Effects of vibration on capacitors

[Terry Given]

The only porovblem then becomes how to tell a hi K from a Low K. Like,
they con't come labeled as to that factor.

I suppose you could put them into a high gain amp circuit and plink on
them and see if they put out something. But then that's too easy, isn't
it.

A simple way is with a capacitance meter and a heat gun - high K
dielectrics drop about 80% at 60C or so. solder a pair of wires (so you
dont melt the tester leads), hook up the cap. measure C at room temp,
then heat up. If it dont change much, its X7R or better. If it changes
dramatically, its Z5U/Y5V.

The other approach is with a series cap (100x larger) and applying DC
bias (or, if you have a fancy LCR bridge, ask it to measure C at varying
DC bias)

Cheers
Terry

 

[Ken Taylor]

Replied to what question? You began a new thread, there is no question.

Just what you'd expect from a Democrat!

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

A simple way is with a capacitance meter and a heat gun - high K
dielectrics drop about 80% at 60C or so. solder a pair of wires (so you
dont melt the tester leads), hook up the cap. measure C at room temp,
then heat up. If it dont change much, its X7R or better. If it changes
dramatically, its Z5U/Y5V.

The other approach is with a series cap (100x larger) and applying DC
bias (or, if you have a fancy LCR bridge, ask it to measure C at varying
DC bias)

I found this:

http://www.kemet.com/kemet/web/homepage/kechome.nsf/vabypagename/militar
yfaq

<<
Question: Are your military ceramic capacitors subject to the
piezoelectric effect?
Answer: Certain classes of ceramic capacitors exhibit a normal
characteristic, called piezoelectricity, than can cause unexpected
effects in certain circuits. In some cases, the piezoelectric effect
may result in the appearance of electrical noise, while in other cases,
an acoustic sound may be heard, coming from the capacitor itself.
Ceramic piezo effects are well known, and were even the basis for the
ceramic phono cartridges used in the past.

Piezoelectricity is a common characteristic of many ceramic chip
capacitors and occurs in those classes of dielectric which are
classified as ferroelectric. Piezoelectric effects can result in noise
for ferroelectric ceramic chips, such as those used for military BX &
BR, as well as commercial EIA Class 2 and Class 3 dielectric, such as
X7R, X5R, X8R, Y5V, Y5U, Z5U, etc. Piezoelectricity occurs in all
ferroelectric dielectrics, regardless of manufacturer. Note that there
are essentially no piezoelectric effects in Class 1 capacitors, such as
C0G, NP0, or military BP - none of which are ferroelectric.
.....
And what i would really like to know more about is what is mentioned in
"commercial EIA Class 2 and Class 3 dielectric, such as X7R, X5R, X8R,
Y5V, Y5U, Z5U, etc." Like, what's this "EIA Class 2 and Class 3
dielectric, such as X7R, X5R, X8R, Y5V, Y5U, Z5U, etc."?? Where can I
find more detailed info? Or is ths another high-priced EIA document
that's not available online? And no one is willing to go into detail
about the subject?

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

Yet another reason to avoid the poxy things. Z5U, Y5V = shite.

Avoid, how? I come across a lot of caps tha aren't marked as such, all
they have is '104' marked on them, ferinstance. How do I know if
they're "shite"? And then there's the problem of how to avoid them, if
a replacment has other serious disadvantages, such as being much bigger?

That's why I want to know more aout the different grades of caps, and
what those designations mean. That way, I can make more intelligent
decisions on whether or not they're suitable for a certain application,
and not just make an uninformed generalization and claim they're all
"shite".

 

[ddwyer]

Or is ths another high-priced EIA document
that's not available online? And no one is willing to go into detail
about the subject?

Only cog NPO plastic and electrolytic arte free from piezo
michrophonics.
Silver mica has scintillation which is random cap variation.
The only way big values are available in small volumes is to reduce the
insulation layers= lower voltage working or doping with piezo materials
= michrophonics..
Many big value HiK (high dilectric constant caps have such poor
variation over temperature down to 20% of value hot and cold!that their
advantage is an illusion.

 

[Terry Given]

Avoid, how? I come across a lot of caps tha aren't marked as such, all
they have is '104' marked on them, ferinstance. How do I know if
they're "shite"? And then there's the problem of how to avoid them, if
a replacment has other serious disadvantages, such as being much bigger?

