[Adam S]

Everyone knows that high dialectic constant ceramic capacitors (X7R,
Z5U, Y5V) exhibit voltage sensitivity. This apparently causes distortion
when any "significant" AC voltage appears across the capacitor. In a
filter this would be the -3dB corner frequency. Does anyone have any
idea how much distortion it actually produces ?
Due to limited PCB space I want to use ceramics for DC blocking at the
input of some FET op-amps. I see that high voltage X7R caps have the
lowest voltage sensitivity but I don't have a THD analyzer to get any
meaningful numbers. For example, would a high pass made with 50V X7R
100nF and a 1M ohm resistor give me THD = 0.02% at 20Hz ?

Adam

[Bob Quintal]

Adam S not.valid@nosuchaddress wrote in
u:

Everyone knows that high dialectic constant ceramic capacitors
(X7R, Z5U, Y5V) exhibit voltage sensitivity. This apparently
causes distortion when any "significant" AC voltage appears
across the capacitor. In a filter this would be the -3dB
corner frequency. Does anyone have any idea how much
distortion it actually produces ? Due to limited PCB space I
want to use ceramics for DC blocking at the input of some FET
op-amps. I see that high voltage X7R caps have the lowest
voltage sensitivity but I don't have a THD analyzer to get any
meaningful numbers. For example, would a high pass made with
50V X7R 100nF and a 1M ohm resistor give me THD = 0.02% at
20Hz ?

Adam

AVX Capacitors offers a comparison between ceramic and tantalum
that includes some data on hte problem
http://www.avxcorp.com/docs/techinfo/mlc-tant.pdf

If you keep the DC Bias high wrt the AC component there isn't a
problem.



--
Bob Quintal

PA is y I've altered my email address.

--
Posted via a free Usenet account from http://www.teranews.com

[Bob Woodward]

Adam S wrote:
Everyone knows that high dialectic constant ceramic capacitors (X7R,
Z5U, Y5V) exhibit voltage sensitivity. This apparently causes distortion
when any "significant" AC voltage appears across the capacitor. In a
filter this would be the -3dB corner frequency. Does anyone have any
idea how much distortion it actually produces ?
Due to limited PCB space I want to use ceramics for DC blocking at the
input of some FET op-amps. I see that high voltage X7R caps have the
lowest voltage sensitivity but I don't have a THD analyzer to get any
meaningful numbers. For example, would a high pass made with 50V X7R
100nF and a 1M ohm resistor give me THD = 0.02% at 20Hz ?

Adam

Some time ago I did some testing on this but with my A1 set that has a
bottom THD reading at 0.002% I could not find any distortion, whatever
capacitor and at any voltage or frequency.

Someone with an AP who can measure a decade lower ??

Andre

[Eeyore]

Adam S wrote:

Everyone knows that high dialectic constant ceramic capacitors (X7R,
Z5U, Y5V) exhibit voltage sensitivity. This apparently causes distortion
when any "significant" AC voltage appears across the capacitor. In a
filter this would be the -3dB corner frequency. Does anyone have any
idea how much distortion it actually produces ?
Due to limited PCB space I want to use ceramics for DC blocking at the
input of some FET op-amps. I see that high voltage X7R caps have the
lowest voltage sensitivity but I don't have a THD analyzer to get any
meaningful numbers. For example, would a high pass made with 50V X7R
100nF and a 1M ohm resistor give me THD = 0.02% at 20Hz ?

Use an electrolytic. It'll way outperform a ceramic.

Graham

[Adam S]

Bob Woodward wrote:

Some time ago I did some testing on this but with my A1 set that has a
bottom THD reading at 0.002% I could not find any distortion, whatever
capacitor and at any voltage or frequency.

Do you remember the component values you tested were ? I assume they
were not NPO or COG ceramics as these behave essentially like a film
capacitors and should be "distortionless".

[Bob Woodward]

Adam S wrote:
Bob Woodward wrote:

Some time ago I did some testing on this but with my A1 set that has a
bottom THD reading at 0.002% I could not find any distortion, whatever
capacitor and at any voltage or frequency.

Do you remember the component values you tested were ? I assume they
were not NPO or COG ceramics as these behave essentially like a film
capacitors and should be "distortionless".

It was a X7R, 100n 0805 from PHILIPS ( today BC components )
Since then I have noticed that almost everyone uses conventional, wired,
film-capacitors for audio-signals.
That includes major pro-audio manufacturers and electronic magazines.

When no space on board-surface let an electrolytic fly above the other
components,.

