X2 AC capacitor failure

[Pimpom]

I devised the circuit below for use as a continuous mains
presence/absence indicator on my front porch.

0.1uF
|| LED
3.3k |----||---| |--->|----|
___ | || | | |
------|___|----| |---| |---
| ___ | | LED | |
---|___|--- ----|<----- |
230VAC 1M |
|
-------------------------------------------|

C = 275VAC X2, LEDs = Blue 5mm

It worked well for some time (a few months maybe) but the LEDs
then grew dimmer and dimmer until they were barely visible in
daytime. Turned out that the capacitor had greatly decreased in
value - down below 1nF.

I replaced the cap and the LEDs glowed brightly once more - for a
few weeks, and then began to dim again. The replacement cap has
already gone down to about a tenth of its nominal value.

These caps were not quite new. I pulled ten of them from another
project after testing it for a day. Each cap initially served as
a snubber in series with a 100-ohm resistor across a triac. But
the load for each triac was light (<5W) and essentially
non-inductive, so I took the caps out after the test, during
which they were switched on and off some dozens of times.

The test was run in a place where there's no public power supply
and I used a square wave UPS. Could that have weakened the caps
during the test? But then why would the effect show up only after
they had been subjected to normal sine wave AC for some months?

 

[Phil Allison]

"Pimpom"

I devised the circuit below for use as a continuous mains presence/absence
indicator on my front porch.

** Those last two words are key here.

Turned out that the capacitor had greatly decreased in value - down below
1nF.

** Then these.

How do the leads on the failed X caps look ??

Nice and shiny or not ?



..... Phil

 

[Maynard A. Philbrook Jr.]

In article <lrklm9$var$1@news.albasani.net>, Pimpom@invalid.invalid
says...

I devised the circuit below for use as a continuous mains
presence/absence indicator on my front porch.

0.1uF
|| LED
3.3k |----||---| |--->|----|
___ | || | | |
------|___|----| |---| |---
| ___ | | LED | |
---|___|--- ----|<----- |
230VAC 1M |
|
-------------------------------------------|

C = 275VAC X2, LEDs = Blue 5mm

It worked well for some time (a few months maybe) but the LEDs
then grew dimmer and dimmer until they were barely visible in
daytime. Turned out that the capacitor had greatly decreased in
value - down below 1nF.

I replaced the cap and the LEDs glowed brightly once more - for a
few weeks, and then began to dim again. The replacement cap has
already gone down to about a tenth of its nominal value.

These caps were not quite new. I pulled ten of them from another
project after testing it for a day. Each cap initially served as
a snubber in series with a 100-ohm resistor across a triac. But
the load for each triac was light (<5W) and essentially
non-inductive, so I took the caps out after the test, during
which they were switched on and off some dozens of times.

The test was run in a place where there's no public power supply
and I used a square wave UPS. Could that have weakened the caps
during the test? But then why would the effect show up only after
they had been subjected to normal sine wave AC for some months?

Modified sine wave UPS/inverters do play havoc on things like that
however, normally if you have other things on the line the wave isn't so
bad but if that is all you have operating at the time I really don't
think the X caps were intended for constant pulsing.

from my understanding of them, X caps are design to slowly degrade
by self healing when excessive pulses are seen but all they do is kind
of chip away at the metalized foil to remove the short. This makes the
cap weaker over time until it no longer works.

Difference manufactures may do their own thing but that is how I
understand it.

Have you thought about using a passive xformer or Common mode choke?
Maybe even a Y cap going to common before the X cap may capture the
fast raise.

Or you could try and use a mica cap which I am sure will handle the
pounding.

Jamie

 

[John Fields]

On Sun, 3 Aug 2014 21:08:04 +1000, "Phil Allison"
<phil_a@tpg.com.au> wrote:

"Pimpom"

I devised the circuit below for use as a continuous mains presence/absence
indicator on my front porch.

** Those last two words are key here.

Turned out that the capacitor had greatly decreased in value - down below
1nF.

** Then these.

How do the leads on the failed X caps look ??

Nice and shiny or not ?



.... Phil

---
Good one!!!

John Fields

 

[Pimpom]

Phil Allison wrote:

"Pimpom"

I devised the circuit below for use as a continuous mains
presence/absence indicator on my front porch.

** Those last two words are key here.

Turned out that the capacitor had greatly decreased in value -
down
below 1nF.

** Then these.

How do the leads on the failed X caps look ??

Nice and shiny or not ?

Do I understand correctly that you suspect environmental effects?
Moisture seeping in around the leads? I can't dismiss that
offhand. And the leads do look like they're lightly oxidised. But
then that's quite common here as this is a tropical area, even in
gadgets and spare parts kept indoors.

