Ceramic capacitor value changes drastically with DC bias

[JW]

I don't think I've seen this discussed before - I did not know this.

http://www.edn.com/design/analog/44...-4-7--F-capacitor-becomes-a-0-33--F-capacitor

 

[Jeroen Belleman]

I don't think I've seen this discussed before - I did not know this.

http://www.edn.com/design/analog/44...-4-7--F-capacitor-becomes-a-0-33--F-capacitor

Yeah, I got acquainted with that particular problem in the 1980's.
High-K ceramics have all sorts of non-ideal behaviour. Capacitance
varies with temperature and bias voltage, they are piezo-electric,
hysteretic and tend to crack. If they couldn't be made so tiny,
no one would want to use them.

There was one thing in your link that *did* surprise me: One
commenter claims to have used a batch of caps that would oscillate
all by themselves at 900 MHz when biased. I won't believe that
until I see it.

Jeroen Belleman

 

[George Herold]

I don't think I've seen this discussed before - I did not know this.

http://www.edn.com/design/analog/4402049/Temperature-and-voltage-vari...

It's been talked about in passing here, where the idea of using a Z5U
(or Y) as part of a low frequency VCO.. (Hey Joerg how 'bout a cermaic
cap for you 10kHz VCO?)

George H.

 

[Uwe Hercksen]

There was one thing in your link that *did* surprise me: One
commenter claims to have used a batch of caps that would oscillate
all by themselves at 900 MHz when biased. I won't believe that
until I see it.

Hello,

a parametric oscillator?
If I would see it, I would try to remove all semiconductors from that
board to see if it still oscillates.

Bye

 

[George Herold]

You could make a parametric amplifier with them. Power gain would be huge..

Scratch, scratch... (google google)... OK I'm not sure how a
parametric amplifer works. I 'did' a parametric oscillator in the
past, but even that is a bit vague now. (I recall a plot of drive
amplitude versus frequency that looked like a semi-circle, with a
threshold.)

Got a link (or pic) that might explain it? (It's a mixer type thing,
no?)

George H.

 

[Joerg]

It's been talked about in passing here, where the idea of using a Z5U
(or Y) as part of a low frequency VCO.. (Hey Joerg how 'bout a cermaic
cap for you 10kHz VCO?)

I have used the effect for moving a filter around. It doesn't get any
cheaper than that.

Right now I am looking for capacitors with the worst microphonics
because I want to use them ... as contact sound sensors. But has to be
low enough in capacitance so I can reasonably squeeze 100Hz bandwidth
out of them.

 

[Joerg]

variation-of-ceramic-capacitors--or-why-your-4-7--F-capacitor-becomes-
a-0-33--F-capacitor

It's been discussed. It's a known issue with high dielectric constant
ceramics, and part of the reason that you want to avoid them unless you
really need the capacitance in the space you can provide.

Also, don't automatically assume that X7R is always golden and never has
that problem. The very high density ones often do and it can be very
tough to pry some data about that out of manufacturers. BTDT, a lot.
Murata is pretty good when it comes to furnishing such data so they are
currently my preferred source.

 

[tm]

I have used the effect for moving a filter around. It doesn't get any
cheaper than that.

Right now I am looking for capacitors with the worst microphonics
because I want to use them ... as contact sound sensors. But has to be
low enough in capacitance so I can reasonably squeeze 100Hz bandwidth
out of them.

--

If you take a ceramic capacitor and apply the rated DC voltage across it and
heat to above the curie point, when it cools it will exhibit the piezo
effect.

 

[Joerg]

Delta-c over C should be independent of C, so if you charge a cap
through a big resistor, and look at the voltage, bandwidth should be
OK for a big cap.

With MLCC one never knows. All kinds of weird effects.

I did a little testing about ceramic caps generating noise from
tapping the board and didn't see anything. We lost phase lock when a
front-panel SMB connector was pulled out... turned out to be the XO.
We had some tiny springs fabricated to shock isolate the dip14
oscillator.

