X-capacitor failure modes?

On 9/22/2013 4:21 PM, Joerg wrote:
> Phil Hobbs wrote:
>> On 9/22/2013 10:36 AM, Joerg wrote:
>>> legg wrote:
>>>> On Sat, 21 Sep 2013 15:17:34 -0700, Joerg <invalid@invalid.invalid>
>>>> wrote:
>>>>
>>>>> Folks,
>>>>>
>>>>> This afternoon I took the old HP-4191A apart. Yesterday in the
>>>>> middle of
>>>>> some measurements it went pop-pop-pop, loud, like firecrackers. I ran
>>>>> over and turned it off, upon which the popping stopped. An immense
>>>>> amount of light-gray smoke wafted out and it had an odd stench to it.
>>>>> The stench lingered for hours.
>>>>>
>>>>> I can't find any source and upon powering it up the analyzer worked
>>>>> fine. That was before I changed anything. As if it had repaired itself.
>>>>> The only thing I could see is a crack in the plastic of the
>>>>> X-capacitor.
>>>>> Snipped the thing out, measured, has almost 0.28uF capacitance (27%
>>>>> more
>>>>> than stated) and no leakage current. There is another X-cap and two
>>>>> Y-caps inside a canned IEC receptacle but those can't be the culprits
>>>>> because that is on the line side of the power switch.
>>>>>
>>>>> Is it possible that an X-cap self-heals to the point where you can
>>>>> barely see a thing, yet release a serious plume of smoke? It is a Rifa
>>>>> GPF-series film cap 0.22uF/250VAC with all kinds of agency logos on
>>>>> there.
>>>>
>>>> Never heard of Rifa GPF.
>>>>
>>>> In fact, if it's Rifa, and manufactured since 1980, it's part type
>>>> will begin with the letter P, even after their amalgamation with Evox.
>>>>
>>>
>>>
>>> Maybe it's PME. The label says this:
>>>
>>> Rifa    GPF
>>> 40/085/56
>>> PME 271 M 622
>>>
>>> Then also "CM1" rotated 90 degrees on the right. No Evox mention on
>>> there.
>>>
>>>
>>>> Metalized paper PM (or PZ), metalized polyester PH.
>>>>
>>>> Perhaps you've mistaken the brand?
>>>>
>>>
>>> No, says Rifa on there and has the logo.
>>>
>>>
>>>> The 4191A was first released in 1980.
>>>>
>>>
>>> The service manual is from 1980 but I believe the machine itself was
>>> built around 1985. It's a good system. Ok, had its quirks. A big
>>> electrolytic in it gave up once. Then they had screwed up the backup
>>> battery charger but since I fixed that it doesn't lose calibration data
>>> anymore. The plastic component clamp levers for the 16092A probe head
>>> are wimpy and, predictably, one snapped in half. Still usable but one
>>> has to have strong fingernails. Other than that it has served me well.
>>>
>>>
>>>> Interesting thing about the older Rifa parts, was that their outer
>>>> layer was transparent - an overstress or corona can often be seen, as
>>>> well as heard, but usually only in locations that are subject to the
>>>> attentions of overly zealous test technicians with manually adjustible
>>>> hipot equipment.
>>>>
>>>
>>> This cap developed a crack across one of the flat sides. Considering the
>>> loud pops and the amount of smoke I'd have expected a total meltdown but
>>> other than the crack I can't see any damage.
>>>
>>
>> I had a very nice hand-wired Krohn-Hite filter croak itself a year or so
>> back--the clamp on one of the 'computer-grade' filter caps was loose,
>> and the cap worked its way down through the chassis till it shorted to
>> the bottom of the case. (No fish paper there, for some silly reason.)
>>
>> This resulted in clouds of transformer smoke filling the lab, which was
>> a shame, because it was a really nice box and the transformer was an
>> irreplaceable custom job.  It even had completely separate power
>> supplies for the two channels, to prevent crosstalk.
>>
>> Good thing I'm nuts about turning all the lab power off when I'm not
>> there--it could have happened any time at all.
>>
>
> Still, watch those X-caps in them boat anchors where you now have so
> many of. I have seen these golden-clear plastic caps in many of them. My
> lab has pretty good line spike filtering and the cap ran at less than
> 50% of its rated voltage yet it decided to blow. Problem is, things such
> as this impedance analyzer can be unattended for long periods of time
> when you run an experiment. Same with a spectrum analyzer when it runs a
> baseline scan over a 100MHz swath with 1Hz BW. You don't want to come
> back from lunch only to see trucks with flashing blue lights surrounding
> your building because a fire alarm went off.
>

I turn it off at the power bar for that sort of reason.   Means that the 
ovenized XOs take awhile to warm up, but oh well.

