Searching \ for '[EE] Bulging Electrolytics' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=bulging+electrolytics
Search entire site for: 'Bulging Electrolytics'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Bulging Electrolytics'
2010\07\14@102756 by Roger Weichert

flavicon
face
Hi guys,

I've been going to ask for a while about possible causes for the failure of
some 10,000uF 35v electros that are used in a basic power supply for a
pulsation controller.

There are 7 of them in parallel and on a couple of occasions I have seen
where nearly all have swollen and split their tops and leaked and dried out.
In fact they rattle quite noticeably when you shake them.

>From my understanding I thought that electros only burst if they are fed
with AC or if their voltage rating is exceeded.

In both cases the rectifiers were ok and on fitting new caps the supplies
worked fine.

The originals are rated at 35 volts and even unloaded have a maximum of 15
volts applied. The 240v input would have to more than double to get the
secondries anywhere near 35v !!!

The originals were also only rated at 85 degrees, and I've fitted 105 degree
replacements, but while excessive temperature can dry them out, I wouldn't
have thought it would cause this problem.

I've uploaded an image at    http://imagebin.ca/view/HX_1Ati.html

Any thoughts please.

Regards,   Roger

2010\07\14@103919 by John Hansen

flavicon
face
I don't know whether you've been following the controversy with Dell
computer, but it involved exactly this issue (though not with power supply
caps).  They had this issue with motherboard capacitors that were defective
and exhibited this failure mode.  It was not because the capacitors were
subjected to overvoltage but rather because they were mis-manufactured.  One
batch of Dell servers experienced a 97% failure rate.  Dell has taken some
heat on this because they went to some lengths to disavow the problem and
occasionally blamed the customers for the issue.

If I remember correctly, the capacitor manufacturer in question was
Nichicon.

John Hansen

On Wed, Jul 14, 2010 at 10:27 AM, Roger Weichert <spam_OUTrweicherTakeThisOuTspambigpond.net.au>wrote:

{Quote hidden}

> -

2010\07\14@103935 by AK

flavicon
face
They are just smiling because they are happy!
Joking aside, Have you compared their maximum ripple current rating,
and the ripple current that the pulsation controller loads them with,
during the pulses?


On Wed, Jul 14, 2010 at 10:27 AM, Roger Weichert
<.....rweicherKILLspamspam@spam@bigpond.net.au> wrote:
{Quote hidden}

>

2010\07\14@104520 by Alan B Pearce

face picon face
> Hi guys,
>
> I've been going to ask for a while about possible causes for the
failure of
> some 10,000uF 35v electros that are used in a basic power supply for a
> pulsation controller.

 ^^^^^^^^^^^^^^^^^^^^
I suspect this bit is the cause of your problem.

What is being done during a pulse? I suspect you are heating the
capacitors due to high IR losses during a pulse causing the electrolyte
to boil off eventually.
--
Scanned by iCritical.

2010\07\14@105432 by Olin Lathrop

face picon face
Roger Weichert wrote:
> I've been going to ask for a while about possible causes for the
> failure of some 10,000uF 35v electros that are used in a basic power
> supply for a pulsation controller.

The basic causes for electrolytic cap failures are:

1 - Poor manufactured quality.  This is one of those things where
   brand does matter.

2 - Voltage stress.  There isn't a hard limit at the voltage rating,
   so creative specsmanship can be a factor (see #1).  For long
   life it's a good idea to derate the max voltage by a comfortable
   margin.

3 - Temperature.  Again, no hard limit, reputable suppliers may
   leave more margin in their specs.  Derate for long life.  Note
   that high ripple current causes dissipation, which increases
   temperature, which decreases life.  Caps meant for this kind of
   application have a ripple current spec, like Panasonic FK
   series.

> I've uploaded an image at http://imagebin.ca/view/HX_1Ati.html

It's a little hard to tell how tall these caps are, but they look
surprisingly small for 10mF and 35V.  Does it look like 6 Joules can fit in
there to you?  Look up the size of caps of the same rating from reputable
vendors.  If those turn out to be bigger, then it's pretty clear what
happened.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\07\14@111446 by Roger Weichert

flavicon
face

----- Original Message -----
From: "Olin Lathrop" <olin_piclistspamKILLspamembedinc.com>
To: "Microcontroller discussion list - Public." <.....piclistKILLspamspam.....MIT.EDU>
Sent: Thursday, July 15, 2010 12:24 AM
Subject: Re: [EE] Bulging Electrolytics


{Quote hidden}

I'm not sure about the brand of the originals, and just for a change I threw
the crook ones away rather than leave them lying around the bench. I guess
if I get any more, I can try and look up their specs.