Andre

[GregS]

In article , Adam S not.valid@nosuchaddress wrote:
Everyone knows that high dialectic constant ceramic capacitors (X7R,
Z5U, Y5V) exhibit voltage sensitivity. This apparently causes distortion
when any "significant" AC voltage appears across the capacitor. In a
filter this would be the -3dB corner frequency. Does anyone have any
idea how much distortion it actually produces ?
Due to limited PCB space I want to use ceramics for DC blocking at the
input of some FET op-amps. I see that high voltage X7R caps have the
lowest voltage sensitivity but I don't have a THD analyzer to get any
meaningful numbers. For example, would a high pass made with 50V X7R
100nF and a 1M ohm resistor give me THD = 0.02% at 20Hz ?

Adam

The Z after the cap will have the most effect.

greg

[Ian Iveson]

Bob Woodward wrote:

Do you remember the component values you tested were ? I assume they
were not NPO or COG ceramics as these behave essentially like a film
capacitors and should be "distortionless".

As blameless as polypropylene, according to extensive tests published
in Elektor by Cyril Bateman.

From memory, other ceramics range around -60 to -80dB distortion with
1V at 1kHz across them. Common polar electrolytics are better, but
poor examples not by much.

Films ranged from around -90 to being unmeasurable with a floor
at -120dB. Order was as you might expect, with polypropylene the
lowest.

Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal says
"If you keep the DC Bias high wrt the AC component there isn't a
problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

cheers, Ian

[Richard Crowley]

See the PDF document on John Hardy's 990 preamp
where there is a section at the end about his use of COG/
NPO ceramics in the signal path. www.johnhardyco.com

[Chris Morriss]

In message , Adam S
writes
Everyone knows that high dialectic constant ceramic capacitors (X7R,
Z5U, Y5V) exhibit voltage sensitivity. This apparently causes
distortion when any "significant" AC voltage appears across the
capacitor. In a filter this would be the -3dB corner frequency. Does
anyone have any idea how much distortion it actually produces ?
Due to limited PCB space I want to use ceramics for DC blocking at the
input of some FET op-amps. I see that high voltage X7R caps have the
lowest voltage sensitivity but I don't have a THD analyzer to get any
meaningful numbers. For example, would a high pass made with 50V X7R
100nF and a 1M ohm resistor give me THD = 0.02% at 20Hz ?

Adam

X7R dielectric isn't too bad from the capacitance versus dc bias point
of view. Z5U and Y5V are absolutely dreadful. If you use RMAA to test
certain soundcards you'll find some that have horrendous IM distortion
when using an LF and HF signal, when driving into relatively low load
impedances. The reason for this is that the small value of the output
cap means that there is a large-ish signal appearing across the output
cap at the low frequencies, thus making its capacitance value alter with
the LF portion of the signal. Result is poor LF distortion, but ok HF
distortion, and also poor intermodulation distortion. Replace the Y5V
output caps with X7R (if you have the space) and the performance is much
improved.
--
Chris Morriss

[Bob Woodward]

Ian Iveson wrote:
Bob Woodward wrote:

Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal says
"If you keep the DC Bias high wrt the AC component there isn't a
problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

cheers, Ian

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having the cap
inversely polarized for half periods of the audio signal.

Andre

[Ian Iveson]

Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal
says "If you keep the DC Bias high wrt the AC component there isn't
a problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having the cap
inversely polarized for half periods of the audio signal.

Particularly with electrolytics as I remember. I wish I could find the
article...actually Electronics World rather than Elektor. At the time
I read it I was not aware of some of the myths he claimed to explode.
I remember wishing that he had included some lower frequency tests
because he said that electrolytic action is too slow to have an
appreciable effect at audio frequencies.

Non-polar electrolytics, used with zero bias, tested well, as did
back-to-back polar electrolytics.

The article may be available from EW, and I believe Cyril produced a
CDROM of his work with capacitors. Not for free though.

cheers, Ian

[Bob Woodward]

Ian Iveson wrote:
Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal
says "If you keep the DC Bias high wrt the AC component there isn't
a problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having the cap
inversely polarized for half periods of the audio signal.

Particularly with electrolytics as I remember. I wish I could find the
article...actually Electronics World rather than Elektor. At the time
I read it I was not aware of some of the myths he claimed to explode.
I remember wishing that he had included some lower frequency tests
because he said that electrolytic action is too slow to have an
appreciable effect at audio frequencies.

Non-polar electrolytics, used with zero bias, tested well, as did
back-to-back polar electrolytics.

The article may be available from EW, and I believe Cyril produced a
CDROM of his work with capacitors. Not for free though.

cheers, Ian

I just found a 1980 article "Picking Capacitors" by Walter Jung and
Richard March. Made scans and packed them in a zip.

http://www.tubesworld.com/inf/Picking_Capacitors.zip

Andre

[Arny Krueger]

"Bob Woodward" "Bob wrote in message

Ian Iveson wrote:
Contrary to what others here have said, distortion
increased, in general, both with signal *and* with
bias. I notice Bob Quintal says "If you keep the DC
Bias high wrt the AC component there isn't a
problem.", which sounds like a contrary view, although
not necessarily. I wonder what the evidence is?