When I got the Fluke 8845A, or rather after UPS (literally!) threw it
over the backyard fence, I took an older ceramic cap and hooked it up to
it. Turned it to AC, all 6-/12 digits of it. Every time I clapped my
hands near that cap I could see the excursion.

This meter is nice, no fan. But the relays clicking around when in
ohms-mode are a bit annoying.

How about a cheap piezo chirper disk?

They are not very good at very low frequencies, sub-Hertz and such.

I once designed a circuit that crashed a product if the cover was
removed, using a phototransistor to detect light. I never used it.

Or design one where it quits working for a minute every time someone
says "s..t!"

 

[tm]

With MLCC one never knows. All kinds of weird effects.



When I got the Fluke 8845A, or rather after UPS (literally!) threw it
over the backyard fence, I took an older ceramic cap and hooked it up to
it. Turned it to AC, all 6-/12 digits of it. Every time I clapped my
hands near that cap I could see the excursion.

This meter is nice, no fan. But the relays clicking around when in
ohms-mode are a bit annoying.



They are not very good at very low frequencies, sub-Hertz and such.



Or design one where it quits working for a minute every time someone
says "s..t!"

--

I wonder what the feedback sounds like with that.

 

[Joerg]

I wonder what the feedback sounds like with that.

Like this: Dang!

For customers in the south: Daiang!

 

[Joerg]

John Larkin wrote:

[...]

When I got the Fluke 8845A, or rather after UPS (literally!) threw it
over the backyard fence, I took an older ceramic cap and hooked it up to
it. Turned it to AC, all 6-/12 digits of it. Every time I clapped my
hands near that cap I could see the excursion.

This meter is nice, no fan. But the relays clicking around when in
ohms-mode are a bit annoying.

I have an 8845A, replacing the horrible Chinese rebrand Keithley I
sent back. Great meter, the Fluke.

The relay click is a good audible continuity indicator!

Ah, that's what it's for

It is a great meter but one thing I'll never understand: It's not the
slowest on the continuity beep but if I drag the probe across a set of
traces too fast it misses. The cheapo Chinese meter doesn't! So for
continuity tests I'll keep using one of the Chinese meters. I wish the
large companies would talk to their clients some more.

I'll still have to try out LAN and PC operation. It shipped with a
little USB-RS232 gizmo. For the LAN I'll wait for warmer weather because
it's a crawl space job to get Ethernet there. It came with some sort of
Fluke software but (as usual with many US companise ...) that seems to
only be some teaser kit.

 

[Tim Williams]

High capacity surface-mount ceramics also make good microphones when
they've got a lot of bias on them. They're great blocking caps if you
have a low-level signal riding on a lot of DC that you want to corrupt
with ambient vibration.

I built a little LNA (not wideband, ends ca. 50MHz IIRC) with 10uF 50V
1206 X7R caps coupling the input and output. Thought it was reading
strange waveforms, maybe power supply transients or something, until I
realized they were microphonic.

On further inspection, the transients observed from tapping the PCB have a
frequency reasonable with respect to the acoustic length of said board.

FWIW, that was around 10mV for a casual finger tap. Think the caps were
TDK. Soldered with big globs on either end, which will tend to increase
strain.

Tim

 

[Joerg]

I built a little LNA (not wideband, ends ca. 50MHz IIRC) with 10uF 50V
1206 X7R caps coupling the input and output. Thought it was reading
strange waveforms, maybe power supply transients or something, until I
realized they were microphonic.

On further inspection, the transients observed from tapping the PCB have a
frequency reasonable with respect to the acoustic length of said board.

FWIW, that was around 10mV for a casual finger tap. Think the caps were
TDK. Soldered with big globs on either end, which will tend to increase
strain.

