Replace capacitors
M Jones
I have been given a faulty PC that freezes and shuts off frequently. The
owner says the fault has been happening more frequently to the point where
the machine does not boot any more. I had a look at the motherboard because
I experienced the fault before in another machine as leaking capacitors.
When I looked I could see indeed 10 capacitors are leaking / bulging.
Since I am quite adventurous and the machine is decent spec. I am going to
attempt to replace the capacitors and use the machine in my home. I have
checked and see they are all 6.3V 1500uF. I've gone into my local maplin
and asked for new ones only to be told they are not available but 16V 2200uF
will do. Since I have never done this before I thought I would ask if these
will be sufficient replacements for the ones that have gone or should I look
online for the correct ones?
Thanks.
Chris Whelan
Plenty of the correct ones available at RS...
Search in the archives of this ng for a thread where Conor shared his method
of doing this.
I've done a couple of motherboards with just one or two faulty caps; I'm not
sure I would attempt 10 however.
Let the group know the outcome.
Chris
--
Remove prejudice to reply.
M Jones
"Chris Whelan" <cawh...@prejudicentlworld.com> wrote in message
news:X1jrj.5275$N53....@newsfe1-win.ntli.net...
Thanks. I'll have a look.
All the ones leaking are of the same type and manufacture. All other types
on this board are OK. I am thinking these same ones installed were of poor
standard. It's an Intel board as well...
thoss
>M Jones wrote:
>
>> Hi all,
>>
>> I have been given a faulty PC that freezes and shuts off frequently. The
>> owner says the fault has been happening more frequently to the point where
>> the machine does not boot any more. I had a look at the motherboard
>> because I experienced the fault before in another machine as leaking
>> capacitors. When I looked I could see indeed 10 capacitors are leaking /
>> bulging.
>>
>> Since I am quite adventurous and the machine is decent spec. I am going to
>> attempt to replace the capacitors and use the machine in my home. I have
>> checked and see they are all 6.3V 1500uF. I've gone into my local maplin
>> and asked for new ones only to be told they are not available but 16V
>> 2200uF
>> will do. Since I have never done this before I thought I would ask if
>> these will be sufficient replacements for the ones that have gone or
>> should I look online for the correct ones?
>>
>> Thanks.
>
>Plenty of the correct ones available at RS...
>
But higher-voltage-rated ones, e.g. 16V 1500uF would be better.
--
Thoss
Lee
news:tLqdnbB4K4jDJjDa...@bt.com...
I have electronic experience mate been taught by some very experienced
engineers and I know its ok the cap has more voltage then original just not
less but not recommended to put a higher micro farad rating in its place.
cheers.
Lee
news:X1jrj.5275$N53....@newsfe1-win.ntli.net...
I have replaced over 10 caps on a motherboard before its not that hard but a
tip from me is to use a fine tip soldering iron ;-)
M Jones
"Lee" <joeo...@hotmail.com> wrote in message
news:FAjrj.71815$801....@fe1.news.blueyonder.co.uk...
I thought so. That's why I thought i'd ask first.
I'll order online and get the correct ones. I've tried RS but they are out
of stock. Do you know of anywhere else I can go and get these things?
M Jones
It doesn't seem all that bad to me either from what i've read. The tricky
part seems to be removal of the existing caps and clearing old solder
without damaging the board!
Dr.Hal0nf1r£$
That would depend upon the capacitor's intended purpose and the circuit in
which it was incorporated:
Certainly you are correct in that it is never a good idea to put a component
with a lower-rated working voltage in place of another as a general rule;
but the extra capacitance mey or may not be a problem.
If the capacitor's intended purpose is as a ripple-smoothing component or an
AC decoupling capacitor then under most circumstances the difference in
capacitance rating will make negligible if any difference to the circuit's
function, working at 16 or less volts. However if the capacitor is part of a
frequency-dependant circuit such as a stabilised ocillator or
pulse-switching multivibrator then the exact value is paramount; although
el;ectrolytic capacitors are so leaky in terms of voltage and have such a
wide tolerance margin that other types of more-stable capacitor are usually
incorporated in such a circuit.; such as mylar-film, polycarbonate, or
drilitic etc.
Having said that; the difference in capacitance between a 1500 microfarad
and a 2200 microfarad capacitor is quite substantial; and therefore to be on
the safe side I'd probably only attempt repair using a maximum of 1800
microfarads. You will find that the physical size of a given rating of
capacitor generally increases in proportion to its working voltage; however
you may just about be able to squeeze a 1600uF 20v capacitor into the
available space, but don't quote me on that. Don't fall into the trap of
size though; as it also holds true that the size of a given rating of
capacitor reduces proportionally along with its working voltage, so don't be
tempted to fit a 1600uF 12v capacitor - You *may* get away with it if you're
very lucky, as the rating tolerance of some components differs slightly from
that marked, but personally I wouldn't risk it. If you were to buy 1600uF
20v capacitors and they didn't physically fit in the available space then it
may be possible to insulate the leads, (Remember to observe the correct
polarity when fitting them.) and use the insulated leads as "stilts" with
the capacitor slightly above the other components.