The indicator is on the porch, but there's a five-foot overhang
above and beyond it. I also have a cheap Chinese 4-ft flourescent
tube light much closer to the edge of the overhang and it's kept
on throughout the night. Neither the tube nor the simple
electronic choke is covered and are often subjected to spray and
fog, but they're quite durable, the current set being at least
five years old. The tube just needs to be rotated in the socket
perhaps once a year. The choke doesn't use an X2 capacitor but I
would have expected some deleterious effect from the weather if
that's what's causing the caps in the indicator to fail so
quickly.

But if you're saying that X2 capacitors are particularly prone to
such degradation, then I'll accept your experienced judgment.

Roof edge
--------------------------------------------
============
Tube light
--------------------------
\
\----------------
Verandah |
|
Mains indicator-> X|
| House
|
|
____________________________|

House

 

[Pimpom]

Maynard A. Philbrook Jr. wrote:

In article <lrklm9$var$1@news.albasani.net>,
Pimpom@invalid.invalid
says...

I devised the circuit below for use as a continuous mains
presence/absence indicator on my front porch.

0.1uF
|| LED
3.3k |----||---| |--->|----|
___ | || | | |
------|___|----| |---| |---
| ___ | | LED | |
---|___|--- ----|<----- |
230VAC 1M |
|
-------------------------------------------|

C = 275VAC X2, LEDs = Blue 5mm

It worked well for some time (a few months maybe) but the LEDs
then grew dimmer and dimmer until they were barely visible in
daytime. Turned out that the capacitor had greatly decreased
in
value - down below 1nF.

I replaced the cap and the LEDs glowed brightly once more -
for a
few weeks, and then began to dim again. The replacement cap
has
already gone down to about a tenth of its nominal value.

These caps were not quite new. I pulled ten of them from
another
project after testing it for a day. Each cap initially served
as
a snubber in series with a 100-ohm resistor across a triac.
But
the load for each triac was light (<5W) and essentially
non-inductive, so I took the caps out after the test, during
which they were switched on and off some dozens of times.

The test was run in a place where there's no public power
supply
and I used a square wave UPS. Could that have weakened the
caps
during the test? But then why would the effect show up only
after
they had been subjected to normal sine wave AC for some
months?

Modified sine wave UPS/inverters do play havoc on things like
that
however, normally if you have other things on the line the wave
isn't
so bad but if that is all you have operating at the time I
really
don't think the X caps were intended for constant pulsing.

from my understanding of them, X caps are design to slowly
degrade
by self healing when excessive pulses are seen but all they do
is kind
of chip away at the metalized foil to remove the short. This
makes the
cap weaker over time until it no longer works.

Difference manufactures may do their own thing but that is how
I
understand it.

That would seem to be a logical explanation. I've tested the rest
of the caps that were removed from the other project and they
measure around 0.07-0.08 uF instead of the nominal 0.1uF. But why
did they continue to degrade while being used in a simple sine
wave circuit? As you can see, the present application even has a
3.3k resistor in series to limit the worst-case inrush current.
I'd love to understand the mechanism that continues to cause more
and more of the metal film to vapourize.

The whole circuit is fitted into a standard 2-pin AC plug which
is plugged into a 230V outlet on my verandah. If my house AC
supply were repeatedly subjected to fast rising spikes and
surges, I'm sure they would have had a noticeable effect on at
least some of the other gadgets around the house.

Have you thought about using a passive xformer or Common mode
choke?
Maybe even a Y cap going to common before the X cap may capture
the
fast raise.

Or you could try and use a mica cap which I am sure will handle
the
pounding.

Well, I'm not using these caps as mains filters. And they're no
longer run from a UPS/inverter. That was only while testing the
other - and much more complex - project for a few hours.

 

[Maynard A. Philbrook Jr.]

In article <lrm1ok$ubg$1@news.albasani.net>, Pimpom@invalid.invalid
says...

degrade
by self healing when excessive pulses are seen but all they do
is kind
of chip away at the metalized foil to remove the short. This
makes the
cap weaker over time until it no longer works.

Difference manufactures may do their own thing but that is how
I
understand it.

That would seem to be a logical explanation. I've tested the rest
of the caps that were removed from the other project and they
measure around 0.07-0.08 uF instead of the nominal 0.1uF. But why
did they continue to degrade while being used in a simple sine
wave circuit? As you can see, the present application even has a
3.3k resistor in series to limit the worst-case inrush current.
I'd love to understand the mechanism that continues to cause more
and more of the metal film to vapourize.