Here in CA, when sudden weird jitters start to appear on a circuit like
that, it's better to get the hell out of the building

 

[Jamie]

I built a little LNA (not wideband, ends ca. 50MHz IIRC) with 10uF 50V
1206 X7R caps coupling the input and output. Thought it was reading
strange waveforms, maybe power supply transients or something, until I
realized they were microphonic.

On further inspection, the transients observed from tapping the PCB have a
frequency reasonable with respect to the acoustic length of said board.

FWIW, that was around 10mV for a casual finger tap. Think the caps were
TDK. Soldered with big globs on either end, which will tend to increase
strain.

Tim

Years ago when I used to do some repairs for my buddies, I had one bring
me a guitar amplifier. He could not use it because it was feed
back, even when nothing plugged into it. He also noted when tapping on
the cabinet, it was perfectly microphonic.

When I finally found the problem, it was a transistor in a plastic
package. The package was fine and so was the transistor but something
inside broke loose and was playing with the beta and made a perfect
microphone.

I used that transistor in an FM oscillator and it worked perfectly as
a bug!..

Jamie

 

[George Herold]

I built a little LNA (not wideband, ends ca. 50MHz IIRC) with 10uF 50V
1206 X7R caps coupling the input and output.  Thought it was reading
strange waveforms, maybe power supply transients or something, until I
realized they were microphonic.

On further inspection, the transients observed from tapping the PCB have a
frequency reasonable with respect to the acoustic length of said board.

FWIW, that was around 10mV for a casual finger tap.  Think the caps were
TDK.  Soldered with big globs on either end, which will tend to increase
strain.

Tim

Grin... well can you turn the lemon into lemonade?
some sort of vibration sensor...
upthread tm said to heat it above it Curie temperature.

George H.

 

[tm]

I built a little LNA (not wideband, ends ca. 50MHz IIRC) with 10uF 50V
1206 X7R caps coupling the input and output. Thought it was reading
strange waveforms, maybe power supply transients or something, until I
realized they were microphonic.

On further inspection, the transients observed from tapping the PCB have a
frequency reasonable with respect to the acoustic length of said board.

FWIW, that was around 10mV for a casual finger tap. Think the caps were
TDK. Soldered with big globs on either end, which will tend to increase
strain.

Tim

Grin... well can you turn the lemon into lemonade?
some sort of vibration sensor...
upthread tm said to heat it above it Curie temperature.

George H.

+++++++++++++++++++++++++++++++

With a DC voltage applied. Allow it to cool with the DC applied and it
should become very microphonic. Put an audio tone on it and you should hear
it sing. Usually a soldering iron will get it hot enough but the material
makes a difference.

 

[George Herold]

Grin... well can you turn the lemon into lemonade?
some sort of vibration sensor...
upthread tm said to heat it above it Curie temperature.

George H.

+++++++++++++++++++++++++++++++

With a DC voltage applied. Allow it to cool with the DC applied and it
should become very microphonic. Put an audio tone on it and you should hear
it sing. Usually a soldering iron will get it hot enough but the material
makes a difference.- Hide quoted text -

- Show quoted text -

Yeah I was going to ask what the curie temperature is?
Is there a better 'flavor'?

Thanks,
George H.

 

[Tim Williams]

Yeah I was going to ask what the curie temperature is?
Is there a better 'flavor'?

AFAIK, barium titanate and related blends are in the 120C+ range, maybe
200C tops. Comparable to ferrites. They're definitely annealed in
soldering, the chips at least.

If you have old discs that aren't measuring right, you can try soldering
them, I suppose the mass and size of the disc may prevent the whole thing
from heating up.

Tim

 

[Nico Coesel]

I have used the effect for moving a filter around. It doesn't get any
cheaper than that.

Right now I am looking for capacitors with the worst microphonics
because I want to use them ... as contact sound sensors. But has to be
low enough in capacitance so I can reasonably squeeze 100Hz bandwidth
out of them.

Piezo elements are too expensive I presume?