capacitor replacement question

[Chris]

I have figured out why my dead system won't boot/post. It turns out,
after a quick look, there are 3 bad caps (they are bulging at the top,
my thanks to the poster who suggested to check the caps). Anyhow, the
motherboard is a ECS K7S5A and the 3 caps that are bad are rated at
6.3volts at 1800 uf. I have an old motherboard with 3 caps that are
the same size and are rated at 6.3 volts but 1200uf. Will these work?
If not, where can I order the new caps from?

Chris
If life seems jolly rotten
There's spmething you've forgotten
and thats to laugh and smile and dance and sing!

 

[Arno Wagner]

I have figured out why my dead system won't boot/post. It turns out,
after a quick look, there are 3 bad caps (they are bulging at the top,
my thanks to the poster who suggested to check the caps). Anyhow, the
motherboard is a ECS K7S5A and the 3 caps that are bad are rated at
6.3volts at 1800 uf. I have an old motherboard with 3 caps that are
the same size and are rated at 6.3 volts but 1200uf. Will these work?
If not, where can I order the new caps from?

Unless the old caps are low-ESR types, they will not work at all.
Even if they are low-ESR, you might have serious troubles, since
they are too small in capacity.

I used Rubycon ZL series caps for replacement with good experiences.
Rubycon ZLH are even better.

You need to replace with the same capacity, if necessary combined
from several capacitors. If you are sure the three are in paralell,
then you can use, e.g. two 2200uf and one 1000uf. If they are
not paralell, try with the closest match you can get, 2200uF
or 1500uF may work.

As to voltage, higher is better.

As to size, if these are 10mm diameter ones, you may not
get replacements that fit. Since bigger diameter caps
have better characteristics, this is not too bad. You can
mount them with a bit of air between cap and board and
bend the leads to fit.

As for a source, that depends on were you are. In Europe,
www.distrelec.com is a possible one.

Arno

 

[Bob Palmer]

I have figured out why my dead system won't boot/post. It turns out,
after a quick look, there are 3 bad caps (they are bulging at the top,
my thanks to the poster who suggested to check the caps). Anyhow, the
motherboard is a ECS K7S5A and the 3 caps that are bad are rated at
6.3volts at 1800 uf. I have an old motherboard with 3 caps that are
the same size and are rated at 6.3 volts but 1200uf. Will these work?
If not, where can I order the new caps from?

Chris
If life seems jolly rotten
There's spmething you've forgotten
and thats to laugh and smile and dance and sing!

my guess - despite being 33% smaller the ones you have will handle the job.
Consider that what was there has been deteriorating for some time before the
machine died and you will have been living with well less than the rated
value. These are only filter caps at any rate and since the application is
the same they can't hurt. Worst case - a little more instability than an
up to snuff machine.

 

[Franc Zabkar]

I have figured out why my dead system won't boot/post. It turns out,
after a quick look, there are 3 bad caps (they are bulging at the top,
my thanks to the poster who suggested to check the caps). Anyhow, the
motherboard is a ECS K7S5A and the 3 caps that are bad are rated at
6.3volts at 1800 uf. I have an old motherboard with 3 caps that are
the same size and are rated at 6.3 volts but 1200uf. Will these work?
If not, where can I order the new caps from?

Chris

Not cheap, but try http://www.badcaps.net/.

- Franc Zabkar

 

[M.I.5¾]

In comp.sys.ibm.pc.hardware.misc Chris


Unless the old caps are low-ESR types, they will not work at all.
Even if they are low-ESR, you might have serious troubles, since
they are too small in capacity.

I used Rubycon ZL series caps for replacement with good experiences.
Rubycon ZLH are even better.

You need to replace with the same capacity, if necessary combined
from several capacitors. If you are sure the three are in paralell,
then you can use, e.g. two 2200uf and one 1000uf. If they are
not paralell, try with the closest match you can get, 2200uF
or 1500uF may work.

As to voltage, higher is better.

No it isn't. Not in electrolytics anyway. Replace with 6.3 volt, though a
volt or two higher would be OK. Anyhting much higher will reduce the
reliability. Replace them with 105°C caps if you can. These are
significantly more reliable than the 85°C caps that are probably fitted.

 

[M.I.5¾]

I have figured out why my dead system won't boot/post. It turns out,
after a quick look, there are 3 bad caps (they are bulging at the top,
my thanks to the poster who suggested to check the caps). Anyhow, the
motherboard is a ECS K7S5A and the 3 caps that are bad are rated at
6.3volts at 1800 uf. I have an old motherboard with 3 caps that are
the same size and are rated at 6.3 volts but 1200uf. Will these work?
If not, where can I order the new caps from?