That's why I want to know more aout the different grades of caps, and
what those designations mean. That way, I can make more intelligent
decisions on whether or not they're suitable for a certain application,
and not just make an uninformed generalization and claim they're all
"shite".

Fair comments.

http://www.avx.com/docs/Catalogs/cy5v.pdf

http://www.avx.com/docs/Catalogs/cx7r.pdf


The two interesting curves are on the first page:
1 - temperature coefficient (its not but thats what its labelled; its
actually a %C change-vs-T

2 - capacitance change vs DC bias. at 20% rated voltage, -60% C. at 40%
rated V, -80%; at 60% rated V, -90%

Z5U dielectric is similarly dreadful; a little web-browsing will show
you how dreadful.

I only design with smt parts, but IME I can get X7R parts (which are a
HUGE improvement - look at the link) in the same package.

The funny thing about Y5V/Z5U is that in most apps I have seen (esp.
smt) the DC bias reduces the actual capacitance well below what you can
get with a similar-sized X7R cap. From the avx data, an 0603 1uF 10V Y5V
cap has about 150nF actual capacitance at 5V DC bias (say as a
decoupling cap) and 20C; at -20C its 75nF; at +85C its 60nF. I can also
get a 0603 220nF 10V X7R cap, which has 220nF at 20C, 215nF at -20C and
203nF at 85C.

I first "discovered" this jabout 15 years ago, when testing an IGBT
gatedrive flyback smps (+25V) with a constant-current load - during the
on-time the output voltage dropped much, much faster than I expected
with my 220nF 50V shite cap. I used a bench psu & ammeter to verify the
load current was right, and then used I/(dV/dt) to calculate C, which
was (IIRC) around 10nF. And then went "what the...?". Out came the
databook (you'll remember them), and voila - there was the answer,
written for all to see. I double-checked the tempco with a heatgun & LCR
bridge, then played silly buggers with the DC bias to measure that too.
Sure enough, Philips were right. My circuit also had to run at 85C,
which just made things worse. I swapped the part for a 1206 220nF 50V
X7R, and got 20x more capacitance. That started my "anti-shite cap"
crusade

As for unmarked caps - test with heatgun/DC bias. If you use these shite
caps, only run them at < 10% of rated voltage.

Cheers
Terry

 

[Rich Grise]

Avoid, how? I come across a lot of caps tha aren't marked as such, all
they have is '104' marked on them, ferinstance. How do I know if
they're "shite"? And then there's the problem of how to avoid them, if
a replacment has other serious disadvantages, such as being much bigger?

That's why I want to know more aout the different grades of caps, and
what those designations mean. That way, I can make more intelligent
decisions on whether or not they're suitable for a certain application,
and not just make an uninformed generalization and claim they're all
"shite".

Z5U and Y5V are very cheap ceramic dielectrics that pack a lot of
picofarads into a relatively small package, at the cost of abysmal
tolerance and stability specs, and possibly ESR - but two out of
three of these are irrelevant for decoupling, and the third might
actually enhance it, by lowering the Q of the decoupling cap,
tending towards a snubber sort of thing. X7R, I guess, is one of
the high-stability ones, either by tolerance or tempco; NPO means
negative-positve-Zero (tempco), which, for your convenience, they
symbolize with '0', which everybody reads as 'O' anyway.

Any more than that, I'd have to recommend to you to Our Friend,
Google The Wise, for charts and stuff.

If you'd like to hear from the bench tech in me, I'd use polyester
or some very expensive thing like that if I wanted to get 1% out
of a 555. For decoupling, use the cheapest ceramics you can get,
and one cap per chip is not too many. And I would try to sprinkle
tantalums about, 1 - 10 uF, and one 100uF aluminum honker at the
board's power entry point. For two rails, of course, double this,
observing polarity.

For RF, I've heard great things about silvered mica, but that
might just be my inner museum bringing them up because all those
millennia ago that was the best they could come up with.

And, of course, as usual, (all together now,) It Depends.

As far as microphonics. Since I am as good as totally clueless
about that compared to the information the other posters have
been kind enough to share, I'll let that go other than to say
that I understand the principle, but I've never encountered
microphonics from any capacitor that I've noticed at the time.
I haven't ever gone _looking_ for them, and whenever I've
encountered microphonics, there's always been a coil or some
component flapping in the breeze or something, to account for
it.

Try shouting at your All-American-Five and see if your voice
comes out the speaker!