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having
the cap inversely polarized for half periods of the
audio signal.

Particularly with electrolytics as I remember. I wish I
could find the article...actually Electronics World
rather than Elektor. At the time I read it I was not
aware of some of the myths he claimed to explode. I
remember wishing that he had included some lower
frequency tests because he said that electrolytic action
is too slow to have an appreciable effect at audio
frequencies. Non-polar electrolytics, used with zero bias, tested
well, as did back-to-back polar electrolytics.

The article may be available from EW, and I believe
Cyril produced a CDROM of his work with capacitors. Not
for free though.

I just found a 1980 article "Picking Capacitors" by
Walter Jung and Richard March. Made scans and packed
them in a zip.
http://www.tubesworld.com/inf/Picking_Capacitors.zip

Too bad you didn't search for the article using google. Here's one of many
posts of it:

http://www.reliablecapacitors.com/pickcap.htm

BTW, the article is not entirely composed of generally agreed-upon truth.

[dizzy]

Ian Iveson wrote:

Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal
says "If you keep the DC Bias high wrt the AC component there isn't
a problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having the cap
inversely polarized for half periods of the audio signal.

Particularly with electrolytics as I remember. I wish I could find the
article...actually Electronics World rather than Elektor. At the time
I read it I was not aware of some of the myths he claimed to explode.
I remember wishing that he had included some lower frequency tests
because he said that electrolytic action is too slow to have an
appreciable effect at audio frequencies.

I know that on Doug Self's designs, he's got electrolytics all over
the place to block DC. There's no biasing to keep them from becoming
reverse-biased by the AC signal. He never mentions how and why that's
okay.

[Bob Quintal]

"Ian Iveson" wrote in
. uk:

Bob Woodward wrote:

Do you remember the component values you tested were ? I
assume they were not NPO or COG ceramics as these behave
essentially like a film capacitors and should be
"distortionless".

As blameless as polypropylene, according to extensive tests
published in Elektor by Cyril Bateman.

From memory, other ceramics range around -60 to -80dB
distortion with 1V at 1kHz across them. Common polar
electrolytics are better, but poor examples not by much.

Films ranged from around -90 to being unmeasurable with a
floor at -120dB. Order was as you might expect, with
polypropylene the lowest.

Contrary to what others here have said, distortion increased,
in general, both with signal *and* with bias. I notice Bob
Quintal says "If you keep the DC Bias high wrt the AC
component there isn't a problem.", which sounds like a
contrary view, although not necessarily. I wonder what the
evidence is?

cheers, Ian

From what I understand of the distortion creation mechanism, the
increase in voltage causes a change in the capacitance. This
change is most pronounced at low voltages especially for Y5V
caps. Since without bias, the change in voltage is equal to the
peak-peak voltage, that represents a large change in the
impedance through the cap.

The bias reduces the percentage change significantly.

Refer to figure 2 on page 3 of the link I posted,
http://www.avxcorp.com/docs/techinfo/mlc-tant.pdf

--
Bob Quintal

PA is y I've altered my email address.

--
Posted via a free Usenet account from http://www.teranews.com

[Eeyore]

dizzy wrote:

Ian Iveson wrote:

Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal
says "If you keep the DC Bias high wrt the AC component there isn't
a problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having the cap
inversely polarized for half periods of the audio signal.

Particularly with electrolytics as I remember. I wish I could find the
article...actually Electronics World rather than Elektor. At the time
I read it I was not aware of some of the myths he claimed to explode.
I remember wishing that he had included some lower frequency tests
because he said that electrolytic action is too slow to have an
appreciable effect at audio frequencies.

I know that on Doug Self's designs, he's got electrolytics all over
the place to block DC. There's no biasing to keep them from becoming
reverse-biased by the AC signal. He never mentions how and why that's
okay.

That's simple.

Electrolytics don't become non-linear until the reverse voltage exceeds ~ 100mV.
Since coupling caps don't have very much AC voltage across them it's easy to
ensure they operate in this situation. The non-linearity seems to be the result
of the cap becoming an electrolytic recitifier AIUI and you have to exceed the
'forward voltage' i.e 100mV for that to occur.

To make sure simply this never happens use large value electrolytics, say 10 -22
uF for inputs and 100- 200 uF for outputs. It works a treat. It also ensures
very low phase shift at LF as a bonus.

Graham

[Eeyore]

Bob Quintal wrote:

From what I understand of the distortion creation mechanism, the
increase in voltage causes a change in the capacitance.

The dielectric constant isn't errrr... constant !