As a general rule though; try to never use a lower voltage rated component
than the original as our friend suggests. In the case of capacitors of this
type, a little extra capacitance is usually ok; but using 2200uF in place of
1600uF may be pushing it, or then again it may not. (?) generally I have
found IMO that spotty teenagers and Ł5-an-hour assistants behind a desk in
electronics shops talk out of the wrong hole and get a power-trip from doing
their job: Just as a dog isn't a horse if it's born in a stable, so a
school-leaver doesn't become a qualified engineer or technician by selling
electronic components. Seek the advice of an electronics engineer if you
wish, or wait until you have the near identical components and you can't go
wrong.
(I'm a qualified electronics technician BTW, FYI.( C&G 300,301))
HTH
That's my piece of eight point five seven nine recurring anyway.
Dr.Hal0nf1r£$
Some of these suppliers may be OK;
http://www.google.co.uk/search?hl=en&q=UK+electronic+component+suppliers&btnG=Google+Search&meta=
generally I stick to RS components and Maplin, though Farnell have an
excellent reputation.
HTH
Dr.Hal0nf1r£$
> M Jones writes:
>> Since I am quite adventurous and the machine is decent spec. I am
>> going to attempt to replace the capacitors and use the machine in my
>> home. I have checked and see they are all 6.3V 1500uF. I've gone
>> into my local maplin and asked for new ones only to be told they are
>> not available but 16V 2200uF will do. Since I have never done this
>> before I thought I would ask if these will be sufficient
>> replacements for the ones that have gone or should I look online for
>> the correct ones?
>
> Sometimes they say that changing the uF value is okay, but the vast
> vast majority of the time it is a Very Bad Thing to do. Voltage
> rating can be increased to whatever you want, but never decreased.
>
> For recapping you need a decent 50-watt soldering station (a 30W iron
> is fine only for some low-melting-point solder)
!!!!That is one enormously over-powered iron!!!! I'm surprised you haven't
overheated neighboring components or seperated track from board where
possible with it! If you can't melt ordinary solder with a 15-watt iron then
there is something wrong with the voltages in your house-wiring. The entire
job should be easy with a 25 watt iron and possible with a 15 watt. Jeez;
it's a sensitive electronic circuit board not an automobile chassis!
Lee
tip to removing old solder with damaging the flux area is to put new solder
around area first then suck that up wi your solder sucker it will help ;-)
cheers
Conor
> It doesn't seem all that bad to me either from what i've read. The tricky
> part seems to be removal of the existing caps and clearing old solder
> without damaging the board!
>
You don't desolder them. Get a pair of pliers, grip the capacitor and
pull. It'll leave the two pins standing proud of the motherboard.
Solder the capacitors directly to them.
--
Conor
As a Brit I'd like to thank the Americans for their help in the war
against terror because if they'd not funded the IRA for 30 years, we
wouldn't know how to deal with terrorists.
Conor
> For recapping you need a decent 50-watt soldering station
Are you sure? I've been using no more than half that for the past two
decades.
John Jordan
>
> I'll order online and get the correct ones. I've tried RS but they are out
> of stock. Do you know of anywhere else I can go and get these things?
I used CPC for this.
--
John Jordan
Mike Tomlinson
m.kustomkomputa.co.uk.invalid> writes
>!!!!That is one enormously over-powered iron!!!! I'm surprised you haven't
>overheated neighboring components or seperated track from board where
>possible with it! If you can't melt ordinary solder with a 15-watt iron then
>there is something wrong with the voltages in your house-wiring. The entire
>job should be easy with a 25 watt iron and possible with a 15 watt. Jeez;
>it's a sensitive electronic circuit board not an automobile chassis!
It's obvious you've never done it.
The capacitors that fail have one or both pins soldered to the voltage
planes in the board - large sheets of copper, which carry heat away from
the iron very rapidly. You need to use a large iron to get the joint
hot enough to melt the solder. The best way is to have two irons - a
25W and a 50W. Use the 25W first and if that won't melt the joint
(because you've hit a voltage plane), then use the 50W.
Clearing the holes can be tricky - as another poster said, adding a bit
of fresh solder before using the solder-sucker usually works.
--
(\__/) Bunny says NO to Windows Vista!
(='.'=) http://www.cs.auckland.ac.nz/~pgut001/pubs/vista_cost.html
(")_(")
linker3000
> (I'm a qualified electronics technician BTW, FYI.( C&G 300,301))
>
Nice summary but you only get 9.5/10 as you forgot to mention that you
shouldn't really run electrolytics at below 10% of their rated voltage
as this may lead to the incomplete re/formation of the oxide insulating
layer and thus higher leakage currents.