Cheers

Phil Hobbs

-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510 USA
+1 845 480 2058

hobbs at electrooptical dot net
http://electrooptical.net

On 22/09/2013 06:42, Jon Elson wrote:
> Joerg wrote:
>
>> Folks,
>>
>> This afternoon I took the old HP-4191A apart. Yesterday in the middle of
>> some measurements it went pop-pop-pop, loud, like firecrackers. I ran
>> over and turned it off, upon which the popping stopped. An immense
>> amount of light-gray smoke wafted out and it had an odd stench to it.
>> The stench lingered for hours.
>>
>> I can't find any source and upon powering it up the analyzer worked
>> fine. That was before I changed anything. As if it had repaired itself.
>> The only thing I could see is a crack in the plastic of the X-capacitor.
>> Snipped the thing out, measured, has almost 0.28uF capacitance (27% more
>> than stated) and no leakage current. There is another X-cap and two
>> Y-caps inside a canned IEC receptacle but those can't be the culprits
>> because that is on the line side of the power switch.
>>
>> Is it possible that an X-cap self-heals to the point where you can
>> barely see a thing, yet release a serious plume of smoke? It is a Rifa
>> GPF-series film cap 0.22uF/250VAC with all kinds of agency logos on there.
>>
> Yes, definitely.  But, it is a bit surprising that it still has above
> rated capacitance.  The crack in the plastic is pretty close to a
> "smoking gun".  I have seen this before, at least once.

X rated capacitors are generally metallised film, such that where the 
film breaks down the metal is vaporised destroying and preventing any 
further current path.  As a result the capacitance will generally 
degrade over time, so find the apparent 0.28uF a little hard to explain.

-- 
Mike Perkins
Video Solutions Ltd
www.videosolutions.ltd.uk

X class caps have improved a lot with fusible links built into thin metal foil pattern so that less heat damage to vaporizing toxic plastic with only fractional loss of capacitance with many fuse links in the entire rolled metalized layer.

Check out Panasonic's design of X class PU and PE caps.

I don't know the MSDS of your part, but smoke from molten PVC wire insulation caused more lethal risk than the radiation at Three Mile Island.

Phil Hobbs wrote:
> On 9/22/2013 4:21 PM, Joerg wrote:
>> Phil Hobbs wrote:
>>> On 9/22/2013 10:36 AM, Joerg wrote:
>>>> legg wrote:

[...]


>>>>> Interesting thing about the older Rifa parts, was that their outer
>>>>> layer was transparent - an overstress or corona can often be seen, as
>>>>> well as heard, but usually only in locations that are subject to the
>>>>> attentions of overly zealous test technicians with manually adjustible
>>>>> hipot equipment.
>>>>>
>>>>
>>>> This cap developed a crack across one of the flat sides. Considering
>>>> the
>>>> loud pops and the amount of smoke I'd have expected a total meltdown
>>>> but
>>>> other than the crack I can't see any damage.
>>>>
>>>
>>> I had a very nice hand-wired Krohn-Hite filter croak itself a year or so
>>> back--the clamp on one of the 'computer-grade' filter caps was loose,
>>> and the cap worked its way down through the chassis till it shorted to
>>> the bottom of the case. (No fish paper there, for some silly reason.)
>>>
>>> This resulted in clouds of transformer smoke filling the lab, which was
>>> a shame, because it was a really nice box and the transformer was an
>>> irreplaceable custom job.  It even had completely separate power
>>> supplies for the two channels, to prevent crosstalk.
>>>
>>> Good thing I'm nuts about turning all the lab power off when I'm not
>>> there--it could have happened any time at all.
>>>
>>
>> Still, watch those X-caps in them boat anchors where you now have so
>> many of. I have seen these golden-clear plastic caps in many of them. My
>> lab has pretty good line spike filtering and the cap ran at less than
>> 50% of its rated voltage yet it decided to blow. Problem is, things such
>> as this impedance analyzer can be unattended for long periods of time
>> when you run an experiment. Same with a spectrum analyzer when it runs a
>> baseline scan over a 100MHz swath with 1Hz BW. You don't want to come
>> back from lunch only to see trucks with flashing blue lights surrounding
>> your building because a fire alarm went off.
>>
> 
> I turn it off at the power bar for that sort of reason.   Means that the
> ovenized XOs take awhile to warm up, but oh well.
> 