If you can put a o-scope across the capacitor maybe you can see some
noise that should be there. Like high freq ripples caused from
generators and things like switching devices on the same line.

Better yet, if you have low level clamp on current probe for a scrope
you can see this happening.

Jamie

 

[Phil Allison]

"Pimpom"

Phil Allison wrote:
"Pimpom"

I devised the circuit below for use as a continuous mains
presence/absence indicator on my front porch.

** Those last two words are key here.

Turned out that the capacitor had greatly decreased in value - down
below 1nF.

** Then these.

How do the leads on the failed X caps look ??

Nice and shiny or not ?

Do I understand correctly that you suspect environmental effects?

** What was your first clue?

> Moisture seeping in around the leads?

** Causing corrosion where the lead is crimped to the metallised film.

Had a whole bunch of 100nF and 220nF 400V caps go open from the same
ause - and they were subjected to only low DC voltage and kept indoors or
inside equipment all the time.


I can't dismiss that
> offhand. And the leads do look like they're lightly oxidised.

** That is just how the ones above looked too.

You have a 3.3kohm resistor in series with your X caps so spike voltages on
the AC supply are not causing any damage.

By their "two in one" design, class X1 and X2 caps are immune from internal
corona so lead corrosion is the only possibility left.


..... Phil

 

[Pimpom]

Maynard A. Philbrook Jr. wrote:

In article <lrm1ok$ubg$1@news.albasani.net>,
Pimpom@invalid.invalid
says...
degrade
by self healing when excessive pulses are seen but all they
do
is kind
of chip away at the metalized foil to remove the short. This
makes the
cap weaker over time until it no longer works.

Difference manufactures may do their own thing but that is
how
I
understand it.

That would seem to be a logical explanation. I've tested the
rest
of the caps that were removed from the other project and they
measure around 0.07-0.08 uF instead of the nominal 0.1uF. But
why
did they continue to degrade while being used in a simple sine
wave circuit? As you can see, the present application even has
a
3.3k resistor in series to limit the worst-case inrush
current.
I'd love to understand the mechanism that continues to cause
more
and more of the metal film to vapourize.

If you can put a o-scope across the capacitor maybe you can see
some
noise that should be there. Like high freq ripples caused from
generators and things like switching devices on the same line.

Better yet, if you have low level clamp on current probe for a
scrope
you can see this happening.

I dug up a batch of caps that had not previously been used in a
square-wave circuit, all weighing in at about 0.11 uF and
replaced the failing one today. That should settle the question
once and for all. Now I just have to wait for a few weeks.

 

[Cydrome Leader]

Pimpom <Pimpom@invalid.invalid> wrote:

Maynard A. Philbrook Jr. wrote:
In article <lrklm9$var$1@news.albasani.net>,
Pimpom@invalid.invalid
says...

I devised the circuit below for use as a continuous mains
presence/absence indicator on my front porch.

0.1uF
|| LED
3.3k |----||---| |--->|----|
___ | || | | |
------|___|----| |---| |---
| ___ | | LED | |
---|___|--- ----|<----- |
230VAC 1M |
|
-------------------------------------------|

C = 275VAC X2, LEDs = Blue 5mm

It worked well for some time (a few months maybe) but the LEDs
then grew dimmer and dimmer until they were barely visible in
daytime. Turned out that the capacitor had greatly decreased
in
value - down below 1nF.

I replaced the cap and the LEDs glowed brightly once more -
for a
few weeks, and then began to dim again. The replacement cap
has
already gone down to about a tenth of its nominal value.

These caps were not quite new. I pulled ten of them from
another
project after testing it for a day. Each cap initially served
as
a snubber in series with a 100-ohm resistor across a triac.
But
the load for each triac was light (<5W) and essentially
non-inductive, so I took the caps out after the test, during
which they were switched on and off some dozens of times.

The test was run in a place where there's no public power
supply
and I used a square wave UPS. Could that have weakened the
caps
during the test? But then why would the effect show up only
after
they had been subjected to normal sine wave AC for some
months?

Modified sine wave UPS/inverters do play havoc on things like
that
however, normally if you have other things on the line the wave
isn't
so bad but if that is all you have operating at the time I
really
don't think the X caps were intended for constant pulsing.

from my understanding of them, X caps are design to slowly
degrade
by self healing when excessive pulses are seen but all they do
is kind
of chip away at the metalized foil to remove the short. This
makes the
cap weaker over time until it no longer works.

Difference manufactures may do their own thing but that is how
I
understand it.