Hope This Helps!
Rich

 

[Rich Grise]

And what i would really like to know more about is what is mentioned in
"commercial EIA Class 2 and Class 3 dielectric, such as X7R, X5R, X8R,
Y5V, Y5U, Z5U, etc." Like, what's this "EIA Class 2 and Class 3
dielectric, such as X7R, X5R, X8R, Y5V, Y5U, Z5U, etc."?? Where can I
find more detailed info? Or is ths another high-priced EIA document
that's not available online? And no one is willing to go into detail
about the subject?

http://www.google.com/search?q=ceramic+dielectric+comparison+chart&btnG=Search
First Hit:
http://www.avxcorp.com/docs/techinfo/dielectr.pdf



Have Fun!
Rich

 

[Ross Herbert]

The time spent under high vibration for the life of this project will
be minutes, at most. I'm not worried about lead or solder fatigue.

Well, if you had provided this info initially it would have saved a
lot of time spent considering this aspect.

snip

Pardon? You're stating that if you took a charged cap, changed the
distance between the plates and measured the voltage across it while
doing so, there'd be no change in voltage at all?

I didn't say "no change in voltage at all" but when we are considering
the proposed subject (in your initial statement) of "voltage
regulation" we must assume (since you also didn't mention exactly what
magnitudes you were referring to) that this would involve voltages of
at least several volts. In this sort of application even a few microns
displacement in an electro cap would make no practical difference in
its output voltage. It may change by a couple of microvolts but that's
about all. Now if you are referring to "voltage regulation" in the
millivolts range then you must be dealling with extremely low currents
to boot, and in this case you wouldn't be using the large type of
electro caps anyway.

Compare that "sub-micron" movement with the actual thickness of the
dielectric material. It is very thin to begin with. A few microns of
compression of the dielectric could be a small percentage of the
thickness of the dielectric. I didn't know. THAT'S WHY I ASKED.

Well why don't you spell out the exact situation together with a
precise description of what you are trying to do and at what voltage
and currents and trhe test setup you are using. We are all guessing
unless you do this.

You statement implies the dielectric would be infinitely hard and
incompressable. No substance is.

- would not be noticeable and would therefore have

Well, I tried this for myself. Using a speaker and running several
different frequencies through it (with the cap attached through a
linkage 10" away), the oscope showed a clear voltage deviation. Some
frequencies (10kHz) were worse than others (500Hz). Maybe before
attacking someone's question, you may want to make sure you're not
making too many assumptions.

Now you are introducing a completely different set of circumstances.
Just what has your original question got to do with alternating
voltages at audio frequencies. Your question was about "voltage
regulation", and that implies DIRECT CURRENT.

What about my question implied that I was looking for an argument?
I'm not sure if you were just in a bad mood when you replied, but it
was a civilized question, did not insult anyone, and had several
people talk about microphonics.

Dave

Well it just seems to me to be that you might be trying to stimulate
some discussion purely for your own entertainment. If you really want
some help on this describe exactly what it is you trying to
accomplish.

 

[David Harper]

Well, if you had provided this info initially it would have saved a
lot of time spent considering this aspect.

I never asked about mechanical fatigue. I was only worried about
effects on voltage.

Now you are introducing a completely different set of circumstances.
Just what has your original question got to do with alternating
voltages at audio frequencies.

This 'shoestring' test measured voltage deviation at 5VDC, not AC.
Secondly, mechanical vibration outside of 'audio frequencies' is
rarely a factor.

Well it just seems to me to be that you might be trying to stimulate
some discussion purely for your own entertainment. If you really want
some help on this describe exactly what it is you trying to
accomplish.

I have a microcontroller reading a dual-channel 12-bit ADC, one
channel reading a ADXL150 accelerometer and the other a motorola 6115
pressure sensor. It is mounted on a high powered rocket (not the
little 'C' engines kids shoot, but rather H, I, and J engines, which
can exceed 100lbs of thrust and vibrate quite a bit). During the
boost phase, I expect to see around 15g's, as well as quite a bit of
vibration especially as it approaches Mach 1.

My question is how much could variation in the 5VDC reference voltage
affect the measurement of the accelerometer during the boost phase?
The pressure sensor isn't read (nor deploys the parachute) until after
the boost phase, so I'm not concerned with that.

Dave

 

[Ross Herbert]

I have a microcontroller reading a dual-channel 12-bit ADC, one
channel reading a ADXL150 accelerometer and the other a motorola 6115
pressure sensor. It is mounted on a high powered rocket (not the
little 'C' engines kids shoot, but rather H, I, and J engines, which
can exceed 100lbs of thrust and vibrate quite a bit). During the
boost phase, I expect to see around 15g's, as well as quite a bit of
vibration especially as it approaches Mach 1.