While these still look reasonably new they have been in daily service for
around 7 years, so have lasted reasonably well  ...  it's just their way of
failing that surprised me.


{Quote hidden}

I actually ordered Panasonic caps from Farnell, but the ones I got had no
Panasonic markings on them that I recognised, and they have supplied two
different types so far.

I went for 105 degree caps, which limited my choices due to size restraints.
Maybe I should have been worrying more about ripple current.

Looks like I better study up on it.


{Quote hidden}

The originals are 25mm diam by 40mm high. The 105 degree replacements are
45mm high. They just :) fit with 2mm headroom!

Thanks to everyone for the response so far.

Regards,  Roger

2010\07\14@113659 by Olin Lathrop

face picon face
Roger Weichert wrote:
> I went for 105 degree caps, which limited my choices due to size
> restraints. Maybe I should have been worrying more about ripple
> current.

Caps that can tolerate higher temperature is a good thing to try, but so is
cutting down on the heat produced in the first place.  These kind of caps do
take more to manufacture, but if you have high ripple current it's either
that or the problems you see.  This is why the Panasonic FK series
(specifically designed for low ESR and high ripple current) are physically
bigger and more expensive than those with otherwise the same spec.  There is
no free lunch.  Many of the other series are meant for other applications,
and if you look carefully they don't even have a ripple current spec.  This
is common for "computer grade" caps, for example.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2010\07\14@115233 by Spehro Pefhany

picon face
At 10:27 AM 14/07/2010, you wrote:


>The originals were also only rated at 85 degrees, and I've fitted 105 degree
>replacements, but while excessive temperature can dry them out, I wouldn't
>have thought it would cause this problem.

When the cap starts to go the ESR will increase, which quickly
increases the internal I^2*R heating, and the cap can then vent.

The "X" embossed into the top of the cap deliberately weakens the
aluminum can so it will release pressure relatively safely and in a
relatively controlled fashion when the end comes. Smaller caps
typically vent out the bottom by blowing the rubber bung out.

For long life you want to keep the caps as cool as possible..
considering ambient temperature, external heat sources and internally
generated heat. As a rule of thumb, every 10 K increase in temperature
halves the life. The rated life is usually only 2-3K hours for the
cheaper caps, which is only a few months of 24/7 operation-- not
acceptable in most applications, so derating is virtually always required.

And there's no free lunch, if you can find caps rated at 5,000 hours or
more they will usually be a lot bigger as well as maybe 10-15dB more
expensive.


>Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
EraseMEspeffspam_OUTspamTakeThisOuTinterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com



2010\07\14@130644 by RussellMc

face picon face
What they all said ..

and

even ecaps involve a degree of dark magic!

1. Too LOW a voltage also kills caps quicker.
For wet aluminum caps, 15V on a 35V cap is too low!

2. Putting another comment into a formula
Rule of thumb to be used with care but useful.

                    Lifetime ~~~= Hours_rated x 2^(Trated-Toperating)/10

so REAL 105C caps are well worthwhile.

(2^(105-85)/10 ~~= 4 times longer lifetime.

3000 hour versus 2000 hr caps = +50% longer lifetime.

All very approximate.


3. If it doesn't say Panasonic on it then there's a reasonable chance
that it hasn't got Panasonic in it. Farnell are reputable but bad
things happen. I'd check that. Really.

4. Ripple current was mentioned by several people. This matters much.
Look it up in the spec sheet and see how it relates to application.

5. Failure as seen is not unusual. As noted, cap dries out, then
overheats and boils or shorts and vents.
Full explosion with guts thrown out messily is not unknown.

6. Cooling can help. Even heatsinking caps can help !!! BUT just
maximising air flow helps.

7 Storage at temperature WITHOUT voltage applied is FAR worse for
lifetime than operation at rated voltage !!!
Forming voltages are not present and dry-out is accelerated.

8  Size as noted is some indication of capability although high
quality caps MAY be smaller.

9 As noted - for demanding applications DEMAND name brand with real
specs and traceability. Farnell are real enough that they should care.