That would seem to be a logical explanation. I've tested the rest
of the caps that were removed from the other project and they
measure around 0.07-0.08 uF instead of the nominal 0.1uF. But why
did they continue to degrade while being used in a simple sine
wave circuit? As you can see, the present application even has a

A square wave inverer does not output a simple sine wave. It outputs
garbage+noise.

3.3k resistor in series to limit the worst-case inrush current.
I'd love to understand the mechanism that continues to cause more
and more of the metal film to vapourize.

dv/dt. I am a bit surprised line rated caps would degrade so fast though.
They are mass produced items so they are "just good enough" though, at
least for plain sine wave use.

The whole circuit is fitted into a standard 2-pin AC plug which
is plugged into a 230V outlet on my verandah. If my house AC
supply were repeatedly subjected to fast rising spikes and
surges, I'm sure they would have had a noticeable effect on at
least some of the other gadgets around the house.

there aren't too many items that just use a capacitor to drop voltage. The
only thing I have that does that that I can think off off hand is a P3
Kill-a-watt meter. See if you can blow one those up somehow.

Inductive loads do weird stuff with square waves, sometimes. Switching
power supplies don't seem to care, but they tend to have lots of filering
on the input anyways. Resisitve loads should not givea hoot as long as the
RMS voltage you feed them is correct.

Have you thought about using a passive xformer or Common mode
choke?
Maybe even a Y cap going to common before the X cap may capture
the
fast raise.

Or you could try and use a mica cap which I am sure will handle
the
pounding.

Mica is good for abusive conditions. It's still interesting the other caps
are getting destroyed so fast.

Well, I'm not using these caps as mains filters. And they're no
longer run from a UPS/inverter. That was only while testing the
other - and much more complex - project for a few hours.

a check with a scope on the porch might be good to see if you're getting
some crazy voltages on your caps from the inductance of your wiring.

 

[Phil Allison]

"Cydrome Leader"

3.3k resistor in series to limit the worst-case inrush current.
I'd love to understand the mechanism that continues to cause more
and more of the metal film to vaporize.

dv/dt.

** Huh??

Metallisation will only vaporise if there is an arc inside the cap and that
series resistor really slugs the peak dv/dt compared to normal use.

I am a bit surprised line rated caps would degrade so fast though.
They are mass produced items so they are "just good enough" though, at
least for plain sine wave use.

** Hang on a mo, class X1/2 caps are designed to be wired continuously
across the AC supply with all its hazards.

The vast majority are polyester or polypropylene film types with a special
wind that makes two in series with a floating electrode in the middle. They
are tested at over 2kV DC and must not arc.

there aren't too many items that just use a capacitor to drop voltage. The
only thing I have that does that that I can think off off hand is a P3
Kill-a-watt meter.

** How about nearly every AC powered smoke alarm ?



..... Phil

 

[Pimpom]

Cydrome Leader wrote:

Pimpom <Pimpom@invalid.invalid> wrote:

That would seem to be a logical explanation. I've tested the
rest
of the caps that were removed from the other project and they
measure around 0.07-0.08 uF instead of the nominal 0.1uF. But
why
did they continue to degrade while being used in a simple sine
wave circuit? As you can see, the present application even has
a

A square wave inverer does not output a simple sine wave. It
outputs
garbage+noise.

I meant that they continued to degrade while being used with
normal domestic AC supply - after they were used for a trial run
with square wave in that other project.

The whole circuit is fitted into a standard 2-pin AC plug
which
is plugged into a 230V outlet on my verandah. If my house AC
supply were repeatedly subjected to fast rising spikes and
surges, I'm sure they would have had a noticeable effect on at
least some of the other gadgets around the house.

there aren't too many items that just use a capacitor to drop
voltage.

I don't mean just those with a series capacitor input. Surely a
line that dirty would also affect other loads.

Well, I'm not using these caps as mains filters. And they're
no
longer run from a UPS/inverter. That was only while testing
the
other - and much more complex - project for a few hours.

a check with a scope on the porch might be good to see if
you're
getting some crazy voltages on your caps from the inductance of
your
wiring.

Will do that when I have time.

 

[Cydrome Leader]

Phil Allison <phil_a@tpg.com.au> wrote:

"Cydrome Leader"

3.3k resistor in series to limit the worst-case inrush current.
I'd love to understand the mechanism that continues to cause more
and more of the metal film to vaporize.

dv/dt.

** Huh??

Metallisation will only vaporise if there is an arc inside the cap and that
series resistor really slugs the peak dv/dt compared to normal use.

well, the RC time constant with the 3.3k is 330uS and depending on how
square that inverter output is, it might be a problem.

I am a bit surprised line rated caps would degrade so fast though.
They are mass produced items so they are "just good enough" though, at
least for plain sine wave use.