This change is most pronounced at low voltages especially for Y5V
caps. Since without bias, the change in voltage is equal to the
peak-peak voltage, that represents a large change in the
impedance through the cap.

The bias reduces the percentage change significantly.

No. Not so. Coupling caps *do not* see the signal voltage across them. They
wouldn't be doing their job if they did.

Graham

[Don Pearce]

On Tue, 17 Jul 2007 23:56:16 GMT, dizzy wrote:

Ian Iveson wrote:

Contrary to what others here have said, distortion increased, in
general, both with signal *and* with bias. I notice Bob Quintal
says "If you keep the DC Bias high wrt the AC component there isn't
a problem.", which sounds like a contrary view, although not
necessarily. I wonder what the evidence is?

Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having the cap
inversely polarized for half periods of the audio signal.

Particularly with electrolytics as I remember. I wish I could find the
article...actually Electronics World rather than Elektor. At the time
I read it I was not aware of some of the myths he claimed to explode.
I remember wishing that he had included some lower frequency tests
because he said that electrolytic action is too slow to have an
appreciable effect at audio frequencies.

I know that on Doug Self's designs, he's got electrolytics all over
the place to block DC. There's no biasing to keep them from becoming
reverse-biased by the AC signal. He never mentions how and why that's
okay.

Coupling caps do not become reverse biased by AC signals. Whatever
voltage is across them when the equipment is switched on remains
across them in the presence of an audio signal. This business of the
being reverse biased during negative-going half cycles (or whatever)
is a myth. Just examine how a circuit operates.

Of course extremely low frequency signals will reverse bias small
coupling caps, but that isn't what we are talking about here, is it?

d

--
Pearce Consulting
http://www.pearce.uk.com

[Bob Woodward]

Arny Krueger wrote:
"Bob Woodward" "Bob wrote in message

Ian Iveson wrote:
Contrary to what others here have said, distortion
increased, in general, both with signal *and* with
bias. I notice Bob Quintal says "If you keep the DC
Bias high wrt the AC component there isn't a
problem.", which sounds like a contrary view, although
not necessarily. I wonder what the evidence is?
Was that related to film capacitors or electrolytics ?
When electrolytics, applying DC-bias would avoid having
the cap inversely polarized for half periods of the
audio signal.
Particularly with electrolytics as I remember. I wish I
could find the article...actually Electronics World
rather than Elektor. At the time I read it I was not
aware of some of the myths he claimed to explode. I
remember wishing that he had included some lower
frequency tests because he said that electrolytic action
is too slow to have an appreciable effect at audio
frequencies. Non-polar electrolytics, used with zero bias, tested
well, as did back-to-back polar electrolytics.

The article may be available from EW, and I believe
Cyril produced a CDROM of his work with capacitors. Not
for free though.

I just found a 1980 article "Picking Capacitors" by
Walter Jung and Richard March. Made scans and packed
them in a zip.
http://www.tubesworld.com/inf/Picking_Capacitors.zip

Too bad you didn't search for the article using google. Here's one of many
posts of it:

uumph ...

http://www.reliablecapacitors.com/pickcap.htm

BTW, the article is not entirely composed of generally agreed-upon truth.

My reflex reaction was caused by the fact that this article
was what made me start measuring caps.

When discussing the influence of caps, keep in mind that most
musicproductions, recordings or live performance, are made using
large mixing-consoles and outboard equipment that is loaded with
caps in the signal-path. Mostly electrolytics, no bias.

Once heard Rupert Neve during a lecturer claim he did apply about 1V
bias to make all electrolytics beeing right polarized for most of
the signal ( amplitude ).

Then I did a complete overhaul of a large console ( DDA DCM232 )
( http://www.ddaconsoles.com/pdf/schem...schematics.pdf )
that was 12 years old.
measuring frequency response and distortion of the single channel-strips
showed a disaster. Still it has been in use till days before it was
taken to my works.

Roll-off starting at 200Hz down giving -10dB at 20Hz, or
Distortion over 1%, or
High-cut at 6kHz, or combination hereof.
All these faults originated from electrolytics. Dried, leaked, exploded.
So finally decided to recap the whole console.

My conclusion: unless special precautions have been taken, it is wise
to check behavior every 5 years and when these faults occur, recap.
In new design; go DC, use servo circuit to correct offset in DC,
use film capacitors, use ( good quality ) electrolytics with bias,
use electrolytics and replace them every 1000 days of use.

So to the OP. how much DC must you block ? is there space on another
spot of the pcb? Do you have an unused op-amp in a dual or quad package
to make a servo-circuit? Or can you accept the DC in the first stage and
block it somewhere further in the circuit?

Andre

[Eeyore]

Bob Woodward wrote:

Once heard Rupert Neve during a lecturer claim he did apply about 1V
bias to make all electrolytics beeing right polarized for most of
the signal ( amplitude ).