In this case, though, if the caps are rated at 6.3v and over the 5V
rail, you'd have to be using 50V/63V rated ones before you run that risk
+ they would probably be too big to fit anyway!
L3K
MIElecIE (lapsed!)
TEng
linker3000
M Jones
"Sam" <a...@ng.invalid> wrote in message
news:uc8sq3tkt44av2lr5...@4ax.com...
> Conor writes:
>> > For recapping you need a decent 50-watt soldering station
>>
>> Are you sure? I've been using no more than half that for the past two
>> decades.
>
> suggested a 50W - obviously with temp. control. A hotter iron can get
> the job done quicker, and the less time it has touching the solder/PCB,
> the less chance of overheating the traces/components.
I have a 25 watt iron and I thought that would be fine for such a small job
as this.
GSV Three Minds in a Can
Mike Tomlinson <mi...@jasper.org.uk> said
<snip>
>Clearing the holes can be tricky - as another poster said, adding a bit
>of fresh solder before using the solder-sucker usually works.
Or adopt plan B - just solder the new capacitor to the leads left over
from the failed on (if you can snip the lead to leave a reasonable size
target).
--
GSV Three Minds in a Can
10,414 Km walked. 2,032 Km PROWs surveyed. 36.9% complete.
M Jones
Thanks for the help guys. Would these fit the bill ? :-
http://uk.farnell.com/jsp/search/productdetail.jsp?sku=1219448&N=401
Looks like it will only cost me a fiver if they are.
Dr.Hal0nf1r£$
25W is perfectly adequate if you use the correct soldering technique.
Johnny B Good
from "M Jones" <mat...@privatemail.me.uk> contains these words:
It does rather depend on the design of the iron. An iron using the
traditional Antex 'sleeve bit' design is much more effective than even
the later Antex version of the 'Other' brand design where the bit slides
_into_ the heating element. Antec's version of the 'Other' brand method
uses a much longer inserted section, but even this is nowhere near as
effective as their origional 'sleeve bit' design.
The problem with the 'Other' bit insert style, especially with the
cheap ones with just a small amount of the bit material used to couple
the heat from the heating element, is the much higher 'runaway'
temperatures reached between applications of the bit to the solder
joint.
--
Regards, John.
Please remove the "ohggcyht" before replying.
The address has been munged to reject Spam-bots.
Johnny B Good
from Sam <a...@ng.invalid> contains these words:
> Conor writes:
> > > For recapping you need a decent 50-watt soldering station
> >
> > Are you sure? I've been using no more than half that for the past two
> > decades.
> My 25W iron wouldn't melt the solder on a router's PCB; that's why I
> suggested a 50W - obviously with temp. control. A hotter iron can get
> the job done quicker, and the less time it has touching the solder/PCB,
> the less chance of overheating the traces/components.
It rather depends on the design of the iron. If you have the early, and
much superior, Antex type (where the bit fits _over_ the heating
element), this would be just about right for the job. If you have a
cheaper brand of iron (including, very sadly, the later models of Antex)
where the heating element _surrounds_ the bit, you'll indeed something
like a 40 to 50 watt rated iron to get the same soldering power effect.
Johnny B Good
from GSV Three Minds in a Can <G...@quik.clara.co.uk> contains these words:
> Bitstring <FCeS0cB+...@jasper.org.uk>, from the wonderful person
> Mike Tomlinson <mi...@jasper.org.uk> said
> <snip>
> >Clearing the holes can be tricky - as another poster said, adding a bit
> >of fresh solder before using the solder-sucker usually works.
> Or adopt plan B - just solder the new capacitor to the leads left over
> from the failed on (if you can snip the lead to leave a reasonable size
> target).
The only problem with this is the extra lead inductance this introduces
into the circuit. This is probably acceptable if you can keep such lead
extension to less than a centimetre in total (5mm per lead), but is a
definite no no with centimetre lengh leads as a means of wiring a much
physically larger component into circuit.
Johnny B Good
from "Dr.Hal0nf1r£$" <fem...@nospam.kustomkomputa.co.uk.invalid>
contains these words:
And an Antex sleeve bit iron. Antex now seem to only use this design
for their lower power range up to 20 watt max with the less optimal bit
insert design for the 25 watt and above models. :-(
I used to use their 16/18 watt sleeve bit design model for this work,
but it was barely up to the job. I reckon a 25/30 watt version would
have been just the ticket but Antex don't do sleeve bit types in the
higher ratings (unless they're still doing the 35 watt 50v models they
used to supply BT with). As a consequence I've had to make do with their
'Bit Insert' type 40 watt model which has only proved to be a marginal
improvement over the 16/18 watt model.
Johnny B Good
from "Dr.Hal0nf1rŁ$" <fem...@nospam.kustomkomputa.co.uk.invalid>
contains these words:
Normally, such repair work at the component level (the component, in
this case, being electrolytic capacitors), isn't usually considered a
cost effective option. However, winXP's WPA process makes such a repair
infinitely more attractive than the usual MoBo replacement exercise,
especially when your planned upgrade is still a good 2 or 3 years away
and you'd rather not have to invest in not only a modern MoBo, but also
new memory and processor right now.