10 7 years sounds OK in that sort of application day to day.

11. Not applicable here, but using several smaller caps to make a
given capacitance often can give better net ripple current rating.


 Russell McMahon

2010\07\14@134225 by Spehro Pefhany

picon face
At 01:06 PM 14/07/2010, you wrote:
{Quote hidden}

These are also Panasonic:

www.happytrees.org/main-images/ic-manuf-logos/ic_manuf_logo--Matsushita_Electronics_Corp.jpg
www.happytrees.org/main-images/ic-manuf-logos/ic_manuf_logo--Matsushita_Electronics_Corp-3.gif
.. latter is a bit distorted, usually it looks like this:

http://media.digikey.com/photos/Panasonic%20Photos/TSUP%20SERIES%2025,35.jpg

The stylized 'M' is for Matsushita, of course.

>Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speffspamspam_OUTinterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com



2010\07\14@182721 by Jake Anderson

flavicon
face
On 15/07/10 03:06, RussellMc wrote:
> What they all said ..
>
> and
>
> even ecaps involve a degree of dark magic!
>
> 1. Too LOW a voltage also kills caps quicker.
> For wet aluminum caps, 15V on a 35V cap is too low!
>    
I was under the impression that under rating them were ok, but the caps
"set" on the lower voltage and if you then try to increase the voltage
even while staying within spec they will fail.
> 9 As noted - for demanding applications DEMAND name brand with real
> specs and traceability. Farnell are real enough that they should care.
>    
who are brand names in this field?

2010\07\14@195209 by Roger Weichert

flavicon
face


Spehro Pefhany wrote

.>The rated life is usually only 2-3K hours for the
> cheaper caps, which is only a few months of 24/7 operation-- not
> acceptable in most applications, so derating is virtually always required.

Yes I saw that while I was looking for replacements and worked out at 8
hours a day they should only last about a year, so to get 7 years was
probably pretty good.


> And there's no free lunch, if you can find caps rated at 5,000 hours or
> more they will usually be a lot bigger as well as maybe 10-15dB more
> expensive.


And thats a problem because there is no spare space now. They are crammed
side by side already. Looks like I'd better hope the ones I fitted were a
good choice and last at least as long as the originals. Thanks.

Regards,  Roger

2010\07\14@195806 by Philip Pemberton

face
flavicon
face
On 14/07/10 18:06, RussellMc wrote:
> 3. If it doesn't say Panasonic on it then there's a reasonable chance
> that it hasn't got Panasonic in it. Farnell are reputable but bad
> things happen. I'd check that. Really.

They don't mark them "Panasonic", they use the Matsushita logo instead,
which is a squared "M" in a box. A bit like this:

+--------+
| .    . |
| |\  /| |
| | \/ | |
| +-  -+ |
+--------+

There's an image of the logo here:
  http://capacitor.web.fc2.com/m.html#matsushita

The FC2 capacitor logo database is pretty handy (when combined with a JP
-> EN translator)...

--
Phil.
@spam@piclistKILLspamspamphilpem.me.uk
http://www.philpem.me.uk/

2010\07\14@202943 by Roger Weichert

flavicon
face

Russell wrote

>>3. If it doesn't say Panasonic on it then there's a reasonable chance
>>that it hasn't got Panasonic in it. Farnell are reputable but bad
>>things happen. I'd check that. Really.

I need to apologise to Farnell. The caps do have the Matsushita logo.I guess
I was expecting the name Panasonic to be splashed across them in big bold
letters.

As Spehro has pointed out their different logos below. The second lot of
caps I received had the third logo stamped on them and I dont remember
exactly but I think the first lot I ordered had the first logo. So both were
likely Panasonic caps as advertised.



Spehro Pefhany wrote

{Quote hidden}

I've learned a lot from asking the question. Thanks to everyone who
responded.

My biggest concern was that I thought the caps had failed suddenly, and I
wanted to know why so it didn't happen again.

Now I realise they had probably been slowly dying with an increase in ESR
... but the final stage was the overheating and bursting because of the IR
effects from the ripple current.

I've been fixing stuff for over 30 years and I've changed 1000's and 1000's
of electros and less than 1% would have burst like these  ...  and the ones
that did could usually be put down to an over revving generator or short
circuit supply diodes  ...  so this has been a real eye opener for me.