Never seen any Neve gear that did that. All the dual supply op-amp kit they've
made that I've seen (quite a bit) uses zero bias electrolytic coupling caps.
Their single supply stuff if course DC biased.

There is one company I know of that used that technique. It was Midas (SR
consoles) in the 70s / early 80s and they did it because they were using
tantalum bead electrolytics that rectify more easily AIUI than aluminium ones.

Graham

[Eeyore]

Bob Woodward wrote:

Then I did a complete overhaul of a large console ( DDA DCM232 )
( http://www.ddaconsoles.com/pdf/schem...schematics.pdf )
that was 12 years old.
measuring frequency response and distortion of the single channel-strips
showed a disaster. Still it has been in use till days before it was
taken to my works.

Roll-off starting at 200Hz down giving -10dB at 20Hz, or
Distortion over 1%, or
High-cut at 6kHz, or combination hereof.
All these faults originated from electrolytics. Dried, leaked, exploded.
So finally decided to recap the whole console.

That's because they dried out with age. The absence of DC bias isn't a factor
there AIUI.

Graham

[Eeyore]

Bob Woodward wrote:

In new design; go DC, use servo circuit to correct offset in DC,

DC servos everywhere ? You have to be joking.


use film capacitors, use ( good quality ) electrolytics with bias,
use electrolytics and replace them every 1000 days of use.

Why not just use 105C electrolytics ? They'll take 4 times as long (compared to
standard 85C types) typically to dry out from elevated temperatures.

It does get warm in busy channel strips. Neve V series was a case in point. A 48
channel console took about 3kW of DC power.

Graham

[Arny Krueger]

"Bob Woodward" "Bob wrote in message

Arny Krueger wrote:

http://www.reliablecapacitors.com/pickcap.htm

BTW, the article is not entirely composed of generally
agreed-upon truth.

My reflex reaction was caused by the fact that this
article was what made me start measuring caps.

I measured a few caps after reading it, once I stopped laughing. More to the
point, I actually listened to some capacitors, and developed more evidence
for laughing.

When discussing the influence of caps, keep in mind that
most musicproductions, recordings or live performance,
are made using large mixing-consoles and outboard equipment that is
loaded with caps in the signal-path. Mostly electrolytics, no bias.

That's because despite all the hubub, most equipment that uses caps is
properly designed.

Once heard Rupert Neve during a lecturer claim he did
apply about 1V bias to make all electrolytics beeing
right polarized for most of the signal ( amplitude ).

Ole Rupert has said a lot of things, some ordinary, some wonderful, and some
pretty strange.


Then I did a complete overhaul of a large console ( DDA DCM232 ) (
http://www.ddaconsoles.com/pdf/schem...schematics.pdf
) that was 12 years old.
measuring frequency response and distortion of the single
channel-strips showed a disaster. Still it has been in
use till days before it was taken to my works.

Roll-off starting at 200Hz down giving -10dB at 20Hz, or
Distortion over 1%, or
High-cut at 6kHz, or combination hereof.
All these faults originated from electrolytics. Dried,
leaked, exploded. So finally decided to recap the whole
console.

A well-known exposure with older equipment, one that has zilch to do with
the topics in the Marsh-Jung article you cited.

My conclusion: unless special precautions have been
taken, it is wise to check behavior every 5 years and
when these faults occur, recap.

That makes as little sense as recommending a complete overhaul of every car
engine after 3-5 years. Most people don't do that. They first check the
condition of the engine to see if its performance warrants the overhaul.

[Bob Quintal]

Eeyore wrote in
:



Bob Quintal wrote:

From what I understand of the distortion creation mechanism,
the increase in voltage causes a change in the capacitance.

The dielectric constant isn't errrr... constant !


This change is most pronounced at low voltages especially for
Y5V caps. Since without bias, the change in voltage is equal
to the peak-peak voltage, that represents a large change in
the impedance through the cap.

The bias reduces the percentage change significantly.

No. Not so. Coupling caps *do not* see the signal voltage
across them. They wouldn't be doing their job if they did.

Graham

Every capacitor has some series impedance. In a properly
designed circuit, a coupling cap should be chosen to have a very
low impedance compared to the rest of the circuit, In the case
of a ceramic cap, the impedance may be 1% of the load impedance
it sees, but if the capacitance changes, that impedance will
change. All I'm saying is that if it changes from 1% when the
signal is at negative peak to 2%, when the signal is at positive
peak, that will introduce a certain amount of distortion.

--
Bob Quintal

PA is y I've altered my email address.

--
Posted via a free Usenet account from http://www.teranews.com

[Eeyore]

Bob Quintal wrote:

Eeyore wrote
Bob Quintal wrote:

From what I understand of the distortion creation mechanism,
the increase in voltage causes a change in the capacitance.