Doesn't always work. Another incident I had was a 6h endurance test of a
new design for a customer. I sat in the office next door doing something
else. Then ... sniff .. sniff ... what on earth ... ran into the lab and
sure enough a transistor in a linear bench supply had shorted. Instead
of a regulated 9V my prototype now got close to 40V at several amps. It
was not a pretty sight and took out several other pieces of equipment.

-- 
Regards, Joerg

http://www.analogconsultants.com/

Mike Perkins wrote:
> On 22/09/2013 06:42, Jon Elson wrote:
>> Joerg wrote:
>>
>>> Folks,
>>>
>>> This afternoon I took the old HP-4191A apart. Yesterday in the middle of
>>> some measurements it went pop-pop-pop, loud, like firecrackers. I ran
>>> over and turned it off, upon which the popping stopped. An immense
>>> amount of light-gray smoke wafted out and it had an odd stench to it.
>>> The stench lingered for hours.
>>>
>>> I can't find any source and upon powering it up the analyzer worked
>>> fine. That was before I changed anything. As if it had repaired itself.
>>> The only thing I could see is a crack in the plastic of the X-capacitor.
>>> Snipped the thing out, measured, has almost 0.28uF capacitance (27% more
>>> than stated) and no leakage current. There is another X-cap and two
>>> Y-caps inside a canned IEC receptacle but those can't be the culprits
>>> because that is on the line side of the power switch.
>>>
>>> Is it possible that an X-cap self-heals to the point where you can
>>> barely see a thing, yet release a serious plume of smoke? It is a Rifa
>>> GPF-series film cap 0.22uF/250VAC with all kinds of agency logos on
>>> there.
>>>
>> Yes, definitely.  But, it is a bit surprising that it still has above
>> rated capacitance.  The crack in the plastic is pretty close to a
>> "smoking gun".  I have seen this before, at least once.
> 
> X rated capacitors are generally metallised film, such that where the
> film breaks down the metal is vaporised destroying and preventing any
> further current path.  As a result the capacitance will generally
> degrade over time, so find the apparent 0.28uF a little hard to explain.
> 

That 0.28uF is the main reason I posted my question. I expected half or
more of the capacitance to be gone up in smoke, literally.

-- 
Regards, Joerg

http://www.analogconsultants.com/

On Sun, 22 Sep 2013 16:58:46 -0700, Joerg <invalid@invalid.invalid>
wrote:

>Mike Perkins wrote:
>> On 22/09/2013 06:42, Jon Elson wrote:
>>> Joerg wrote:
>>>
>>>> Folks,
>>>>
>>>> This afternoon I took the old HP-4191A apart. Yesterday in the middle of
>>>> some measurements it went pop-pop-pop, loud, like firecrackers. I ran
>>>> over and turned it off, upon which the popping stopped. An immense
>>>> amount of light-gray smoke wafted out and it had an odd stench to it.
>>>> The stench lingered for hours.
>>>>
>>>> I can't find any source and upon powering it up the analyzer worked
>>>> fine. That was before I changed anything. As if it had repaired itself.
>>>> The only thing I could see is a crack in the plastic of the X-capacitor.
>>>> Snipped the thing out, measured, has almost 0.28uF capacitance (27% more
>>>> than stated) and no leakage current. There is another X-cap and two
>>>> Y-caps inside a canned IEC receptacle but those can't be the culprits
>>>> because that is on the line side of the power switch.
>>>>
>>>> Is it possible that an X-cap self-heals to the point where you can
>>>> barely see a thing, yet release a serious plume of smoke? It is a Rifa
>>>> GPF-series film cap 0.22uF/250VAC with all kinds of agency logos on
>>>> there.
>>>>
>>> Yes, definitely.  But, it is a bit surprising that it still has above
>>> rated capacitance.  The crack in the plastic is pretty close to a
>>> "smoking gun".  I have seen this before, at least once.
>> 
>> X rated capacitors are generally metallised film, such that where the
>> film breaks down the metal is vaporised destroying and preventing any
>> further current path.  As a result the capacitance will generally
>> degrade over time, so find the apparent 0.28uF a little hard to explain.
>> 
>
>That 0.28uF is the main reason I posted my question. I expected half or
>more of the capacitance to be gone up in smoke, literally.

