Power Supply cap suggestions?

[slap_happy]

I've read quite a few threads here, but I'm still confused about cap choice.

The device is a cooler master ultimate ucp 700w power supply. It recently started having cold start troubles after 3 years of mostly continuous use.

Inspection revealed some ltec 2200uf 10v that are crowning, 10mmx25mm.

I looked at the lists here, and some of the threads, and a lot of the recommendations seem to be for 5000hr caps. I'm concerned about continuous use with this life length of cap.

So, what is a good power supply cap choice with longer life specs.

I'm planning to replace 6 2200uf 10v and 4 2200uf 16v. All 10mmx25mm (space is tight). I'm guessing low-esr isn't needed for PSU? I'm confused about all the various series. I really like this power supply or else I wouldn't bother. Thanks in advance for any help.

[PCBONEZ]

Re: Power Supply cap suggestions?

The 5000hr rating is Endurance Lifetime and it tells you absolutely nothing about the life expectancy of a cap.

That number is how long it can handle -MAX- voltage, ripple current & temp [at the same time] WITHOUT going out of spec.

Doesn't have anything to do with the cap's longevity in real world situations - AT ALL.

-
One year is 8760 hrs.
5000 Hrs = 208.3 days.
.

It's not uncommon for caps with a 1000hr Endurance Lifetime to last 15-20 years.
That's because in the real world all the parameters aren't maxed out as they are in an Endurance Test.
.

[PCBONEZ]

Re: Power Supply cap suggestions?

If you want caps that will last:
Use only 105C rated caps.
Use caps with the same or less ESR.
Use caps rated for the same or more Ripple Current. <- BIGGIE

If you had a cap with an Endurance Lifetime rating of 50,000 hours and you stuck it in a circuit that exceeds it's Ripple Current rating it might not last one hour.
The test they do to find the Endurance Lifetime assumes you don't exceed the Ripple Current rating.
- Thus the Ripple Current rating is what you should be concerned with first.

Step one: Look up the ESR and Ripple Current specs of the old caps.
Without that you are just guessing.
.

[PCBONEZ]

Re: Power Supply cap suggestions?

I'm guessing low-esr isn't needed for PSU?

Wrong Guess, minus 5.
-
The larger caps [not counting the 1 or 2 huge inputs cap[s]] are Ripple Filters.
Low ESR is critical there.
.
Can't tell you much more without knowing the series and diameter of the original LTECs.
.

[Toasty]

Re: Power Supply cap suggestions?

>>I'm guessing low-esr isn't needed for PSU?<<

Not just needed, -required- for SMPS use.
Going by the size you gave, those are some of the output caps.

LTEC site: http://www.ltec.com.tw/

Their page is not working as it should. I'm finding the links broken.

Identify your caps by series/capacitance/voltage.
Someone here will have the datasheets to provide the information for you.

Welcome!

[mariushm]

Re: Power Supply cap suggestions?

The number of hours listed for a capacitor is how many hours the capacitor can stay at that temperature within its life and still be within the specifications ( ### uF, +/- 20%, with that ripple and impedance). Something like that... trying to simplify it as much as possible.

The scale is logarithmic, meaning if you keep the capacitors during their life at less temperature, their life goes up a lot. For example it's 5000h at 105c, 8k at 95c, 14k at 85c, 30k at 75c and so on (bogus values, just to get an idea)...

As there's almost always a fan blowing air over the surface of the capacitors, they'll stay on average at about 50-60c. You don't have to worry a lot about that temperature rating.

The ripple and low ESR are very important, try to stick as close as possible to the original specs or exceed them (higher ripple, lower esr).

There's a special case where you replace regular electrolytic capacitors with polymer ones but if don't have experience it's better to stick to the regular caps.

later edit : you can find here a bunch of datasheets ... it's a temporary location, thinking of making something nicer: ftp://helpedia.com/pub/temp/datasheets/capacitors/

[slap_happy]

Re: Power Supply cap suggestions?

Thanks to all the above, I've learned a lot already - which is good.

The 6 ltec are 2200uf 10v LYZ 105c 10x25

I'm thinking of replacing these with rubycon mbz

The 4 ost are 2200uf 16v RSP 105c 10x25

Was also thinking of going with rubycon mbz, but apparently the 16v version doesn't grow on trees. Maybe I can somehow squeeze in ncc kze 12.5mm.

[PCBONEZ]

Re: Power Supply cap suggestions?

The number of hours listed for a capacitor is how many hours the capacitor can stay at that temperature within its life and still be within the specifications ( ### uF, +/- 20%, with that ripple and impedance). Something like that... trying to simplify it as much as possible.