Having stated why such repair work is even worth attempting, it's worth
remembering that we're talking about _electrolytic_ capacitors, not
mylar film or paper or polyester types and the first thing to keep in
mind is the standard value tolerance range of -20/+80%. The second thing
to keep in mind is that they are used in circuits where 'more is
better', i.e a larger value will do no harm, and will generally improve
performance.
It's also worth bearing in mind that the value worked out in the design
phase has been arrived at by making due allowance of that -20% tolerance
figure and picking the next standard available value above the minimum
desired value.
Since there can be a very wide range of values that can be used in any
one such circuit the manufacturer can often reduce manufacturing costs
by 'overspecifying' to a more restricted value range wherein a much
larger quantity of each capacitor value can be ordered into stock at
more favourable bulk discount rates.
The usual voltage ratings are mostly 6.3 and 16 volts with the odd 10v
rated capacitor used here and there. At these voltage ratings, a 16 volt
capacitor will be roughly twice as bulky as its 6.3 volt counterpart, so
it's worth avoiding the use of higher voltage versions of replacements
unless you know those replacements will physically fit in the space of
the originals.
Although it's quite obvious (or damn well should be!), that the voltage
rating of the replacement should match or exceed that of the original,
you can sometimes specify a lower voltage than that of the original, but
only if the originals were overspecified to begin with.
On MoBos you can spot this state of affairs where the manufacturer has
used 10v caps as a substitute for the more optimally rated 6.3v parts
since there aren't any other voltage levels between the 5 and 12 volt
buses. 10 volt capacitors can't be specified on a 12v line so just
_have_ to be substituting for 6.3v parts on the 5v bus.
I've repaired quite a few MoBos over the years and have never had to
actually buy any capacitors since I've acquired quite a large collection
of scrapped MoBos from which to recover perfectly good quality
capacitors from.
I can be pretty confident that this source of supply is no more likely
to fail than those bought in 'brand new' from a component supplier. In
fact, I can be more confident that they'll be just as good, if not
better than, 'brand new' capacitors!
Having said that, even if I were to specially order in supplies, I'd be
inclined to rather restrict the capacitance range to a very small one,
especially in the 6.3v range. Most likely, restricted to just 1800uF and
2200uF values since this covers most situations (the 1800uF would, for
instance, nicely substitute for 1000, 1200, 1500 uF values).
If I'm replacing 1800uF caps, and I can't find enough of that value
which would physically fit, I'd choose 1500uF as a substitute since (all
other considerations aside[1]), a slightly lower value is infinitely
better than the existing high resistance versions I'm trying to replace.
[1] The other considerations being that -20/+80% tolerance figure and
the manufacturer's choice of a value in excess of actual design needs,
which could well be 2 or 3 times larger due to inventory considerations.
HTH
CJK
For a motherboard you really need low ESR capacitors, others risk
overheating
or poor smoothing performance. Remember these are passing significant
currents
on the low voltage rails (processor, 3.3 & 5v supplies).
Farnell do these.
Chris K
M Jones
"CJK" <dono...@spamtrap.com> wrote in message
news:13qud1k...@corp.supernews.com...
I've checked the site and farnell say they are low ESR. I've done another
search and can't find anything else other than these.
M Jones
I found some info. It would appear they are.
CJK
Yes, they look OK, good luck with the soldering.
Chris K
fred
<mat...@privatemail.me.uk> writes
Panasonic are a good make.
105degC rating is premium quality for long life.
High ripple current (2.2A) quoted at high frequency (100kHz).
Ultra low impedance (20milliohms at 100k).
If you find FEC funny about dealing with joe public, their consumer
friendly arm CPC will have something similar.
Funny this thread should come up now, my movie PC croaked and I've just
changed a couple of the caps to resurrect it, running memtest right now.
Just be gentle trying to get the caps out, make sure the solder is
molten right through the hole & give it a little wiggle to be sure
before pulling. I ease one side at a time pulling half way each side at
a time so the cap comes out on 2 hits per side. Solid power planes will
soak up a lot of heat.
--
fred
Plusnet - I hope you like vanilla
Dr.Hal0nf1r£$
Do Antex still make the X25? That was my first soldering iron; 25 watts with
a sleeve-bit.
Mike Tomlinson
<G...@quik.clara.co.uk> writes
>Or adopt plan B - just solder the new capacitor to the leads left over
>from the failed on (if you can snip the lead to leave a reasonable size
>target).
Yeah, it's a terrible bodge though, and the purist in me rather objects
to it. I'm a great believer in "if the job's worth doing" etc.
M Jones
"fred" <n...@for.mail> wrote in message news:mMvpfHDxi0rHFw76@y.z...