Thanks to everyone again.

Regards,   Roger

2010\07\14@224310 by John Chung

picon face
Just curious. Did you hook a temp probe on the capacitor and determine
the temperature the caps are subjected to while running the circuit?


John

--- On Wed, 7/14/10, Roger Weichert <KILLspamrweicherKILLspamspambigpond.net.au> wrote:

{Quote hidden}

> -

2010\07\14@230206 by Kenneth Lumia

picon face

----- Original Message -----
From: "Roger Weichert" <TakeThisOuTrweicherEraseMEspamspam_OUTbigpond.net.au>
To: "Microcontroller discussion list - Public."
<RemoveMEpiclistspamTakeThisOuTmit.edu>
Sent: Wednesday, July 14, 2010 7:52 PM
Subject: Re: [EE] Bulging Electrolytics


{Quote hidden}

As you may be aware, this rating is based on maximum
temperature and ripple current.

Two rules used for derating are:
1.  Life expectancy roughly doubles for each 10C drop
in temperature from max temp.
2.  You can calculate lifetime it a little more
accurately if you know the temperatures
     and the ripple currents using the Arrhenius
equation.  I think someone already posted
     the temperature portion of the equation.  For a
good summary of equations see:
     http://www.elna.co.jp/en/capacitor/alumi/catalog/pdf/tecnote_e.pdf

Ken

2010\07\14@233229 by Roger Weichert

flavicon
face

John Chung wrote ..


> Just curious. Did you hook a temp probe on the capacitor and determine
> the temperature the caps are subjected to while running the circuit?


No I didn't, but it's a great idea. That's something I can do when/if I get
the next one in.

The units do run quite hot, in fact there are warnings on the device about
the danger of burns from it !!!

The mains transfomer and rectifier diodes are encapsulated and there is a
huge heatsink on the back in an effort to keep it a bit cooler.

Regards,  Roger

2010\07\14@233908 by Roger Weichert

flavicon
face

Kenneth Lumia wrote

{Quote hidden}

Thanks Ken, that looks like a good place for me to start. I shall read it a
few more times and see what sinks in.

Probably my biggest problem will be getting an accurate substitute for the
load to run the tests. These things run up to about 80 pulsators or so.

Regards,  Roger

2010\07\15@023853 by RussellMc

face picon face
> .>The rated life is usually only 2-3K hours for the
> > cheaper caps, which is only a few months of 24/7 operation-- not
> > acceptable in most applications, so derating is virtually always required.

2000 or 3000 hours is industry standard.
This is at "rated temperature" and about nobody uses or is expected to
use them at that temperature.
Running at some mean level well below rated is the norm.

Even 3000 hours at max rated temperature is special and higher is very special.
Much higher is fraudulent :-)

> Yes I saw that while I was looking for replacements and worked out at 8
> hours a day they should only last about a year, so to get 7 years was
> probably pretty good.

As above.
More an issue (probably) of ripple current.
IF temperature is ramped up by repetitive use then air cooling
(perhaps add a fan) MAY be useful.
But quality caps sounds good enough - do you expect/want/need it to
last another 7 years?

> > And there's no free lunch, if you can find caps rated at 5,000 hours or
> > more they will usually be a lot bigger as well as maybe 10-15dB more
> > expensive.

They may well exist. I've never seen them.


                 Russell

2010\07\15@061222 by Roger Weichert

flavicon
face

RussellMc wrote

{Quote hidden}

Yes, it will be good to get another 7 years. Depends when the dairy farmers
are financial enough to upgrade their system..

If the repair lasts as long as the originals I will be happy.


>
>> > And there's no free lunch, if you can find caps rated at 5,000 hours or
>> > more they will usually be a lot bigger as well as maybe 10-15dB more
>> > expensive.
>
> They may well exist. I've never seen them.
>


I did actually see some that claimed 8000hrs when I was searching, and in my
ignorance thought that was pretty low!!! If I recall, they were physically
much bigger than the originals, so there was no way of fitting them in
anyway.

Thanks for your input.

Regards,  Roger

2010\07\15@065318 by Spehro Pefhany

picon face
At 02:38 AM 7/15/2010, you wrote:

> > > And there's no free lunch, if you can find caps rated at 5,000 hours or
> > > more they will usually be a lot bigger as well as maybe 10-15dB more
> > > expensive.
>
>They may well exist. I've never seen them.

