overcharging a ceramic cap

[John Larkin]

Supposed we had a hi-K ceramic cap, say 10 uF at 10 volts, and we
charge it from a smallish, say 10 uA, current source. I know that at
some voltage past 10, it will start to conduct and limit terminal
voltage, sort of like a very sloppy zener, so it won't punch through
like, say, a film cap would. But is there a longterm damage mode?

Oh, and of I use a jfet as a super-low leakage diode, and zenered the
gate at that same 10 uA, would the junction be affected in the long
term?

John

 

[Pooh Bear]

Supposed we had a hi-K ceramic cap, say 10 uF at 10 volts, and we
charge it from a smallish, say 10 uA, current source. I know that at
some voltage past 10, it will start to conduct and limit terminal
voltage, sort of like a very sloppy zener,

It will ? Is that a property of ceramic dielectrics ?

so it won't punch through
like, say, a film cap would. But is there a longterm damage mode?

With 10uA I'd be surprised if any damage happened. The power is *very*
limited.

Oh, and of I use a jfet as a super-low leakage diode, and zenered the
gate at that same 10 uA, would the junction be affected in the long
term?

Reverse bias leakage ? At 10uA I'd expect you'd be safe again. I've
avoided using the avalanche effect in bipolars to make a 'cheap zener'
though at even the mA level since no-one can typify the long term
junction characteristics.

Graham

 

[Derek Potter]

It will ? Is that a property of ceramic dielectrics ?

Good question. Film caps punch through because there is an avalanche
effect in the plastic. Surely this is also pretty likely with ceramics
too? Whether there's a reliable pre-avalalanche leakage I don't know.
In my experience "leaky" 1uF ceramics hold their charge for days. Most
ceramics have a very high overload capacity, 3x for example. Maybe
this is to ensure that you don't use the "sloppy zener" region?
However, it would be wrong to jump to any conclusion like that as Hi-K
ceramics lose their Hi-K when a voltage is applied.

With 10uA I'd be surprised if any damage happened. The power is *very*
limited.

Unfortunately, it's not. There is a certain amount of energy in the
capacitor, making the momentary power effectively unlimited. If any
positive feedback mechanism exists there will be current crowding
followed by destruction.

If the "sloppy zener" effect is real, I would be suspicious of what
sort of conduction is occurring. Odd thermal electrons are OK, but you
do not want 10uA of metallic ions moving through the lattice

Reverse bias leakage ? At 10uA I'd expect you'd be safe again. I've
avoided using the avalanche effect in bipolars to make a 'cheap zener'
though at even the mA level since no-one can typify the long term
junction characteristics.

I was taught that there's no particular failure mechanism associated
with zener or avalanche breakdown as long as the temperature rise is
not excessive. Unfortunately, breakdown cannot be guaranteed to be
simultaneous across the whole junction, so you can't push the device
to its nominal dissipation. I wouldn't guarantee long-term stability
though.

 

[[email protected]]

Good question. Film caps punch through because there is an avalanche
effect in the plastic. Surely this is also pretty likely with ceramics
too? Whether there's a reliable pre-avalalanche leakage I don't know.
In my experience "leaky" 1uF ceramics hold their charge for days. Most
ceramics have a very high overload capacity, 3x for example. Maybe
this is to ensure that you don't use the "sloppy zener" region?
However, it would be wrong to jump to any conclusion like that as Hi-K
ceramics lose their Hi-K when a voltage is applied.



Unfortunately, it's not. There is a certain amount of energy in the
capacitor, making the momentary power effectively unlimited. If any
positive feedback mechanism exists there will be current crowding
followed by destruction.

If the "sloppy zener" effect is real, I would be suspicious of what
sort of conduction is occurring. Odd thermal electrons are OK, but you
do not want 10uA of metallic ions moving through the lattice


I was taught that there's no particular failure mechanism associated
with zener or avalanche breakdown as long as the temperature rise is
not excessive. Unfortunately, breakdown cannot be guaranteed to be
simultaneous across the whole junction, so you can't push the device
to its nominal dissipation. I wouldn't guarantee long-term stability
though.

Zeners (above about 7.5V) don't break down by the Zener mechanism, but
by avalanching, and people have photographed interesting looking light
emissions from very lcalised "microplasmas". The interesting question
is why one of these localised avalanches doesn't heat up the narrow
tube of silicon involved, carry all the current available and heating
the silicon above its melting point.

Something like this does happen in laser diodes, which can be ruined by
a microsecond or so of over-current.

 

[Derek Potter]

Zeners (above about 7.5V) don't break down by the Zener mechanism, but
by avalanching, and people have photographed interesting looking light
emissions from very lcalised "microplasmas". The interesting question
is why one of these localised avalanches doesn't heat up the narrow
tube of silicon involved, carry all the current available and heating
the silicon above its melting point.