Once I get them i'll have a go and post the results back here.
Mike Tomlinson
<jcs.comp...@plugzetnet.co.uk> writes
>The second thing
>to keep in mind is that they are used in circuits where 'more is
>better', i.e a larger value will do no harm, and will generally improve
>performance.
Up to a point, Lord Copper.
> I've repaired quite a few MoBos over the years and have never had to
>actually buy any capacitors since I've acquired quite a large collection
>of scrapped MoBos from which to recover perfectly good quality
>capacitors from.
>
> I can be pretty confident that this source of supply is no more likely
>to fail than those bought in 'brand new' from a component supplier. In
>fact, I can be more confident that they'll be just as good, if not
>better than, 'brand new' capacitors!
I'm sorry Johnny, I have to disagree here. Electrolytic capacitors
start to deteriorate from the moment they are installed in a circuit.
They have a finite life, usually measured in a few thousands of hours,
and particularly in CPU power circuits, where they: 1) are expected to
absorb very large changes in current as the CPU load changes, and 2) are
subjected to localised heat from the MOSFET transistors to whose legs
they are soldered (a direct heat transfer path), they lead a pretty hard
life. To re-use capacitors in such applications is IMHO a false
economy, especially as they are available cheaply new, branded and with
elevated temperature spec (105degC).
>Most likely, restricted to just 1800uF and
>2200uF values since this covers most situations (the 1800uF would, for
>instance, nicely substitute for 1000, 1200, 1500 uF values).
Too much of a bodge. I'd always try and get as close as poss to the
original installed value. I might sub 1800uF for 1500uF but not vice
versa.
> If I'm replacing 1800uF caps, and I can't find enough of that value
>which would physically fit, I'd choose 1500uF as a substitute since (all
>other considerations aside[1]),
I wouldn't.
> a slightly lower value is infinitely
>better than the existing high resistance versions I'm trying to replace.
Swapping one problem for another IMHO.
Mike Tomlinson
>Thanks for the help guys. Would these fit the bill ? :-
>
>http://uk.farnell.com/jsp/search/productdetail.jsp?sku=1219448&N=401
>
>Looks like it will only cost me a fiver if they are.
Yes, they're a good choice.
M Jones
Farnell don't accept orders below £20.00 and CPC don't have them. I've
tried RS and this model number is out of stock. Would this one suit? It
looks the same to me.
http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=4490851#header
al
> In article <2--dnUHWlKUfcDPanZ2dnUVZ8rGdn...@bt.com>, M Jones
Have you tried just removing the bad caps? you may just get away with
it.
Al
fred
<mat...@privatemail.me.uk> writes
>>>> http://uk.farnell.com/jsp/bespoke/bespoke7.jsp?bespokepage=farnell/en/ed_wor
>ld/products/new/2006/wk7/panasonic.jsp
>>>
>>> I found some info. It would appear they are.
>>>
>>
>> Yes, they look OK, good luck with the soldering.
>>
>> Chris K
>>
>
>Farnell don't accept orders below £20.00 and CPC don't have them. I've
>tried RS and this model number is out of stock. Would this one suit? It
>looks the same to me.
>
>http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct
>&R=4490851#header
>
motherboard which is still going strong after 2yrs. They're not as good
as the ones you spotted at FEC (half the ripple rating and 2.5x the
impedance) but still acceptable in my view. I'd say ok to use.
Sorry about the pointer to CPC, their range of high spec caps really has
turned to shit.
><(((°>
"Mike Tomlinson" <mi...@jasper.org.uk> wrote in message
news:FCeS0cB+...@jasper.org.uk...
> In article <BeudncaOHN8...@bt.com>, Dr.Hal0nf1rŁ$ <femail@nospa
Agreed regarding the iron wattage - I've had trouble with a temp controlled
Weller 175W iron on some legs on the ground plane.
Although I can always get the bad component out, I sometimes struggle to
suck out the hole.
Occasionally I've resorted to drilling it out using a 0.5mm drill in a pin
chuck carefully hand turned.
I also rob components from dead motherboards.
Note also that capacitor plague is not restricted to motherboards.
I've often repaired graphics cards with a few bulgers and I suspect it
occurs in PSU's too although I never bother attempting to repair them.
Johnny B Good
from Mike Tomlinson <mi...@jasper.org.uk> contains these words:
> In article <3130303037373...@plugzetnet.co.uk>, Johnny B Good
> <jcs.comp...@plugzetnet.co.uk> writes
> >The second thing
> >to keep in mind is that they are used in circuits where 'more is
> >better', i.e a larger value will do no harm, and will generally improve
> >performance.
> Up to a point, Lord Copper.
> > I've repaired quite a few MoBos over the years and have never had to
> >actually buy any capacitors since I've acquired quite a large collection
> >of scrapped MoBos from which to recover perfectly good quality
> >capacitors from.