The dielectric constant isn't errrr... constant !


This change is most pronounced at low voltages especially for
Y5V caps. Since without bias, the change in voltage is equal
to the peak-peak voltage, that represents a large change in
the impedance through the cap.

The bias reduces the percentage change significantly.

No. Not so. Coupling caps *do not* see the signal voltage
across them. They wouldn't be doing their job if they did.

Every capacitor has some series impedance. In a properly
designed circuit, a coupling cap should be chosen to have a very
low impedance compared to the rest of the circuit, In the case
of a ceramic cap, the impedance may be 1% of the load impedance
it sees, but if the capacitance changes, that impedance will
change. All I'm saying is that if it changes from 1% when the
signal is at negative peak to 2%, when the signal is at positive
peak, that will introduce a certain amount of distortion.

Let's suppose the signal voltage is 1V. 1% of that (per your example) of 1V is
10mV. That's the kind of signal that'll be *across* the cap.

Your statement "the change in voltage is equal to the peak-peak voltage" is
simply incorrect. Perhaps you could simply have accepted that without seeming to
obfuscate ?

FYI I typically use a 10uF (electrolytic) coupling cap for inputs. That has an
impedance of -j795 ohms @ 20 Hz. With a 1V rms signal and a typical 10k input
impedance the voltage *across* the cap in those conditions is ~ 30mV. In this
case that 30mV is well below the threshold at which electrolytic rectification
takes place and the signal is not distorted. I have performed measurements to
confirm this btw.


Graham

[Bob Quintal]

Eeyore wrote in
:



Bob Quintal wrote:

Eeyore wrote
Bob Quintal wrote:

From what I understand of the distortion creation
mechanism, the increase in voltage causes a change in the
capacitance.

The dielectric constant isn't errrr... constant !


This change is most pronounced at low voltages especially
for Y5V caps. Since without bias, the change in voltage is
equal to the peak-peak voltage, that represents a large
change in the impedance through the cap.

The bias reduces the percentage change significantly.

No. Not so. Coupling caps *do not* see the signal voltage
across them. They wouldn't be doing their job if they did.

Every capacitor has some series impedance. In a properly
designed circuit, a coupling cap should be chosen to have a
very low impedance compared to the rest of the circuit, In
the case of a ceramic cap, the impedance may be 1% of the
load impedance it sees, but if the capacitance changes, that
impedance will change. All I'm saying is that if it changes
from 1% when the signal is at negative peak to 2%, when the
signal is at positive peak, that will introduce a certain
amount of distortion.

Let's suppose the signal voltage is 1V. 1% of that (per your
example) of 1V is 10mV. That's the kind of signal that'll be
*across* the cap.

Your statement "the change in voltage is equal to the
peak-peak voltage" is simply incorrect. Perhaps you could
simply have accepted that without seeming to obfuscate ?

FYI I typically use a 10uF (electrolytic) coupling cap for
inputs. That has an impedance of -j795 ohms @ 20 Hz. With a
1V rms signal and a typical 10k input impedance the voltage
*across* the cap in those conditions is ~ 30mV. In this case
that 30mV is well below the threshold at which electrolytic
rectification takes place and the signal is not distorted. I
have performed measurements to confirm this btw.


Graham

Graham, The thread is about ceramic capacitors used for
coupling. I usually use a 6.8mF cap for coupling a microphone
input,and 22mF for interstage . It's a fact that ceramic caps do
have a capacitance/voltage sensitivity. If the capacitance
changes then the series resistance (the real part of impedance)
will change too. It doesn't happen with electrolytics, so you
don't have to worry unless you change your designs. The Original
Poster was asking about cercaps because of their smaller
physical dimensions per unit capacitance. Using those as
coupling caps requires additional considerations be entertained
over yours. Not everybody remains stuck with a 1960s mentality.


--
Bob Quintal

PA is y I've altered my email address.

--
Posted via a free Usenet account from http://www.teranews.com

[Eeyore]

Bob Quintal wrote:

Graham, The thread is about ceramic capacitors used for
coupling.

Actually it's about coupling caps and whether ceramics are suitable.

I think they're not suitable (in the types available for the required value) and
I'm recommending an alumimium electrolytic instead.


I usually use a 6.8mF cap for coupling a microphone input,

For a low impedance mic input ????

m means milli btw. I assume you don't mean 6800uF?


and 22mF for interstage .

A larger value for interstage ? That's curious.


It's a fact that ceramic caps do have a capacitance/voltage sensitivity.

Medium and high K ceramic dielectrics do. Low K ceramics don't exhibit this
problem.


If the capacitance changes then the series resistance (the real part of
impedance)
will change too.