If you design things like relatively high power inverter motor drives
that run from single phase power, large power supplies etc. that are
expected to work reliably 24/7 you need this kind of part.. used to be
called "computer grade" capacitors.

Eg.

http://www.kemet.com/kemet/web/homepage/kechome.nsf/vapubfiles/F3304_ALC10.pdf/$file/F3304_ALC10.pdf
10,000uF/40V
15,000 hours at 85°C 3.6A ripple ,  not expensive
but not much ripple current rating. 35mm dia x 35 mm tall

http://www.cde.com/catalogs/500C.pdf 100,000uF/35V 20A ripple current
5,000 hours at 85°C (could replace several of the
others)  51mm dia x 145mm tall

or 500C103T050AC2B  10,000 uF 50V 35mm diameter x
105mm tall 5000 hours at 85C. 122 in stock
at Digikey for about $20 in 10's. Doesn't seem to be in current datasheet








>                   Russell

2010\07\15@072023 by RussellMc

face picon face
> http://www.kemet.com/kemet/web/homepage/kechome.nsf/vapubfiles/F3304_ALC10.pdf/$file/F3304_ALC10.pdf
> 10,000uF/40V
> 15,000 hours at 85°C 3.6A ripple ,  not expensive
> but not much ripple current rating. 35mm dia x 35 mm tall

About same life as 4000 hour 105C cap.

> http://www.cde.com/catalogs/500C.pdf 100,000uF/35V 20A ripple current
> 5,000 hours at 85°C (could replace several of the
> others)  51mm dia x 145mm tall

About same as 105C 1250 hour cap

> or 500C103T050AC2B  10,000 uF 50V 35mm diameter x
> 105mm tall 5000 hours at 85C. 122 in stock
> at Digikey for about $20 in 10's. Doesn't seem to be in current datasheet

Same.

ie the first seems to have a very real advantage over 2000 hour 105C
cap, the others apparently less so.
It would be interesting to know how an eg 85C 4000 hours and 105C 1000
hour cap compared in practice 9as they are nominally the same in
lifetime performance. Not that you'd usually expect to see a 105C 1000
hour part.

A good look at the Elna sheet referenced suggests that even very
modest fan colling could double lifetimes.



          Russell

2010\07\15@073347 by Spehro Pefhany

picon face
At 07:19 AM 7/15/2010, you wrote:

>A good look at the Elna sheet referenced suggests that even very
>modest fan colling could double lifetimes.

And what's the life of the fan? ;-)

>Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speffEraseMEspam.....interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com



2010\07\15@075800 by Jake Anderson

flavicon
face
On 15/07/10 21:33, Spehro Pefhany wrote:
> At 07:19 AM 7/15/2010, you wrote:
>
>    
>> A good look at the Elna sheet referenced suggests that even very
>> modest fan colling could double lifetimes.
>>      
> And what's the life of the fan? ;-)
>
>    
3000-5000 hours for sleeve 10-50k hours for ball bearing ;->

2010\07\15@081129 by RussellMc

face picon face
> >A good look at the Elna sheet referenced suggests that even very
> >modest fan cooling could double lifetimes.
>
> And what's the life of the fan? ;-)

Good question.
If you have a $10 head, buy a $10 helmet.

No name junk sleeve fan - low.
Quality ball bearing fan with certified MTBF from reputable manufacturer.
50,000 - 250,000 hours at 20 C say
20,000 - 50,000 at 70 C
Or more if you pay more.
One example of many:

             http://www.sepa-europe.com/pdfs/sheets/en/sfb25a05h-mfb25f12.pdf

Tiny. 50000 hr MTBF at 20C
http://www.switchesplus.com.au/products/sepa/ultraminature/sepa.mf10a.php

Rather cute
http://www.switchesplus.com.au/products/sepa/ultraminature/MF10A_doc_revised.pdf

ie you CAN get a fan that suits. And you can get junk.
BUT a fan's operation can be checked occasionally and as the job here
is lifetime extension of caps, fan failure is not a fatal event. If
fan can be replaced easily then doing this rather than replacing caps
may be attractive.
Overall design needs to look at implications of various failures.

Caps every 7 years MAY be entirely acceptable. Or not.


                       Russell

More... (looser matching)
- Last day of these posts
- In 2010 , 2011 only
- Today
- New search...