> >
> > I can be pretty confident that this source of supply is no more likely
> >to fail than those bought in 'brand new' from a component supplier. In
> >fact, I can be more confident that they'll be just as good, if not
> >better than, 'brand new' capacitors!
> I'm sorry Johnny, I have to disagree here. Electrolytic capacitors
> start to deteriorate from the moment they are installed in a circuit.
> They have a finite life, usually measured in a few thousands of hours,
The actual service lifetime is very dependent on how close to their
ripple current and temperature limits they are operated at, but I'd have
expected a service life in the region of 25 to 30 thousand hours when
run close to, but not exceeding these limits. The increased service
lifetime can easily exceed an order of magnitude improvement with a
quite modest reduction of temperature and ripple current limits.
> and particularly in CPU power circuits, where they: 1) are expected to
> absorb very large changes in current as the CPU load changes, and 2) are
> subjected to localised heat from the MOSFET transistors to whose legs
> they are soldered (a direct heat transfer path), they lead a pretty hard
> life. To re-use capacitors in such applications is IMHO a false
> economy, especially as they are available cheaply new, branded and with
> elevated temperature spec (105degC).
I appreciate your concerns, but the ones taken from scrapped boards are
rated at 105 deg C and have cleared the infant mortality part of the
'bathtub' curve thus, effectively, passing an extended 'burn in' test.
> >Most likely, restricted to just 1800uF and
> >2200uF values since this covers most situations (the 1800uF would, for
> >instance, nicely substitute for 1000, 1200, 1500 uF values).
> Too much of a bodge. I'd always try and get as close as poss to the
> original installed value. I might sub 1800uF for 1500uF but not vice
> versa.
That, too, is my preferred choice (using a slightly higher value than
the original), but, knowing the very wide tolerance range for
electrolytic capacitor values and the tendency to 'over-spec' (typically
by half an order of magnitude or more in this type of application), it's
extremely unlikely that a slightly lower value substitute would cause
any problems whatsoever.
> > If I'm replacing 1800uF caps, and I can't find enough of that value
> >which would physically fit, I'd choose 1500uF as a substitute since (all
> >other considerations aside[1]),
> I wouldn't.
> > a slightly lower value is infinitely
> >better than the existing high resistance versions I'm trying to replace.
> Swapping one problem for another IMHO.
I don't see that. Would you care to expand?
Johnny B Good
Not as far as I could see when I was looking for higher wattage
versions of my CS18 (16/18 watts 220/240v) iron last year. :-(
Mike Tomlinson
> The actual service lifetime is very dependent on how close to their
>ripple current and temperature limits they are operated at,
Agreed. But you forget that the caps used in CPU power ccts do have to
deal with high ripple and temperatures. The current (several amps)
drawn by the CPU rises and falls near instantaneously as the processor
load rises and falls. The capacitors in the CPU voltage regulator
circuit are there to accommodate and smooth out this rapidly-changing
demand. They are also closely thermally coupled to the hot-running
MOSFETs that form part of the switching power supply circuit, so they
run hot too.
> but I'd have
>expected a service life in the region of 25 to 30 thousand hours
Heh. Try again. For a typical high-quality, branded, low-ESR cap used
in the circuits we're discussing, it's between 1000 hours and 5000
hours. And use in a CPU VRM is going to be at the low end of that
scale.
http://docs-europe.electrocomponents.com/webdocs/00ba/0900766b800bafec.p
df
> I appreciate your concerns, but the ones taken from scrapped boards are
>rated at 105 deg C and have cleared the infant mortality part of the
>'bathtub' curve thus, effectively, passing an extended 'burn in' test.
Caps don't suffer from infant mortality, and the "extended burn in test"
only serves to reduce their total working lifetime. They're so cheap to
buy that reusing them is a false economy. Re-capping a board originally
fitted with mediocre Taiwanese caps with high-quality Panasonic
replacements only costs 5 to 8 quid plus an hour of your time.
The trick, of course, is identifying the caps that do need replacement -
it's only necessary to do the ones that form part of the CPU voltage
regulator circuitry and the memory 2.5V supply. Usually no more then
ten in total, and it's important to do them all, not just the ones that
have obviously (visually) failed, since that ones that look OK won't be
far behind.
> That, too, is my preferred choice (using a slightly higher value than
>the original), but, knowing the very wide tolerance range for
>electrolytic capacitor values and the tendency to 'over-spec' (typically
>by half an order of magnitude or more in this type of application), it's
>extremely unlikely that a slightly lower value substitute would cause
>any problems whatsoever.
Until you upgrade the CPU to the fastest (= most power hungry) supported
by that board and wonder why the machine won't run stably, 'cos the
voltage regulation now can't cope with the current demands of the
replacement CPU. That in turn puts additional strain on the power
supply.
> I don't see that. Would you care to expand?
See my last comment above.
If I were your customer and I know you were fixing my board with re-used
caps and charging me for it, I would not be a happy bunny.