Possibly because the TC of avalanche breakdown is -ve so you get
automatic current sharing/ spreading?

I'm skeptical about "microplasmas". Plasma just means "fully ionised"
- there's no absolute requirement for it to be a hot gas. I seem to
recall the term being applied to silicon during avalanche even though
the crystal streucture is not disrupted.

You say "light emissions". Do you mean IR or can silicon emit in the
visible (other than due to incandescence!)?

 

[Rene Tschaggelar]

Supposed we had a hi-K ceramic cap, say 10 uF at 10 volts, and we
charge it from a smallish, say 10 uA, current source. I know that at
some voltage past 10, it will start to conduct and limit terminal
voltage, sort of like a very sloppy zener, so it won't punch through
like, say, a film cap would. But is there a longterm damage mode?

John,
the involved 10V sound small, but considering the
tiny distances between conductors, the field strength
is rather big. I wouldn't be surprised if it reached
MV/m. And then depending on the "arc-ing" mechanism,
the stored power is rather big to created damage.

Oh, and of I use a jfet as a super-low leakage diode, and zenered the
gate at that same 10 uA, would the junction be affected in the long
term?

Again here the microscopic "arc-ing" mechanism
defines the possible damage. Is is just a tunnel
current through some potential barrier, or does
it involve local ionization, oxidation, ...
There may be microscopic conducting channels that
overheat and change their doping.

Rene

 

[Joerg]

Hello John,

Supposed we had a hi-K ceramic cap, say 10 uF at 10 volts, and we
charge it from a smallish, say 10 uA, current source. I know that at
some voltage past 10, it will start to conduct and limit terminal
voltage, sort of like a very sloppy zener, so it won't punch through
like, say, a film cap would. But is there a longterm damage mode?

Just my 2 cents. I wouldn't rely on a certain leakage behavior. If it
decides to really break down the available current is going to be very
high because it holds a lot of charge by the time you are up there in
voltage.

Once when I had to find field failure causes on a Doppler unit I found
that an unintended charge pump effect was trickling up charges into a
Z5U cap. It destroyed those quite spectacularly.

Regards, Joerg

 

[John Larkin]

Bummer. I connected a 1 uF, 16 volt 0603 ceramic cap to a power
supply, through an ammeter. Stepping the voltage up, and after waiting
for annoying dielectric absorption things, current goes up sort of
linearly with voltage. I've run out of power supply at 120 volts, 35
nA. I'll let that cook for a while but, after 40 minutes, the current
only seems to be going down. So I don't have enough voltage, with this
supply, to see anything dramatic.

John

 

[Terry Given]

Bummer. I connected a 1 uF, 16 volt 0603 ceramic cap to a power
supply, through an ammeter. Stepping the voltage up, and after waiting
for annoying dielectric absorption things, current goes up sort of
linearly with voltage. I've run out of power supply at 120 volts, 35
nA. I'll let that cook for a while but, after 40 minutes, the current
only seems to be going down. So I don't have enough voltage, with this
supply, to see anything dramatic.

John

variac + neon xfmr + string of 1N4007s.......

actually I recall seeing a paper (marcon?) that discussed how high these
things can go, but alas cant recall any details, other than the actual
voltage rating gives the mfg a *lot* of wiggle room.

also, dynamic would be worse than static. next I shall teach my
grandmother to suck eggs....

Cheers
Terry

 

[Derek Potter]

Bummer. I connected a 1 uF, 16 volt 0603 ceramic cap to a power
supply, through an ammeter. Stepping the voltage up, and after waiting
for annoying dielectric absorption things, current goes up sort of
linearly with voltage. I've run out of power supply at 120 volts, 35
nA. I'll let that cook for a while but, after 40 minutes, the current
only seems to be going down. So I don't have enough voltage, with this
supply, to see anything dramatic.

Probably surface moisture. You need to de-salt and then dehydrate the
device before testing it - preferably in dry air.

Still, 120V is quite impressive for a 16V cap

 

[[email protected]]

Possibly because the TC of avalanche breakdown is -ve so you get
automatic current sharing/ spreading?

I'm skeptical about "microplasmas". Plasma just means "fully ionised"
- there's no absolute requirement for it to be a hot gas. I seem to
recall the term being applied to silicon during avalanche even though
the crystal streucture is not disrupted.

You say "light emissions". Do you mean IR or can silicon emit in the
visible (other than due to incandescence!)?

I imagine that it was near infra-red - the papers I saw showed
photgraphs of of localised light emission. It certainly wasn't
incandescence.