No, the complex part changes (on account of K changing).


It doesn't happen with electrolytics,

That's exactly my point. Electrolytics exhibit a different 'problem' that's very
easily completely avoided.


so you don't have to worry unless you change your designs. The Original
Poster was asking about cercaps because of their smaller
physical dimensions per unit capacitance.

It's got to be pretty desperate if there's no room for a 10uF electrolytic.


Using those as coupling caps requires additional considerations be entertained

over yours. Not everybody remains stuck with a 1960s mentality.

What 1960s mentality ?

Graham

[Mr.T]

"Arny Krueger" wrote in message
...
My conclusion: unless special precautions have been
taken, it is wise to check behavior every 5 years and
when these faults occur, recap.

That makes as little sense as recommending a complete overhaul of every
car
engine after 3-5 years. Most people don't do that. They first check the
condition of the engine to see if its performance warrants the overhaul.

And I agree with you both. As he said "check behaviour" and "when faults
occur, recap"
Similar to doing a compression test on a car engine and when crook, rebuild.

However in both cases it is usually obvious something is wrong, so you do
more thorough checking whenever necessary, not at any particular time
interval IMO.

MrT.

[Bob Woodward]

Mr.T wrote:
"Arny Krueger" wrote in message
...
My conclusion: unless special precautions have been
taken, it is wise to check behavior every 5 years and
when these faults occur, recap.
That makes as little sense as recommending a complete overhaul of every
car
engine after 3-5 years. Most people don't do that. They first check the
condition of the engine to see if its performance warrants the overhaul.

And I agree with you both. As he said "check behaviour" and "when faults
occur, recap"
Similar to doing a compression test on a car engine and when crook, rebuild.

However in both cases it is usually obvious something is wrong, so you do
more thorough checking whenever necessary, not at any particular time
interval IMO.

MrT.

Very capable and respected producers have been working on that console
for twelve years. Wear of caps, and so sound becoming worse goes slow.
In one channel that looses the low frequencies you plug in a microphone
for vocals and it appears to be ok.
Eventually they find one or two channels to be "defective" and don't use
these for a while, stick some tape over the fader and carry on working.

This was at a radio and tv station. The producers and programtechnicians
haven't bought the console, they don't own it, they don't care.
I care, I checked it and proposed a complete overhaul. And so did.

Found that half of all electrolytics needed to be replaced. Concluded
that the other half would be in that state within a few years and so
replaced all electrolytics.

Did replace electrolytics,... no, let's say; _needed_ to replace
electrolytics in various other apparatus like distribution amplifiers
( 1-in 6-out ) and all kind of outboard.

Therefore I propose to check every now and then.

When _MY_ car makes a strange noise or has a bit less compression,
I'll notice.

Andre

[Mr.T]

"Bob Woodward" "Bob wrote in message
...
Very capable and respected producers have been working on that console
for twelve years.

Yes, I always find that a laugh too, unless they complained, nothing was
done, and they simply gave up.
That's all too common IME.

Wear of caps, and so sound becoming worse goes slow.

And these capable producers have NEVER listened to anything else in all that
time?
I'll bet most could tell something was wrong.

In one channel that looses the low frequencies you plug in a microphone
for vocals and it appears to be ok.
Eventually they find one or two channels to be "defective" and don't use
these for a while, stick some tape over the fader and carry on working.

Well if you can accept the loss of those channels, it's not a problem then.
And they obviously DID notice a problem if they went to the trouble of
marking it (for the technician to fix later).
Of course they carried on working for that session, I do exactly the same if
I can still work around the problem with spare channels.

This was at a radio and tv station. The producers and programtechnicians
haven't bought the console, they don't own it, they don't care.

I would have thought the technicians were being paid to care. Kick them up
the arse!

I care, I checked it and proposed a complete overhaul. And so did.
Found that half of all electrolytics needed to be replaced. Concluded
that the other half would be in that state within a few years and so
replaced all electrolytics.

A reasonable course of action.

Therefore I propose to check every now and then.

Won't hurt.

When _MY_ car makes a strange noise or has a bit less compression,
I'll notice.

And you don't think you will notice the audio problems you mention?
In fact I'm more likely to notice a small loss of bass or treble than a
small loss of compression myself.
If the problems are big enough, anyone will notice in either case.

MrT.

[Adam S]

Eeyore wrote:

It doesn't happen with electrolytics,

That's exactly my point. Electrolytics exhibit a different 'problem' that's very
easily completely avoided.


so you don't have to worry unless you change your designs. The Original
Poster was asking about cercaps because of their smaller
physical dimensions per unit capacitance.

It's got to be pretty desperate if there's no room for a 10uF electrolytic.


And where will you put a 10uF electro in a portable MP3 player for
example ? The answer is you don't. you stick with DC coupling.