No.
The Endurance Lifetime has nothing at all to do with expected lifespan.
Because many of the factors are the same there is a formula to get from one to the other but [according to caps manufacturers] it is only accurate to +/- 40%.
If you do the math and come up with 5 years it really means anywhere from 3 years to 7 years.
- Not very useful.
-THEN-.
If you do the math and come up with 7 years in means anywhere from 4.2 years to 9.8 years.
--
Every single '7 years' cap could fail at 4.2 years and every single '5 years' cap could fail at 7 years and the math would be dead-on correct for both.
[IOW: The numbers aren't useful and the math is a waste of time.]
This is why -NO- caps manufacturer advertises an expected useful lifetime.
AND it's why looking at the Endurance Lifetime is a waste of your time unless you are a design engineer building for extremely harsh environments.

The scale is logarithmic, meaning if you keep the capacitors during their life at less temperature, their life goes up a lot. For example it's 5000h at 105c, 8k at 95c, 14k at 85c, 30k at 75c and so on (bogus values, just to get an idea)...

Specifically a 10C rise in temp reduces life by 50%.
[-Changes- in Endurance Lifetime are *proportional* to -changes- in Useful Lifetime so if you double one then the other will be double also.]

As there's almost always a fan blowing air over the surface of the capacitors, they'll stay on average at about 50-60c. You don't have to worry a lot about that temperature rating.

Wrong.
All else being equal a 2000hr 85C cap will last 25% as long as a 2000hr 105C cap if they are both at the same actual temp.
Bad 85C caps are fairly common after motherboards hit about 4-5 years of use. Even the little 4 and 5 mm 85C caps that don't 'work hard' begin to fail about then. [You can -sort of- take from that, that 105C caps under similar operating conditions are good for around 16-20 years. REALLY *sort of* that is. It's a big leap..]

The ripple and low ESR are very important, try to stick as close as possible to the original specs or exceed them (higher ripple, lower esr).

YES YES YES

[In a way...]
The Ripple Current rating mostly affects longevity.
The ESR rating mostly affects effectiveness/quality of the cap doing it's job.

~~~~~

You can bypass LTEC's bad links in their web pages and go straight to the file by entering this in a google search:

site:ltec.com.tw XXX.pdf

Just replace the "XXX" with the series name like so:


.

[PCBONEZ]

Re: Power Supply cap suggestions?

Didn't see the LYZ
Here it is:

[mariushm]

Re: Power Supply cap suggestions?

No.
The Endurance Lifetime has nothing at all to do with expected lifespan.
Because many of the factors are the same there is a formula to get from one to the other but [according to caps manufacturers] it is only accurate to +/- 40%.
If you do the math and come up with 5 years it really means anywhere from 3 years to 7 years.
- Not very useful.
-THEN-.
If you do the math and come up with 7 years in means anywhere from 4.2 years to 9.8 years.
--
Every single '7 years' cap could fail at 4.2 years and every single '5 years' cap could fail at 7 years and the math would be dead-on correct for both.
[IOW: The numbers aren't useful and the math is a waste of time.]
This is why -NO- caps manufacturer advertises an expected useful lifetime.
AND it's why looking at the Endurance Lifetime is a waste of your time unless you are a design engineer building for extremely harsh environments.


Specifically a 10C rise in temp reduces life by 50%.
[-Changes- in Endurance Lifetime are *proportional* to -changes- in Useful Lifetime so if you double one then the other will be double also.]

.

I was trying to make it very, very simple. I probably explained it in a too simple way.

Here's a document explaining all the terms and parameters of capacitors:



I'm quoting from page 9-10:

ENDURANCE TEST
In “IEC 60384-4” or “EN130300” the criteria for the acceptable drift of electrical parameters after the endurance test at UR and upper category temperature are defined.
Test duration and conditions per series are stated in the relevant detail specification.
The endurance test does not provide information about the useful life of a capacitor, as no failure percentage is defined for this test.

USEFUL LIFE

Useful life (other names: load life, life time or typical life time) is that period of time, during which a given failure percentage may occur, under well defined conditions and requirements.
Useful life data are usually calculated with a confidence level of 60 %.

[...]

For non-solid aluminum capacitors the influence of temperature on useful life is approximated by the so-called ‘10 K-rule’. The ‘10 K-rule’ states that double the life time can be expected per 10 K temperature decrease; this principle is derived from the well known law of Arrhenius about acceleration of reaction processes.
The exact temperature dependence of useful life for a particular range is given in the corresponding detail specification in the ‘life-time nomogram’ (‘Multiplier of useful life’ graph in the detail specifications). Detailed performance requirements, on which the definition ‘useful life’ is based,
are also stated in the relevant detail specifications.
[...]

Statistically, capacitors rated for 105c will have a longer life, but that's not necessarily because they resist better to heat but more often because of the better quality of the manufacturing, better electrolytic solutions and so on. In my opinion a 85c capacitor in a well cooled circuit can very well live just as much as a 105c capacitor in less ideal conditions. Your view may be biased due to the fact that most 85c capacitors you see are from low quality brands (cheap power supplies etc)

See page 10, it shows some really nice graphs and some methods to determine the useful life based on ripple, temperature and rated hours @ temp. There's even a problem + solution there.

[PCBONEZ]

Re: Power Supply cap suggestions?

And quoted right from you quote:
The endurance test does not provide information about the useful life of a capacitor, as no failure percentage is defined for this test.

The part you made bold is exactly what I said.
A 10C change doubles or halves the life. [Depending on which way temp changes.]

~~

No my opinion about 85C caps isn't biased it's based having a concept of how it works and basic math skills.

105C cap rated for 2000Hrs
105C - 2000hr
95C - 4000hr
85C - 8000hr

85C cap rated for 2000Hrs
85C - 2000hr

At 85C 2000Hr is 25% of 8000Hr [which is what I said.]
And the 85C cap will be at 25% of the 105C at -any- temp.
That's how it works.

~~

Most of the blown 85C I see are Nichicon.
The crap brand 4-5mm 105C don't have the problem and last longer.
I've posted photos in other threads of a few boards with the 85C cap problem.
Those just happened to be Nichicons.
.
.

[PCBONEZ]

Re: Power Supply cap suggestions?

Taking it one step further [IAW "All else being equal", as I said] and assume both caps have a 'Useful Life' of 5 years at their max temp.

105C cap rated for 2000Hrs
105C - 2000hr - 5yrs
95C - 4000hr - 10yrs
85C - 8000hr - 20yrs
75C - 16000Hr - 40yrs
65C - 32000Hr - 80yrs

85C cap rated for 2000Hrs
85C - 2000hr - 5yrs
75C - 4000hr - 10yrs
65C - 8000Hr - 20yrs

