it yesterday and atleast 4 caps are blown open, some other bulging.

Building in my test psu brought the system back alive so the rest seems save. 😊

I will try to get decent pictures.. yesterday evening I couldn't shoot decent ones and I had a hard time disassembling the unit since it's pretty cramped full. I only have a so-so soldering skill so I will need some advice. The blown capacitors are if I observed them well from Teapo.

Now I have the questions: A) What causes so many failures on the psu's? I am running the same psu in much heavier loaded system and it runs alright since I bought it.
B) Where and what caps do I need to order, and how can I perhaps improve the PSU.

As soon as I am home again I will add decent pictures.

Nvm1 wrote:

I will need some help too, but not with an ancient psu. A good friend of me has a first generation I7 build with an Asus P6T del […]
Show full quote

I will need some help too, but not with an ancient psu.
A good friend of me has a first generation I7 build with an Asus P6T deluxe V2 mainboard that fried it's third psu in 4 years. 😵 😢
One thing that is for sure not guilty is it's graphics card since he had three different ones in the system.. each time another when a psu blew.

They are all decent psu's from what I can tell so I am curious if recapping brings them alive again.
The last blown PSU in question is a Chieftec CTF-850g-df, I opened it yesterday and atleast 4 caps are blown open, some other bulging.

Building in my test psu brought the system back alive so the rest seems save. 😊

I will try to get decent pictures.. yesterday evening I couldn't shoot decent ones and I had a hard time disassembling the unit since it's pretty cramped full. I only have a so-so soldering skill so I will need some advice. The blown capacitors are if I observed them well from Teapo.

Now I have the questions: A) What causes so many failures on the psu's? I am running the same psu in much heavier loaded system and it runs alright since I bought it.
B) Where and what caps do I need to order, and how can I perhaps improve the PSU.

As soon as I am home again I will add decent pictures.

I wrote a much longer more detailed answer then my Firefox crashed and I lost it all. - Here's a short version so you know I'm on it.

Chieftec is absolutely NOT a quality PSU and presumably the others you had weren't either.
(Not meant to be insulting. I knew no better when I started out.)

I suspect the problem is that you are using PSUs with multiple 12v rails and the video card is overloading ONE of the rails.
From an engineering standpoint multiple 12v rails is a bad idea.
Multiple 12v rails was forced upon us by ignorant and stupid law makers in Europe.
(I can explain that in detail if anyone is up for a long rant.)

I need more info to offer suggestions as to a good PSU to use.
Namely, What CPU (or it's TDP), what video card (or it's TDP), how many and what slots are in use and how many drives are used.
.

PCBONEZ wrote:

I suspect the problem is that you are using PSUs with multiple 12v rails and the video card is overloading ONE of the rails. Fro […]
Show full quote

I suspect the problem is that you are using PSUs with multiple 12v rails and the video card is overloading ONE of the rails.
From an engineering standpoint multiple 12v rails is a bad idea.
Multiple 12v rails was forced upon us by ignorant and stupid law makers in Europe.
(I can explain that in detail if anyone is up for a long rant.)

I'm actually curious. Is is an amperage limit? Like no circuitry in an electronic device should draw more than X amps? Because I can't imagine another reason, even though limiting amps on some 12V line is pretty stupid.

alexanrs wrote:

PCBONEZ wrote:

I suspect the problem is that you are using PSUs with multiple 12v rails and the video card is overloading ONE of the rails. Fro […]
Show full quote

I suspect the problem is that you are using PSUs with multiple 12v rails and the video card is overloading ONE of the rails.
From an engineering standpoint multiple 12v rails is a bad idea.
Multiple 12v rails was forced upon us by ignorant and stupid law makers in Europe.
(I can explain that in detail if anyone is up for a long rant.)

I'm actually curious. Is is an amperage limit? Like no circuitry in an electronic device should draw more than X amps? Because I can't imagine another reason, even though limiting amps on some 12V line is pretty stupid.

Some EU politicians got the bright idea that more than 20 amps available on any kind circuit could kill you - which is insanely stupid. (Back to that momentarily.)
So they passed a law such that no consumer accessible circuit could be more than 20 amps. (I think it was 20 amps. Something close to that anyway.)
Thus to be 100% legal in the EU a PSU can't have rails more than 20 amps. (I think many manufacturers are cheating now as no one in authority really looks.)
And since PSU manufacturers don't want to build special models just for the EU, EVERYONE else -world wide- was "blessed' with multiple rail PSUs.