---
It could be that the pressure exerted on the film before the case
cracked was enough to squeeze/compress it to the point where the
thinner dielectric caused the capacitance increase.


-- 
JF

"Mike Perkins"
>
> X rated capacitors are generally metallised film, such that where the film 
> breaks down the metal is vaporised destroying and preventing any further 
> current path.  As a result the capacitance will generally degrade over 
> time, so find the apparent 0.28uF a little hard to explain.


** Class X caps are specially wound so as to not suffer from internal corona 
discharge at 250VAC  -  the technique is called "two in series" which 
creates a floating electrode inside the cap that sits at half the applied 
voltage.

Corona is caused by tiny trapped air pocket inside the cap during winding, 
only avoided by winding in vacuo. Discharge initiates at about 150VAC ( 
depending on the air pressure) - so having series pair solves the problem.

A damaged X cap might lose one of the pair and so the capacitance goes up.


...   Phil 

On Mon, 23 Sep 2013 10:34:34 +1000, the renowned "Phil Allison"
<phil_a@tpg.com.au> wrote:

>
>"Mike Perkins"
>>
>> X rated capacitors are generally metallised film, such that where the film 
>> breaks down the metal is vaporised destroying and preventing any further 
>> current path.  As a result the capacitance will generally degrade over 
>> time, so find the apparent 0.28uF a little hard to explain.
>
>
>** Class X caps are specially wound so as to not suffer from internal corona 
>discharge at 250VAC  -  the technique is called "two in series" which 
>creates a floating electrode inside the cap that sits at half the applied 
>voltage.
>
>Corona is caused by tiny trapped air pocket inside the cap during winding, 
>only avoided by winding in vacuo. Discharge initiates at about 150VAC ( 
>depending on the air pressure) - so having series pair solves the problem.
>
>A damaged X cap might lose one of the pair and so the capacitance goes up.
>
>
>...   Phil 
>

Interesting information, Phil! 



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

Joerg <invalid@invalid.invalid> wrote:
> 
> Doesn't always work. Another incident I had was a 6h endurance test of
> a new design for a customer. I sat in the office next door doing
> something else. Then ... sniff .. sniff ... what on earth ... ran into
> the lab and sure enough a transistor in a linear bench supply had
> shorted. Instead of a regulated 9V my prototype now got close to 40V
> at several amps. It was not a pretty sight and took out several other
> pieces of equipment. 

I just lost a brand new motherboard to a defective power supply. I'm 
thinking of putting 50 amp SCR's across all the critical voltages for the 
pc and lab supplies. Trigger the SCR with a TL431 set to 10% above nominal.  
Real cheap - may only cost a few pennies. Better to blow a fuse than lose 
an important piece of equipment.

JK

"Spehro Pefhany"
 >"Phil Allison"
>
>>"Mike Perkins"
>>>
>>> X rated capacitors are generally metallised film, such that where the 
>>> film
>>> breaks down the metal is vaporised destroying and preventing any further
>>> current path.  As a result the capacitance will generally degrade over
>>> time, so find the apparent 0.28uF a little hard to explain.
>>
>>
>>** Class X caps are specially wound so as to not suffer from internal 
>>corona
>>discharge at 250VAC  -  the technique is called "two in series" which
>>creates a floating electrode inside the cap that sits at half the applied
>>voltage.
>>
>>Corona is caused by tiny trapped air pocket inside the cap during winding,
>>only avoided by winding in vacuo. Discharge initiates at about 150VAC (
>>depending on the air pressure) - so having series pair solves the problem.
>>
>>A damaged X cap might lose one of the pair and so the capacitance goes up.
>>
>
>
> Interesting information, Phil!

 ** You mean you did not already know that ?

FYI:

There are also " three in series " X caps for higher voltages like 440VAC.

It is possible to vacuum impregnate some types of X caps ( eg film, foil, 
paper ) and eliminate all trapped air.

But few cap makers will attempt winding caps in vacuo.

Staff hate wearing those clumsy space suits ...



....  Phil