~~~
See what I was saying now?
.
Since they don't cost much [if any] different buying 85C for replacement caps is NOT the way to go.
.

[PCBONEZ]

Re: Power Supply cap suggestions?

I can't find the data for the OST RSP Series.
If those 4x 16v are all in the VRM input then 4x 330uF or 470uF 16v polys would do fine.
The uF isn't the main concern there, ESR is.
.

[c_hegge]

Re: Power Supply cap suggestions?

Maybe he meant RLP series

[PCBONEZ]

Re: Power Supply cap suggestions?

Maybe R.I.P. [hehehehe]
.
OST's Poly series is called PUS.
.

[Toasty]

Re: Power Supply cap suggestions?

Here's the RLP sheet.

Looks like they're dropping the old designations for new ones, but don't have the sheet links working yet.

Without knowing that the old sheets were still there, that info would be lost.

Good suggestion there PCBONEZ!
Go you one better and drop the XXX.

site:ltec.com.tw .pdf

Gets them all in a list

Works with OST too:

site:ost.com.tw .pdf

.

[PCBONEZ]

Re: Power Supply cap suggestions?

Yes but the goal was to find RSP.
[Lacking an update/correction from the requester.]

>> site:ltec.com.tw .pdf
I couldn't get that to work last night.
Maybe I misspelled 'pdf' ?
.

[mockingbird]

Re: Power Supply cap suggestions?

Is it possible to test a cap's voltage on a PSU with it in standby, or does the motherboard have to be drawing power from it?

[Toasty]

Re: Power Supply cap suggestions?

There's no RSP. I pulled their site down a year ago and that sheet never showed. Nor does a Google for it. Gotta be a typo.

@ MB: Standby will not have the secondary running. Supply has to be "ON" to see that. Does not have to be hooked to anything to see secondary voltages if you jumper it "ON".


BTW: Here's a link to OST directory listing: http://www.ost.com.tw/PDF/EC/

PS: BTW: PUWS is their -other- polymer cap series. Both attached FYC.

.

[PCBONEZ]

Re: Power Supply cap suggestions?

I know, that's why I said I couldn't find RSP.
Might be old or an custom/OEM only series.

The 'S' key is a long way from the 'L' key though.

~~
Pardon my VRM comment.
Was late and I forgot this was a PSU.
.