First it takes WAY less than 20 amps to kill you - so the 20 amp limit was an ignorant idea to start with. Also the current through the body is determined by the voltage and the bodies resistance, NOT the current capacity of the voltage source.
Second, 12v DC (regardless of the circuits possible amperage) will never kill anyone because the resistance of the human body is such that 12v DC can not possibly create enough current through the body to the heart to put anyone in mortal danger.

Depending on where you look the "Kill Current" for DC is given as 300 to 500 milliamps ( 0.3 to 0.5 amps).
(AC is worse because it cycles. DC doesn't cycle. The usual number for AC is 100 milliamps ( 0.1 amp) for 1 second.)
- The 300 ohms below comes from USN training manuals for Electricians and Electronic Techs.
A human body that is soaking wet with sweat or seawater is 300 ohms or more from arm to arm or arm to foot - most likely routes through the heart when working on equipment.
Sweat and seawater are used by the Navy as they have salt content and are thus worse than being wet with plain water. ( IOW: A worst case scenario.)
Using Ohms Law and 300 Ohms, 12v is only going to result in 40 milliamps ( 0.04 amps) through the body (heart).
40 milliamps would hurt like a bugger but it's 7.5 times (750%) less DC current than what would kill you.

Also note that most people aren't going to be sopping wet or standing in seawater while they work on their PC.
Using 300 ohms is overkill in more realistic situations.

At least when I was 'in", no special precautions were required for working on -live- -DC- under 30 volts.
30v DC and up required rubber gloves, rubber mats, barriers, signs and so forth. Below 30v DC those were optional.
At 30v DC you would only have 100 milliamps (on 300 ohms) so that requirement still has a 3x safety margin.

There are places that say 30v may kill you, however that is ONLY for AC voltage. They usually leave that part out.

...
Being deliberately stupid is an exception.
I know of one case where a moron managed to kill himself with 9v DC.
He used sharpened meter probes and stuck them in his veins supposedly to measure the resistance. (Most of a human's resistance is in the skin.)
Blood has salt content and is literally a conductor that runs straight to the heart, thus the resistance was way lower than 300 Ohms.
No Law is going to protect people THAT stupid from killing themselves.
[ I did not know anyone involved in that. I read it in an incident report.]

...
Anyway - My point is that multiple 12v rails would not even exist were it not for politicians with no technical knowledge what-so-ever passing laws regarding technical matters.
THEY are STUPID so WE have to deal with more PITA technical problems (like balancing rail loading) and pay out more for PSUs that are more costly to manufacture.
- Just pisses me off. (Can you tell?)
.

Haven't been home yet so no pictures as for now.
The split rails is limited to 25A on this PSU. As far as the real producer of the PSU I came out it should be CWT?

The split rails is the first time you have to use it tricky to make sure you don't plug everything on one of them. 😒

Hope I get the time to visit the dead psu this evening to take the pictures.

Yup, so the law is very stupid.

Nvm1 wrote:

I... first generation I7 build with an Asus P6T deluxe V2 mainboard that fried it's third psu in 4 years. 😵 😢
One thing that is for sure not guilty is it's graphics card since he had three different ones in the system.. each time another when a psu blew.

... A) What causes so many failures on the psu's? I am running the same psu in much heavier loaded system and it runs alright since I bought it.

Overheating perhaps? How's the cooling in that particular system?

Nvm1 wrote:

... No Law is going to protect people THAT stupid from killing themselves ...

Agreed.

TELVM wrote:

Nvm1 wrote:

I... first generation I7 build with an Asus P6T deluxe V2 mainboard that fried it's third psu in 4 years. 😵 😢
One thing that is for sure not guilty is it's graphics card since he had three different ones in the system.. each time another when a psu blew.

... A) What causes so many failures on the psu's? I am running the same psu in much heavier loaded system and it runs alright since I bought it.

Overheating perhaps? How's the cooling in that particular system?

The cooling is very well in his tower. Big 13cm front fan sucking in air, big 13cm rear fan blowing out and two 92mm fans, one for the hdd cage and one next to his soundcard and under his 970gtx. And no obstructions around the case 😒

TELVM wrote:

Nvm1 wrote:

... No Law is going to protect people THAT stupid from killing themselves ...

Agreed.

Finally the first pictures:

Will dismantly the unit further this weekend to get the board better within reach.