Chris

Not cheap, but try http://www.badcaps.net/.

 

[Arno Wagner]

No it isn't. Not in electrolytics anyway.

A higher voltage will result in lower ESR and in lower temperature,
hence longer lifetime. Look into the datasheet of the ZL and ZLH
capacitors....

Replace with 6.3 volt, though a
volt or two higher would be OK. Anyhting much higher will reduce the
reliability.

This is complete nonsense.

Replace them with 105°C caps if you can. These are
significantly more reliable than the 85°C caps that are probably fitted.

There are no 85C low-ESR capacitors on the market. 85C capacitors
will a) not do the job b) explode pretty fast.

It strikes me that you do not know what you are talking about.

Arno

 

[Arno Wagner]

On Tue, 21 Nov 2006 22:13:44 GMT, Chris
<[email protected]> put finger to keyboard and
composed:

Not cheap, but try http://www.badcaps.net/.

But he's already got bad capacitors. What he wants are good ones.

You really have no clue, do you?

Arno

 

[M.I.5¾]

A higher voltage will result in lower ESR and in lower temperature,
hence longer lifetime. Look into the datasheet of the ZL and ZLH
capacitors....


This is complete nonsense.

Not so. If the capacitors are too high a voltage, then the insulating layer
does not reform fully when the polarising voltage is applied. This in fact
considerably lowers the reliability of the capacitors. Just check any
decent manufacturer's reliability data for voltage derating (assuming you
could understand it).

There are no 85C low-ESR capacitors on the market. 85C capacitors
will a) not do the job b) explode pretty fast.

Really? Then it is you who doesn't know what you are talking about. The
first catalogue I grabbed off my shelf has 3 pages of them made by Siemens
and EPCOS, and they are specifically sold for use in computer power supply
systems (from very large to extremely small). 4 pages later turns up a page
and a half of the same thing made by Panasonic. So presumably they will do
the job, just at a lower reliability. Many suppliers of even very expensive
systems prefer to use 85 °C capacitors because they are almost half the
price. Personally, I would never use them as they only have about a third
to half of the MTTF of a 105°C device. Looking inside this very computer
reveals that all the capacitors that can be seen without a magnifying glass
are all rated at 85°C.

It strikes me that you do not know what you are talking about.

In that case I would stop travelling anywhere by aeroplane. I am an
electronics engineer in the aerospace industry, and I most certainly do know
what I am talking about. A substantial amount of hardware goes over your
head to prove it.

 

[M.I.5¾]

You really have no clue, do you?

Whoosh!

 

[Arno Wagner]

Not so. If the capacitors are too high a voltage, then the insulating layer
does not reform fully when the polarising voltage is applied. This in fact
considerably lowers the reliability of the capacitors. Just check any
decent manufacturer's reliability data for voltage derating (assuming you
could understand it).

Really? Then it is you who doesn't know what you are talking about. The
first catalogue I grabbed off my shelf has 3 pages of them made by Siemens
and EPCOS, and they are specifically sold for use in computer power supply
systems (from very large to extremely small).

Ok, my knowledge may be outdated. I admit I only looked at my
primary component source. www.distrelec.ch.

Care to give me a model number or series designation? I cannot find
any low-ESR elecrolytes on the EPCOS page. They do have low-ESR
tantalum and ceramics, but they have different tempersture
characteristics.

4 pages later turns up a page
and a half of the same thing made by Panasonic. So presumably they will do
the job, just at a lower reliability. Many suppliers of even very expensive
systems prefer to use 85 °C capacitors because they are almost half the
price. Personally, I would never use them as they only have about a third
to half of the MTTF of a 105°C device. Looking inside this very computer
reveals that all the capacitors that can be seen without a magnifying glass
are all rated at 85°C.

Have you looked at the low-ESR types in the CPU VR? I have not seen
any 85C capacitors or standard-ESR types used there.

Arno

 

[M.I.5¾]

Ok, my knowledge may be outdated. I admit I only looked at my
primary component source. www.distrelec.ch.

Care to give me a model number or series designation? I cannot find
any low-ESR elecrolytes on the EPCOS page. They do have low-ESR
tantalum and ceramics, but they have different temperature
characteristics.

There are plenty to chose from. Someone has borrowed my first catalogue,
but grabing another and allowing it to flop open open shows me the 056/7
series from BC Components (or for higher voltages the 157 series). ELP
Components ALP 20 and ALP 22 series (the difference is purely tolerance).