Mike Tomlinson
.com>, al <alista...@gmail.com> writes
>Have you tried just removing the bad caps? you may just get away with
>it.
What a stupid suggestion.
M Jones
I was thinking that also. I would never just remove the caps and then turn
the power on!
John Jordan
> In article <3130303037373...@plugzetnet.co.uk>, Johnny B Good
> <jcs.comp...@plugzetnet.co.uk> writes
>
>> The actual service lifetime is very dependent on how close to their
>> ripple current and temperature limits they are operated at,
>
> Agreed. But you forget that the caps used in CPU power ccts do have to
> deal with high ripple and temperatures. The current (several amps)
> drawn by the CPU rises and falls near instantaneously as the processor
> load rises and falls. The capacitors in the CPU voltage regulator
> circuit are there to accommodate and smooth out this rapidly-changing
> demand. They are also closely thermally coupled to the hot-running
> MOSFETs that form part of the switching power supply circuit, so they
> run hot too.
Absolutely nowhere near 105C though. Electrolytic capacitor lifespan
rises exponentially as temperature falls, so most boards are designed to
run their capacitors closer to 50C than 100C, even under full load.
>> but I'd have
>> expected a service life in the region of 25 to 30 thousand hours
>
> Heh. Try again. For a typical high-quality, branded, low-ESR cap used
> in the circuits we're discussing, it's between 1000 hours and 5000
> hours. And use in a CPU VRM is going to be at the low end of that
> scale.
If it actually ran at 105C. Even capacitors from the bad-caps era
generally lasted 5-10 thousand hours in practice.
> The trick, of course, is identifying the caps that do need replacement -
> it's only necessary to do the ones that form part of the CPU voltage
> regulator circuitry and the memory 2.5V supply.
Not sure where you get this idea from. All VRM caps can and do fail. In
many cases the CPU and memory VRM caps are fine, and it's the lesser
VRMs for northbridge/southbridge/AGP etc that fail.
Often these are the boards that are most worth fixing, because there's
just one unique cap on the northbridge that's failed.
--
John Jordan
al
> "Mike Tomlinson" <m...@jasper.org.uk> wrote in message
>
> news:oWMZRQBp...@jasper.org.uk...
>
> > In article <d70ae74f-accd-493c-b78a-42a67e3f6...@e4g2000hsg.googlegroups
> > .com>, al <alistairle...@gmail.com> writes
>
> >>Have you tried just removing the bad caps? you may just get away with
> >>it.
>
> > What a stupid suggestion.
>
> > --
> > (\__/) Bunny says NO to Windows Vista!
> > (='.'=) http://www.cs.auckland.ac.nz/~pgut001/pubs/vista_cost.html
> > (")_(")
>
> I was thinking that also. I would never just remove the caps and then turn
> the power on!
Why not, constructive criticism welcome, if its only for smoothing you
can get away with removing a fair amount, I've seen it done (as a cost
reduction).
PS there are no bad ideas, some are better than others.
Regards
Al
al
> In article <d70ae74f-accd-493c-b78a-42a67e3f6...@e4g2000hsg.googlegroups
>
> >Have you tried just removing the bad caps? you may just get away with
> >it.
>
> What a stupid suggestion.
>
If the caps are faulty it won't make it any worse (boards not posting
anyway). My thinking is the 10 caps didn't fail all at once, probably
only one caused the no post. Removing some may get the board running
again.
Al
Dr.Hal0nf1r£$
> In article
> <d70ae74f-accd-493c...@e4g2000hsg.googlegroups .com>,
> al <alista...@gmail.com> writes
>
>> Have you tried just removing the bad caps? you may just get away
>> with it.
>
> What a stupid suggestion.
I was thinking that too; but I was so gobsmacked at it that I didn't know
what to say in reply
.
Dr.Hal0nf1r£$
Why not remove a RAM stick too? There's at least one more and it'll still
work with just the one - And why bother with a graphics card? The chipset
can run the onboard graphics: Who needs it eh?...And a DVD-RAM? - Why are
people so lazy that they can't be bothered to type in megabytes of program?
as for the power indicator light; we know when the bloody thing's on because
it's working. - And we can do away with the HD indicator too: With a little
concentration we can hear it working. As for that cooling fan; exercise your
lungs and use your breath to keep the CPU cool ...
The money-saving computing alternative from al: No components, no wasted
electricity; just an open case with a pencil, paper, and a pocket calculator
inside. Save ŁŁŁs!
Dr.Hal0nf1r£$
Was that you I saw on that oil rig throwing bread to the helicopters?
Mike Tomlinson
<ju...@jaj22.org.uk> writes
>Absolutely nowhere near 105C though.
They don't need to. Heat causes the electrolyte to dry out, the
capacitance diminishes and the ESR (equivalent series resistance)
increases. The hotter the cap is, the faster this process occurs. And
caps in CPU VRMs do run hot. They also tend to be bunched together,
which doesn't help.