In my case I don't have such extreme size constraints. I decided to fit
in some conventional through hole 63V 100nF polyester caps and forget
about using X7R 0805 100nF 50V caps. This is for DC blocking the inputs
from the outside world. With a 470kohm load resistance that'll form a
3.1Hz cutoff HPF. The DC offset due to input bias current across a 470k
ohm resistor is still well under the intrinsic DC offset voltage of the
op-amp. The remainder of the signal chain is DC coupled right through to
the outputs. At very worst case, one can use a servo amplifier to
correct DC offsets. And probably cheaper too than a bunch of largish
capacitors scattered throughout.

[Eeyore]

Adam S wrote:

Eeyore wrote:

It doesn't happen with electrolytics,

That's exactly my point. Electrolytics exhibit a different 'problem' that's very
easily completely avoided.


so you don't have to worry unless you change your designs. The Original
Poster was asking about cercaps because of their smaller
physical dimensions per unit capacitance.

It's got to be pretty desperate if there's no room for a 10uF electrolytic.

And where will you put a 10uF electro in a portable MP3 player for
example ?

You get a miniature type.


The answer is you don't. you stick with DC coupling

I very much doubt that but there's no need to scatter them about like confetti.

Graham

[Arny Krueger]

"Adam S" not.valid@nosuchaddress wrote in message
u
Eeyore wrote:

It doesn't happen with electrolytics,

That's exactly my point. Electrolytics exhibit a
different 'problem' that's very easily completely
avoided.
so you don't have to worry unless you change your
designs. The Original Poster was asking about cercaps
because of their smaller physical dimensions per unit capacitance.

It's got to be pretty desperate if there's no room for a
10uF electrolytic.

And where will you put a 10uF electro in a portable MP3
player for example ?

Near the headphone jack, and across the battery, for starts.

Since most MP3 players use power with just one polarity, there are generally
at least output coupling caps. Typcially they are more like 100 uF, than 10
uF.

A standard 10 uF, 10 volt SMT electrolytic is about 3 x 5.5 x 3.3 mm or 4
x 5.5 x 4.3 mm, according to
http://panasonic.com/industrial/comp...ABA0000CE2.pdf . That's
pretty small!

[Adam S]

Arny Krueger wrote:

And where will you put a 10uF electro in a portable MP3
player for example ?

Near the headphone jack, and across the battery, for starts.

Since most MP3 players use power with just one polarity, there are generally
at least output coupling caps. Typcially they are more like 100 uF, than 10
uF.

A standard 10 uF, 10 volt SMT electrolytic is about 3 x 5.5 x 3.3 mm or 4
x 5.5 x 4.3 mm, according to
http://panasonic.com/industrial/comp...ABA0000CE2.pdf . That's
pretty small!


For the battery you use a 10uF 6.3V 0805 ceramic which is cheaper,
smaller and more reliable. But why bother for the headphones when
National, Maxim and others have offered low cost capacitorless headphone
driver ICs for a many years now. Adding a 1/2 Vcc virtual ground doesn't
take up much more silicon area.

http://www.national.com/pf/LM/LM4924.html
http://www.national.com/pf/LM/LM4982.html
http://www.national.com/pf/LM/LM4985.html
....

http://datasheets.maxim-ic.com/en/ds/MAX9725.pdf
http://datasheets.maxim-ic.com/en/ds...2-MAX9722B.pdf
....


Adam

[Arny Krueger]

"Adam S" not.valid@nosuchaddress wrote in message
u
Arny Krueger wrote:

And where will you put a 10uF electro in a portable MP3
player for example ?

Near the headphone jack, and across the battery, for
starts. Since most MP3 players use power with just one polarity,
there are generally at least output coupling caps.
Typcially they are more like 100 uF, than 10 uF.

A standard 10 uF, 10 volt SMT electrolytic is about 3 x
5.5 x 3.3 mm or 4 x 5.5 x 4.3 mm, according to
http://panasonic.com/industrial/comp...ABA0000CE2.pdf
. That's pretty small!


For the battery you use a 10uF 6.3V 0805 ceramic which is
cheaper, smaller and more reliable. But why bother for
the headphones when National, Maxim and others have
offered low cost capacitorless headphone driver ICs for a
many years now. Adding a 1/2 Vcc virtual ground doesn't
take up much more silicon area.
http://www.national.com/pf/LM/LM4924.html
http://www.national.com/pf/LM/LM4982.html
http://www.national.com/pf/LM/LM4985.html
...

http://datasheets.maxim-ic.com/en/ds/MAX9725.pdf
http://datasheets.maxim-ic.com/en/ds...2-MAX9722B.pdf

The MAX 4410 caught my eye. Thanks for the technology update!