The confusion may arise because ESR is not a function of the maximum
temperature range of a capacitor. It is a function of the actual capacitor
temperature (roughly a rectangular asymptotic hyperbolic function). Thus
105 °C capacitors have the capability of possessing a lower ESR than their
85 °C counterparts - but only when operated at those higher temperatures.
Officially many manufacturers also claim that ESR is frequency dependant,
but resistance should not be a function of frequency. I believe they really
mean ESI (Equivalant Series Impedance), but this is probably too complex a
concept.

I have worked on a project that required capacitors that required a far
lower ESR and ESI than anything you can ever find in any catalogue. We had
to get them specially made, but I can't tell you how they were made, because
I would then have to kill you.

Have you looked at the low-ESR types in the CPU VR? I have not seen
any 85C capacitors or standard-ESR types used there.

A specific application for which I am not in the slightest bit surprised.
CPU voltage regulator capacitors are usually positioned extremely close to
the CPU itself, often have to operate at quite high temperatures, and are
purpose made for that application. Only the garbage turned out by Intel
won't operate at much more than 40-50 °C without soft errors. But most
serious processors will opearate at 140 °C or more. Some of our PowerPC
boards operate the processor so hot that the board itself has turned a nice
biscuit brown colour. An 85 °C capacitor would be a complete liability in
this environment.

 

[Arno Wagner]

There are plenty to chose from. Someone has borrowed my first catalogue,
but grabing another and allowing it to flop open open shows me the 056/7
series from BC Components (or for higher voltages the 157 series). ELP
Components ALP 20 and ALP 22 series (the difference is purely tolerance).

Ok. Thanks.

The confusion may arise because ESR is not a function of the maximum
temperature range of a capacitor. It is a function of the actual capacitor
temperature (roughly a rectangular asymptotic hyperbolic function). Thus
105 °C capacitors have the capability of possessing a lower ESR than their
85 °C counterparts - but only when operated at those higher temperatures.

Hmm. Maybe also that some manufacturers classify capacitors as low-ESR
at resistance values others don't find low enough for that attribute.

Looking at the 056 series from BC components I see that they have,
e.g., 79mR 4700uf/16V @ 100Hz. A Rubycon ZLH with 4700uF/16V has 13mR
at 100kHz. A NIC components NRSZ 4700uF/16V has 18mR at 100kHz.
Besides the Rubycon ZL (18mR at 100kHz for 3900uF/16V) these
are the only ones my distributor classifies as "low-ESR". Not really
the same class of component. However if you say that the 056 series is
sold as "low-ESR" by some people, I am willing to believe that....

Ok, sorry. While I have no source, I believe you that there are some
85C "low-ESR" capacitors on the market. Not a good choice for replacement
of CPU voltage regulator filter capacitors though. I have measured
65C on top of some Rubycon ZL capacitors I used as replacement.
They would probably not live long if they were 85C types with an ESR
like the 056 series...

Officially many manufacturers also claim that ESR is frequency dependant,
but resistance should not be a function of frequency. I believe they really
mean ESI (Equivalant Series Impedance), but this is probably too complex a
concept.

Probably. I have seen ESR measured values at 100Hz and at 100kHz.
In the end you are right that what is needed would be ESI or
at least a plot of the ESR over frequency. That also explains that
switchers that would need 10uF or less output filter capacitors
come with several 1000uF of low-ESR electrolyte filter capacitors.

I have worked on a project that required capacitors that required a
far lower ESR and ESI than anything you can ever find in any
catalogue. We had to get them specially made, but I can't tell you
how they were made, because I would then have to kill you.

Hmm. Current electrolyte-replacement ceramics are pretty good.
More expensive though.

A specific application for which I am not in the slightest bit surprised.
CPU voltage regulator capacitors are usually positioned extremely close to
the CPU itself, often have to operate at quite high temperatures, and are
purpose made for that application.

Indeed. See the 65C figure above.

Only the garbage turned out by Intel won't operate at much more than
40-50 °C without soft errors. But most serious processors will
opearate at 140 °C or more. Some of our PowerPC boards operate the
processor so hot that the board itself has turned a nice biscuit
brown colour. An 85 °C capacitor would be a complete liability in
this environment.

AMD starts to get flaky at 110C or so. My Nvidia 7600GT starts
to produce graphics errors at around 70C. Pretty bad, but
no real affordable alternatives on the market.

Arno

 

[Trent]

Ok, my knowledge may be outdated.

<GUFFAW>