power audio op amp amplifier capacitor

danadak

Joined Mar 10, 2018
4,057
Unloaded, the amplifier and capacitor will produce almost no distortion. When loaded the capacitor reduces the levels of low frequencies and when its capacitance changes with the momentary varying voltage across the capacitor then even harmonics are produced which is distortion.

Odd harmonics would be produced if the capacitor had AC only with no DC across it.
@Audioguru, can you point me to any papers, ap notes, rigorous investigations where this
even / odd harmonic generation by the cap has been measured/observed ?


Regards, Dana.
 

Audioguru

Joined Dec 20, 2007
11,248
is it better to use an oamp with a pot to vary the audio signal voltage by varying gain or vary the voltage using an ordinary pot straight into the tda2030?

The opamp method prevents the audio signal from going thru the pot.
The TDA2030 and the newer TDA2030A are obsolete and are not made anymore. They are power audio amplifiers, not opamps.
The datasheet for the TDA2030A shows external resistors producing a gain of 26dB which is 20 times. It oscillates at a very high frequency if the gain is reduced but they don't say the minimum allowed gain for no oscillation. The datasheet for the LM386 power amplifier says its minimum allowed gain is more than 9 times, a gain of 20 times or more is recommended. Most ordinary opamps are designed not to oscillate with any amount of gain.

Audio amplifiers use a logarithmic volume control at the input. The volume control must not have any DC in it (so it has coupling capacitors) and forms a voltage divider with the input signal. At maximum, the volume control connects the amplifier input directly to the input signal and at minimum the volume control connects the amplifier input to the circuit ground.
 

Audioguru

Joined Dec 20, 2007
11,248
@Audioguru, can you point me to any papers, ap notes, rigorous investigations where this
even / odd harmonic generation by the cap has been measured/observed ?


Regards, Dana.
I saw no tests and I made no tests myself. Make your own test. Experts say that an electrolytic output capacitor causes low frequency distortion and I believe it. Most audio amplifiers today use a dual polarity supply so that they do not have an output capacitor.
 

danadak

Joined Mar 10, 2018
4,057
Seems unfortunate no key capacitor manufacturer has done measurements
of this phenomena and explained physically how this is happening. Maybe
someone else on the forum can shed some light on this ?

This would have profound impact on attempts to do hi resolution DC voltage
measurements with advanced A/Ds as well.

Regards, Dana.
 
Last edited:

wayneh

Joined Sep 9, 2010
17,496
Seems unfortunate no key capacitor manufacturer has done measurements
of this phenomena and explained physically how this is happening.
Absence of evidence is not evidence of absence. Just because we cannot find the reports doesn’t mean it’s not out there waiting for us to find it.
 

danadak

Joined Mar 10, 2018
4,057
Totally concur. So maybe a couple of emails out to key C manufacturers might be in
order.

We as a community need to know if electrolytic C's should be considered an active device,
eg C = f( V ), and not passive, as conjectured.

Regards, Dana.
 

wayneh

Joined Sep 9, 2010
17,496
Totally concur. So maybe a couple of emails out to key C manufacturers might be in
order.

We as a community need to know if electrolytic C's should be considered an active device,
eg C = f( V ), and not passive, as conjectured.

Regards, Dana.
I imagine they would direct you to the literature. You can find yourself bedtime reading for months just by searching on "capacitor audio amplifier distortion" or similar. Here's just one of the many hits that pop up:
http://stephan.win31.de/capdist.htm

I don't think there are any untrodden grounds in the audio electronics field.
 

Dave Lowther

Joined Sep 8, 2016
224
Totally concur. So maybe a couple of emails out to key C manufacturers might be in
order.

We as a community need to know if electrolytic C's should be considered an active device,
eg C = f( V ), and not passive, as conjectured.

Regards, Dana.
Here's a series of detailed articles by Cyril Bateman on the Linear Audio site.
Douglas Self has quite a lot to say on this topic in his book Small Signal Audio Design Chapter 2 “Capacitor non-linearity examined".
Best regards
Dave
 

Parkera

Joined May 3, 2016
106
Many article on the web say that electrolytic coupling capacitors with a fairly high voltage swing across them (electrolytic types at the output of an audio amplifier, at low frequencies) cause distortion. An example is here: http://sound.whsites.net/articles/coupling-caps.htm
For what it is worth, the electrolytics in the power supply are also in the audio chain, but nobody seems to complain about those. If you want convincing, just follow the CURRENT path from the speaker output ALL the way around; that power has to come from somewhere.
 