Output filter caps are bloated and at least two are leaking.
Needs a recap.
.
Personally I would upgrade to a Corsair TX or a Corsair TX V2.
Good quality and they have single 12v rails.
.

Definitely worth recapping. It won't be easy though.

wave wrote:

Definitely worth recapping. It won't be easy though.

It's kind of a first attempt to salvage this psu to be used as my test psu. I never recapped a motherboard of psu so this should give me a good first go. 😈
I will take better pictures as soon as I have it dismantled further. The only real challange seems to be that the whole where all the cable go through is so tight that I can't get the plastic ring loose that keeps everything in place.

Any input regarding what caps to use/order will be welcome when it's so far. And input to explain to me why it is bad or good at points are welcome. I Always found this topic very interesting and aside from it being a dual rail design I like to learn what to look out for and how it can/could have been better.

And here I am,with three (of my four 🤣 ) PSUs I've got to share with you.

1.Premier LC-B450E

Ah yes. The LC-B450E. Of course,we all over here know who's behind this PSU. Hint:It's L&C,which in turn is Deer.

2.Delux ATX-450W P4

Again,an usual (de)luxury around my town. This one is OEM'd by Sun Pro,which is behind a few older Coolmax units,some Spire units and Gembird PSUs. (Gembird PSUs are the best example of overbuild. They rate them pretty honestly)This one was a pretty good design (apart from the transformer,that is) let down by thin heatsinks. For god's sake Sun Pro,I thought you could do better for the heatsinks,as I can handle replacing the rectifiers and caps. This came with a 1.7GHz Willamette.

3.Codegen 300X1 300W

Hey,this one is actually decent. Pretty old PCB design (reminds me of AT days) but it's crowded with approved parts. Yes,all parts have approved markings.

And for TELVM (and any other expert as well) - you can take a look and say about the quality of each. 😀

They are all 250 watters at best..

Nvm1 wrote:

... Any input regarding what caps to use/order will be welcome ...

From a distance they look like Teapo SC. Quality jap series of similar ESR are Chemi-Con KY, Nichicon PW, Panasonic FC, Rubycon YXG.

If I were to replace that PSU with something new of same wattage, nowadays (2015) I'd get an EVGA Supernova G2 850.

PcBytes wrote:

1.Premier LC-B450E

Oh Deer! Unless there are PI coils on the secondary (can't see them in the pic) don't even think on wasting effort on it. Efficiency is atrocious (70% or less), I wouldn't trust it above ~250W tops, and I wouldn't power anything '+12V heavy' (P4 or later) with it.

That said, with PI coils and decent caps voltage regulation and ripple would be acceptable for powering a PIII or earlier.

Delux ATX-450W P4

Only the PI coils on the secondary absolve it from a total Gutless Wonder award. Almost no input filtering, the infamous four diode treatment (with the diodes lying flat, not raised like in the Deer, which worsens cooling), roasted PCB, etc. Scrounge it for parts then junk it.

3.Codegen 300X1 300W

Have you noticed those two discrete diodes on a bracket (epic ghettoness) for +12V rectification? 😵

PcBytes wrote:

And here I am,with three (of my four :lol: ) PSUs I've got to share with you. […]
Show full quote

And here I am,with three (of my four 🤣 ) PSUs I've got to share with you.

1.Premier LC-B450E

Ah yes. The LC-B450E. Of course,we all over here know who's behind this PSU. Hint:It's L&C,which in turn is Deer.

2.Delux ATX-450W P4

Again,an usual (de)luxury around my town. This one is OEM'd by Sun Pro,which is behind a few older Coolmax units,some Spire units and Gembird PSUs. (Gembird PSUs are the best example of overbuild. They rate them pretty honestly)This one was a pretty good design (apart from the transformer,that is) let down by thin heatsinks. For god's sake Sun Pro,I thought you could do better for the heatsinks,as I can handle replacing the rectifiers and caps. This came with a 1.7GHz Willamette.

3.Codegen 300X1 300W

Hey,this one is actually decent. Pretty old PCB design (reminds me of AT days) but it's crowded with approved parts. Yes,all parts have approved markings.