** Hang on a mo, class X1/2 caps are designed to be wired continuously
across the AC supply with all its hazards.

That is a standard AC supply, not a cheapo inverter.

The vast majority are polyester or polypropylene film types with a special
wind that makes two in series with a floating electrode in the middle. They
are tested at over 2kV DC and must not arc.


there aren't too many items that just use a capacitor to drop voltage. The
only thing I have that does that that I can think off off hand is a P3
Kill-a-watt meter.

** How about nearly every AC powered smoke alarm ?

Yeah, but again, it's not standard to run a building off square waves.

It's known that these caps perform as expected with normal sine waves from
utility power. The uknown factor here is the quality of power being fed to
the LED and possibly environmental issues as you or somebody else
mentioned.

 

[Cydrome Leader]

Pimpom <Pimpom@invalid.invalid> wrote:

Cydrome Leader wrote:
Phil Allison <phil_a@tpg.com.au> wrote:

"Cydrome Leader"

Metallisation will only vaporise if there is an arc inside the
cap
and that series resistor really slugs the peak dv/dt compared
to
normal use.

well, the RC time constant with the 3.3k is 330uS and depending
on how
square that inverter output is, it might be a problem.

--------[snip]-----------

** Hang on a mo, class X1/2 caps are designed to be wired
continuously across the AC supply with all its hazards.

That is a standard AC supply, not a cheapo inverter.

--------[snip]-----------

Yeah, but again, it's not standard to run a building off square
waves.

It's known that these caps perform as expected with normal sine
waves
from utility power. The uknown factor here is the quality of
power
being fed to the LED and possibly environmental issues as you
or
somebody else mentioned.

You seem to have missed the part where I said that the caps
continued to degrade while being run from normal sine wave
utility power. They were run from square wave _only_ for a single
test run (albeit with many on/off cycles) while installed in
_another_ product. The caps were then removed and used in the
present circuit with the LEDs, 3.3k resistor, etc.

I did miss that. Once they're worn out from whatever destroyed them in the
first place, they're still at end of life, so I don't see a reason they'd
be OK again. Abuse can't be undone.

The caps served a different purpose in the _other_ circuit,
namely as snubbers in series with a 100-ohm resistor for each
capacitor. I removed them because they were not needed after all.

I'm lost now as to what was going on at this point. It still stands, that
once you destroy a component, it's done and the failure rate will be
pretty high, or the specs will no longer meet the original requirements.

 

[Pimpom]

Cydrome Leader wrote:

Phil Allison <phil_a@tpg.com.au> wrote:

"Cydrome Leader"

Metallisation will only vaporise if there is an arc inside the
cap
and that series resistor really slugs the peak dv/dt compared
to
normal use.

well, the RC time constant with the 3.3k is 330uS and depending
on how
square that inverter output is, it might be a problem.

--------[snip]-----------

** Hang on a mo, class X1/2 caps are designed to be wired
continuously across the AC supply with all its hazards.

That is a standard AC supply, not a cheapo inverter.

--------[snip]-----------

Yeah, but again, it's not standard to run a building off square
waves.

It's known that these caps perform as expected with normal sine
waves
from utility power. The uknown factor here is the quality of
power
being fed to the LED and possibly environmental issues as you
or
somebody else mentioned.

You seem to have missed the part where I said that the caps
continued to degrade while being run from normal sine wave
utility power. They were run from square wave _only_ for a single
test run (albeit with many on/off cycles) while installed in
_another_ product. The caps were then removed and used in the
present circuit with the LEDs, 3.3k resistor, etc.

The caps served a different purpose in the _other_ circuit,
namely as snubbers in series with a 100-ohm resistor for each
capacitor. I removed them because they were not needed after all.

 

[Phil Allison]

"Cydrome Leader"

Yeah, but again, it's not standard to run a building off square waves.

It's known that these caps perform as expected with normal sine waves from
utility power. The uknown factor here is the quality of power being fed to
the LED and possibly environmental issues as you or somebody else
mentioned.

** There is no inverter - the LEDs are being powered from 230V AC on the
OP's porch.



..... Phil

 

[Cydrome Leader]

Phil Allison <phil_a@tpg.com.au> wrote:

"Cydrome Leader"


Yeah, but again, it's not standard to run a building off square waves.

It's known that these caps perform as expected with normal sine waves from
utility power. The uknown factor here is the quality of power being fed to
the LED and possibly environmental issues as you or somebody else
mentioned.


** There is no inverter - the LEDs are being powered from 230V AC on the
OP's porch.

they apparently aren't doing a damn thing now.