My question is how much could variation in the 5VDC reference voltage
affect the measurement of the accelerometer during the boost phase?
The pressure sensor isn't read (nor deploys the parachute) until after
the boost phase, so I'm not concerned with that.

Dave

Now we can start to understand what your question really is about.
Thanks for the welcome info.

For a start you should determine the worst case current drain on your
5VDC reference and endeavour to select components and circuitry to
keep the drain as low as possible, buffering the supply if necessary.
You should then use capacitors with the least volume possible to do
the job. Preferably they should have solid electrolyte while being
physically small and you can parallel as many as needed of the
selected component as are required. This approach will eliminate any
tendancy towards voltage variation due to plate displacement within
the capacitors themselves. That's where I would start.

 

[David Harper]

Now we can start to understand what your question really is about.
Thanks for the welcome info.

For a start you should determine the worst case current drain on your
5VDC reference and endeavour to select components and circuitry to
keep the drain as low as possible, buffering the supply if necessary.
You should then use capacitors with the least volume possible to do
the job. Preferably they should have solid electrolyte while being
physically small and you can parallel as many as needed of the
selected component as are required. This approach will eliminate any
tendancy towards voltage variation due to plate displacement within
the capacitors themselves. That's where I would start.

Thanks for the advice. I assume by solid electrolyte, you mean
something like tantalum? I'll probably end up putting a couple in
parallel and see if I can dampen the board from vibe as much as
possible.

Dave

 

[Roger Hamlett]

Does anyone know how severely vibration can affect a capacitor's
ability to regulate voltage? (i.e. how much the voltage can deviate
as a function of vibration) What types of caps are better at
regulating voltage under high vibration?

Thanks in advance!
Dave

Lots of comments so far about the various types of component, and the
effects. However I wonder if you could be 'ingenious', and make the
effects cancel?. If (for instance), you arranged four capacitors, in a
tight circle, each rotated 90 degrees to the next, and mechanically
coupled them together (epoxy), and electrically connected them in
parallel, then presumably (within the limits of the mechanical propogation
speed of the actual vibration), the effects would largely cancel. It might
be worth considering, if the capacitor sizes required, were larger than
available in types that have little response to vibration.

Best Wishes

 

[Ross Herbert]

Thanks for the advice. I assume by solid electrolyte, you mean
something like tantalum? I'll probably end up putting a couple in
parallel and see if I can dampen the board from vibe as much as
possible.

Dave

Do Google search for 'solid electrolyte capacitor' and you will see
plenty of hits.

They don't have to be tantalum but this type is common in smd. Solid
electrolyte is also used in aluminium capacitors.

http://www.vishay.com/docs/28355/123sala.pdf

 

[David Harper]

Lots of comments so far about the various types of component, and the
effects. However I wonder if you could be 'ingenious', and make the
effects cancel?. If (for instance), you arranged four capacitors, in a
tight circle, each rotated 90 degrees to the next, and mechanically
coupled them together (epoxy), and electrically connected them in
parallel, then presumably (within the limits of the mechanical propogation
speed of the actual vibration), the effects would largely cancel. It might
be worth considering, if the capacitor sizes required, were larger than
available in types that have little response to vibration.

Best Wishes

I guessing that it would not work much better than 4 caps in parallel
with no specific orientation... your idea assumes that the vibration
affects capacitors arranged in different angles in opposite manners.
It may affect them identically. Secondly, the overall effects could
change significantly depending on the frequency of vibration.

Dave

 

[Roger Hamlett]

I guessing that it would not work much better than 4 caps in parallel
with no specific orientation... your idea assumes that the vibration
affects capacitors arranged in different angles in opposite manners.
It may affect them identically. Secondly, the overall effects could
change significantly depending on the frequency of vibration.

Yes. It was a 'thought exercise', but with some capacitors exhibiting
piezo effects, might be worth considering.

Best Wishes

 

[ddwyer]

Yes. It was a 'thought exercise', but with some capacitors exhibiting
piezo effects, might be worth considering.

Best Wishes

Practical experience in your application.
SM components helpful also a thin layer of flexible encapsulant to damp
board vibration.
As a low cost approach single sided sm with the other side bonded
(double sided foam?) to a structural member would be beneficial.