And for TELVM (and any other expert as well) - you can take a look and say about the quality of each. 😀

For about the last 10 years when my horse trading adventures 'blesses' me with those kinds of PSUs I replace them with high efficiency models and just get rid of them.
I'm not saying there is anything wrong with them, just that I don't -work on- those so I can't tell you much about them.
You can find a whole passel of people that know those things inside-out though.
They frequent http://hardwareinsights.com and/or http://badcaps.net. (Many members are on both sites. I've seen some of them here too.)
Both sites have people that do actual reviews of old PSUs like those. They LOVE those things.
Me, I love working on motherboards and I only rework a PSU when I HAVE to.
.

Nvm1 wrote:

It's kind of a first attempt to salvage this psu to be used as my test psu. I never recapped a motherboard of psu so this should […]
Show full quote

wave wrote:

Definitely worth recapping. It won't be easy though.

It's kind of a first attempt to salvage this psu to be used as my test psu. I never recapped a motherboard of psu so this should give me a good first go. 😈
I will take better pictures as soon as I have it dismantled further. The only real challange seems to be that the whole where all the cable go through is so tight that I can't get the plastic ring loose that keeps everything in place.

Any input regarding what caps to use/order will be welcome when it's so far. And input to explain to me why it is bad or good at points are welcome. I Always found this topic very interesting and aside from it being a dual rail design I like to learn what to look out for and how it can/could have been better.

The first thing you must learn is that the caps needed are not general purpose caps.
They need low ESR caps and there are 'grades' of low ESR. (Grades is my own word. Not some Official Standard.)
- The reason I brought that up is that sourcing the correct kind of caps in the correct grade is not always easy and may take you a while.
- That means this PSU may be out of service quite a while waiting on parts deliveries.
That is part of why I suggested replacing it. - To get the system back on-line sooner.

~ Will note here because it always comes up eventually~
Terms like: "Ultra Low ESR", "Very Low ESR", "Super Low ESR", "Extra Low ESR" - mean ABSOLUTELY NOTHING.
There is no Industry Standard or convention that defines these terms so they can mean whatever the manufacturer wants them to mean.
As a result company A's "Ultra Low ESR" may be exactly equivalent to company B's "Very Low ESR".
~

So you are going to try this:

Step 1 in any kind of recap job is to make a Cap-Map. - NEVER EVER SKIP THE CAP-MAP! - Sh*t Happens!
If there is an interruption part way through the job such that you have to come back to it later you won't remember which cap goes were. (Trust me on this.)
The Cap-Map is just a drawing of the board that shows:
- Where the caps go.
- WHICH WAY they go (The polarity.)
(The polarity marks on PCBs are not always correct. Look which way the cap is installed before you remove it and mark it on your map.)
- The original cap's uF, Volts, Manufacturer, Series, diameter, length. (I only include length if too long would be a problem.)
If any have a weird lead spacing you might want to note that too. This is pretty rare though.
That PSU appears to have daughter boards. You should do a map for each board you will work on.
Doesn't work with PSUs but with motherboards you can usually find a photo or drawing online to make the map. (Print and draw right on it.)

Step 2 is to make your ordering worksheet. This is simply a list of all the caps you need to replace followed by possible replacements.
- The following is of course just one way of doing this.
From your Cap-Map info list the original cap's. Include uF, Volts, Manufacturer, Series, diameter, (length), how many you need.
Then add the original cap's ESR and Ripple - You will need to look ESR/Ripple up in the original cap's datasheet.
(Not all cap datasheets are published and not all actually produced caps are found in the datasheets that are published. There are work-a rounds if you run into that.)
- Doesn't hurt to have other columns for where you are buying (the dealer/seller) and the price.
(On the paper) Each cap you need to replace should have several blank lines under it to enter possible replacements you come across while you're shopping.
I use a pencil for that part 'cause it gets changed often as I come across better replacements or better deals.
[ I've attached a screen-shot of one I cleaned up and saved for future reference. The worksheets aren't this 'pretty' while I'm using them. Normally there are several possible replacements listed under each original cap but before I archived this I removed all the ones I decided against and only left the final choice. I also don't usually include serial numbers but I had 6 of these to recap and their cap situations were all different. ]

Do you know how to solder?
This is a little different than general electronics work as the boards are extra thick. You really need a 50-60 or even 70 watt iron. (Especially for motherboard work.)
If you're experienced you can 'make it work' with a 40 watt but it's still easier with an iron that has more 'grunt'.
The thicker boards suck the heat out of the iron quickly. If the iron doesn't have enough grunt to keep things hot enough to flow the solder completely you'll have a mess.
.