LowQCab

Joined Nov 6, 2012
4,023
I really don't understand how any of you guys can make a response without
knowing anything about the expected application or what is being pursued.
IF........
You are concerned with "low frequency distortion"......
Why are you using anything with an output capacitor in the first place ?
Second, "low frequencies" at very low power, are not going to be very useful,
unless you're talking about headphones.
Third, no one has mentioned what this amp will be driving or possibly sensing.
Forth, if it will be driving a speaker, the ~10 to ~20% distortion created by
the speaker and it's cabinet, and the environment that it is in,
should be the first priority if you must have low distortion.
Fifth, IF,
if this project will be fine with a low power amp,
but you have actual known problems caused by capacitor distortion,
just use 2 amps in a bridged configuration with zero capacitors, ( DC-coupled ),
( you may still need the output compensating network (one on each amp),
(or across the load),
especially if the load is a speaker).
Sixth, the compensating network should use a non-inductive (carbon composition) resistor, in any case.
Seventh, IF,
If this is for reproducing music,
the major distortion problem is not going to be in the selection of components used in the amplifier.
Possibly, this entire post really has to do with someone learning the hard way,
why they are not getting the bass response they would like with their sound system.
If this is the case, you must start with understanding how sound reproduction can be accomplished,
in the anticipated environment,
( out-doors, auditorium, living room, bed room, head phones ),
and with what other mechanical constraints,
( size, cost, available power ),
may be imposed or desired.
The electrical (electronic) part of the equation is usually NOT the problem,
with the possible exception of very limited power being available.
Eighth,
Do you actually know that you have a problem with distortion caused by capacitor selection ?
Did you know that resistors can be a source of noise (hiss) ?
Why do you have any concerns about frequencies below ~30hz ??
Are you aware that, if you are inside of a building or home, that the space you are in
has more influence over bass response, and bass distortion, and low frequency noise,
than any other factor ?
.
So....
What, exactly, are you trying to accomplish ??
( this should have been explained in detail in the initially posted question ).
.
 

Parkera

Joined May 3, 2016
106
Unloaded, the amplifier and capacitor will produce almost no distortion. When loaded the capacitor reduces the levels of low frequencies and when its capacitance changes with the momentary varying voltage across the capacitor then even harmonics are produced which is distortion.

Odd harmonics would be produced if the capacitor had AC only with no DC across it.
Odd harmonics are also distortion and not desirable either.

Also, the output coupling capacitor ALWAYS has a DC voltage across it, but that DC voltage varies from a volt or two above zero to a volt or two below Vcc. Is this important? You would have to know how much the capacitance changes over the maximum voltage swing. From there you can calculate how it affects the low-frequency response. From there, you can make a judgement if that change is important or not.

If it only makes a 1% change at 20 Hz - it probably doesn't matter. If it makes a 50% change at 20 Hz it most certainly will matter. It is not JUST the change in capacitance, but the absolute capacitance change is important in any given circuit and application.
 

adx

Joined Dec 9, 2010
1
The level of distortion often complained about from electrolytic capacitors is relevant down in the 0.1% measurement region where people get bent out of shape about slight changes in THD of an amplifier. The cap distortion may only be 0.001% or unmeasurable.

The capacitance change of a small ceramic (small for the application) can be 80% of its value, which can cause enormous nonlinearity, and easily finds its way into the output as signal distortion.
 

danadak

Joined Mar 10, 2018
4,057
@DaveLowther, looks like a great ref.

Interesting the background needed to do a distortion analyzer design
to get at the low levels of distortion being measured.

Hopefully the Electronics World articles will comment on the stim source
distortion, as making measurements in the .001% area imply the source
has to have much lower distortion than DUT.

Thanks for the post.

Regards, Dana.
 

Audioguru

Joined Dec 20, 2007
11,248
The distortion caused by an electrolytic capacitor is only at low frequencies where the capacitor is reducing low frequencies. Its capacitance changes by the instantaneous DC voltage of the signal so that the top half of the waveform becomes larger or smaller than the lower half which is even harmonics distortion. If the capacitor value or the frequency is high enough then the distortion will be very low.
 

danadak

Joined Mar 10, 2018
4,057
The distortion caused by an electrolytic capacitor is only at low frequencies
If the phenomena is C = f( V ) then that distortion should be there at "higher" frequencies as well.
Not sure I understand why the Cap knows only to distort at low frequency.....?

top half of the waveform becomes larger or smaller than the lower half which is even harmonics distortion
Help me understand this, why a subset of a waveform produces even or odd harmonics ?


Regards, Dana.
 

Audioguru

Joined Dec 20, 2007
11,248
Assume that the input signal is a sinewave with no distortion. Then its top and bottom are not squashed equally which would produce odd harmonics distortion and the top half and bottom half have equal amplitudes so that there is no even harmonics distortion like an old vacuum tube.

But as the signal DC voltage swings up and down, the value of the output capacitor changes. On the top or bottom part of the swing when the capacitance is high then that part of the output sinewave has a high amplitude at a low frequency. At the other half of the output voltage swing when the capacitance is low then the highpass filter action on low frequencies reduces the amplitude. The result is even harmonics distortion.

The problem can be reduced if you use a whopping huge capacitor that passes earthquake frequencies and takes all day to charge.

Somebody said that the power supply filter capacitors also cause the same distortion but I disagree since an amplifier has excellent power supply interference rejection.
 
This looks like a project driven by some "golden ears"-society with a strong believe in the appearance of such distortion.

One can wonder why there is no instruments available for meassuring theese practically non-existing phenomena.

The practical use of an instrument to meassure some possible distorsion of an level lower than 0,0000001% in an amplifier wich already have far more detectable distortioncomponents in the area of 0,001% area is more than questionable.

I know no such instrument ever been buildt.

Wonder why distorsions of theese types only appear when the listener know what capasitor is in the signalpath, but always fails to detect any distorsion beyound statistic criterias when the capasitortype is unknown?
 

Audioguru

Joined Dec 20, 2007
11,248
Fletcher-Munson curves show that human hearing is not sensitive to low levels of low frequencies. Also, since even harmonics are exactly one octave higher or lower than each other, many people do not say that even harmonics are distortion.
 
Top