> Electrolytic capacitor lifespan
>rises exponentially as temperature falls, so most boards are designed to
>run their capacitors closer to 50C than 100C, even under full load.
So? 105C is the maximum temperature rating of such a capacitor, not a
target. You seem to think I'm saying that they all run at 105C. I'm
not.
>If it actually ran at 105C.
No. Read the datasheet I linked to.
> Even capacitors from the bad-caps era
>generally lasted 5-10 thousand hours in practice.
I think not, if Panasonic only rate theirs at 1000 to 5000 hours.
>Not sure where you get this idea from.
Experience. 20 years in the electronic repair trade, including repair
of motherboards, printers monitors etc. to component level.
>Often these are the boards that are most worth fixing, because there's
>just one unique cap on the northbridge that's failed.
Not in my experience. It's usually the CPU VRMs as that works the
hardest and runs hottest (think about why there has been a recent trend
toward putting heatsinks on the processor VRM.)
Mike Tomlinson
s.com>, al <alista...@gmail.com> writes
>If the caps are faulty it won't make it any worse (boards not posting
>anyway). My thinking is the 10 caps didn't fail all at once, probably
>only one caused the no post. Removing some may get the board running
>again.
Sure, and what is the resultant waveform to your expensive CPU going to
look like?
John Jordan
> In article <61bgb8F...@mid.individual.net>, John Jordan
> <ju...@jaj22.org.uk> writes
>
>> rises exponentially as temperature falls, so most boards are designed to
>> run their capacitors closer to 50C than 100C, even under full load.
>
> So? 105C is the maximum temperature rating of such a capacitor, not a
> target. You seem to think I'm saying that they all run at 105C. I'm
> not.
And I'm saying that if you run them substantially below 105C, they work
for orders of magnitude longer. Do you disagree with this?
Any motherboard that runs a capacitor at anywhere near 105C under any
circumstances is very badly designed.
>> If it actually ran at 105C.
>
> No. Read the datasheet I linked to.
I did. It gives the lifespan at 105C, and no values for lower
temperatures at all.
>> Even capacitors from the bad-caps era
>> generally lasted 5-10 thousand hours in practice.
>
> I think not, if Panasonic only rate theirs at 1000 to 5000 hours.
At 105C.
> Not in my experience. It's usually the CPU VRMs as that works the
> hardest and runs hottest (think about why there has been a recent trend
> toward putting heatsinks on the processor VRM.)
Yes, CPU VRM failures are more common, but I've seen (and in a few
cases, fixed) numerous boards with dead caps that weren't on the CPU or
memory VRMs. All of these were manufactured in the bad caps era though.
--
John Jordan
thoss
Tell us what happened when you tried it.
--
Thoss
Daniel James
> > Even capacitors from the bad-caps era
> >generally lasted 5-10 thousand hours in practice.
>
> I think not, if Panasonic only rate theirs at 1000 to 5000 hours.
Well, 1000 hours is only 6 months at 8 hours a day, five days a week;
5000 hours is 2.5 years. Even boards that have failed because of faulty
caps have usually lasted that long -- boards with good caps last much
longer (e.g. 8 years and counting for this PC, which often runs for
weeks at a time without rebooting).
So, either you're missing a zero or two from those Panasonic ratings,
or they last much longer at the actual working temps than the 105C
figure quoted.
Cheers,
Daniel.
Johnny B Good
from Daniel James <waste...@nospam.aaisp.org> contains these words:
I agree, when I saw the figures for those Panasonic caps, I thought
"Domestic grade - not fit for purpose!".
John Jordan
The latter. I found a general formula for electrolytic capacitor
lifespan before, but Google is failing me at the moment. It was
something like:
actual lifespan = rated lifespan * 2^((rated temp - actual temp) / 10)
I did find an example graph. See 5.3.1.3 here:
http://books.google.com/books?id=a988UyrJttYC&pg=PA90&lpg=PA90&dq=electrolytic+capacitor+lifespan+temperature&source=web&ots=kPNBii0AhO&sig=vkxc4x8C9Dx0qn88yq9qxCtjAfE#PPA89,M1
http://tinyurl.com/ypp4qa
--
John Jordan
al
> In article <6d0907ad-9ed8-4292-918e-164019ebe...@f10g2000hsf.googlegroup
> s.com>, al <alistairle...@gmail.com> writes
>
> >If the caps are faulty it won't make it any worse (boards not posting
> >anyway). My thinking is the 10 caps didn't fail all at once, probably
> >only one caused the no post. Removing some may get the board running
> >again.
>
> Sure, and what is the resultant waveform to your expensive CPU going to
> look like?
>
> --
> (\__/) Bunny says NO to Windows Vista!
> (='.'=) http://www.cs.auckland.ac.nz/~pgut001/pubs/vista_cost.html
> (")_(")
No worse than before :-)