 

[John Devereux]

Bummer. I connected a 1 uF, 16 volt 0603 ceramic cap to a power
supply, through an ammeter. Stepping the voltage up, and after waiting
for annoying dielectric absorption things, current goes up sort of
linearly with voltage. I've run out of power supply at 120 volts, 35
nA. I'll let that cook for a while but, after 40 minutes, the current
only seems to be going down. So I don't have enough voltage, with this
supply, to see anything dramatic.

I tried this a while ago. I was running a bit close to the capacitor
voltage rating, and also I was worried about momentary over-voltage
from spikes (e.g. ESD events).

I couldn't get high enough either!

 

[John Larkin]

I tried this a while ago. I was running a bit close to the capacitor
voltage rating, and also I was worried about momentary over-voltage
from spikes (e.g. ESD events).

I couldn't get high enough either!

I guess I'll have to dig the old Spellman 30 kilovolt supply out of
the dungeon.

John

 

[Joerg]

Hello John,

I tried this a while ago. I was running a bit close to the capacitor
voltage rating, and also I was worried about momentary over-voltage
from spikes (e.g. ESD events).

I couldn't get high enough either!

Last time I had a ceramic capacitor accidentally run well above its
rated voltage there was a loud bang, the circuit breaker tripped, much
of the cap was gone and what remained of it had turned into a blob of
streaky green glass. Almost like a piece for jewelry.

Regards, Joerg

 

[Michael A. Terrell]

Last time I had a ceramic capacitor accidentally run well above its
rated voltage there was a loud bang, the circuit breaker tripped, much
of the cap was gone and what remained of it had turned into a blob of
streaky green glass. Almost like a piece for jewelry.

Well, it makes it easy to find the bad part when that happens. Kind
of like the custom made sheet silvered mica capacitors used in old high
power TV transmitters. When they fail, they really fail. A loud bang,
pieces of mica flying out of the transmitter, and the high voltage power
supply tripping a lot of overload relays and circuit breakers. Its
rather impressive to observe up close. I was about ten feet from a UHF
PBS TV station in Oxford, Ohio one day when one exploded, and the whole
engineering staff got busy. I was just visiting the station, but I got
the impression that this happened three or four times a year because of
the way they rushed to get back on the air without doing the required
preventative maintenance to prevent the next failure. I know they had a
lot of equipment problems, and most of the staff were college students.
I remember that the custom made sheet silvered mica capacitors for the
RCA TTU-25B 25 KW UHF transmitter were over $600 each, when they were
still available.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[John Devereux]

Hello John,


Last time I had a ceramic capacitor accidentally run well above its
rated voltage there was a loud bang, the circuit breaker tripped, much
of the cap was gone and what remained of it had turned into a blob of
streaky green glass. Almost like a piece for jewelry.

So... *are* ceramic capacitors ESD sensitive? Would be slightly ironic
if so. A bit like the ESD warnings you see on Transorb packaging.

 

[Joerg]

Hello John,

So... *are* ceramic capacitors ESD sensitive? Would be slightly ironic
if so. A bit like the ESD warnings you see on Transorb packaging.

Depends on the wave form and energy. A large spike into a smallish cap
without adequate resistance up front can certainly fry the cap. And has :-(

Regards, Joerg

 

[Joseph2k]

I imagine that it was near infra-red - the papers I saw showed
photgraphs of of localised light emission. It certainly wasn't
incandescence.

I suspect that the microplasma phenomona are at one boundary or the other of
the depletion region and thus quickly suck up all the availble carriers.
Incandescance is a bulk effect, but i'll bet like all plasma emmissions
that they radiate from infrared through ultraviolet and would look white to
the uv only protected eye.

 

[Joseph2k]

John,
the involved 10V sound small, but considering the
tiny distances between conductors, the field strength
is rather big. I wouldn't be surprised if it reached
MV/m. And then depending on the "arc-ing" mechanism,
the stored power is rather big to created damage.


Again here the microscopic "arc-ing" mechanism
defines the possible damage. Is is just a tunnel
current through some potential barrier, or does
it involve local ionization, oxidation, ...
There may be microscopic conducting channels that
overheat and change their doping.

Rene

Yes, returning to OP topic, yes they will break down, and at successively
lower voltages. They will arc around the grains of the ceramic material,
and the 10 uF will certainly store enough energy to see to the eventual
total destruction of the capacitor. If you are (un)lucky enough the cap
may explode on its first breakdown.

 

[John Larkin]

Yes, returning to OP topic, yes they will break down, and at successively
lower voltages. They will arc around the grains of the ceramic material,
and the 10 uF will certainly store enough energy to see to the eventual
total destruction of the capacitor. If you are (un)lucky enough the cap
may explode on its first breakdown.

Do you know that as fact, or is it conjecture?

I'd be surprised if a ceramic cap could store enough energy to
explode. We're talking millijoules here.

Guess I'll have to dig out a HV power supply.

John