Since a capacitor is made of a dielectric and plates and there are several types of capacitors differentiated by their dielectric, I would like to ask which types are suitable for audio signal processing? I do not wish that this thread transforms into claims not based on the science of dielectrics and capacitor construction.
Oh boy, you just know this thread will end the way you do not wish for...
I'd say (almost) all of them.
Some time ago I experimented swapping out electrolytics in the signal path with MKTs and found that I personally couldn't hear any difference. Sure dielecttrics and construction are different, and they will probably even measure differently, I just don't hear it.
On the more practical side, I regard MKTs (and similar types) as fit and forget items whereas electrolytics have a far more finite lifespan, even if not abused by high ripple current or high temps.
I'd say (almost) all of them.
Some time ago I experimented swapping out electrolytics in the signal path with MKTs and found that I personally couldn't hear any difference. Sure dielecttrics and construction are different, and they will probably even measure differently, I just don't hear it.
On the more practical side, I regard MKTs (and similar types) as fit and forget items whereas electrolytics have a far more finite lifespan, even if not abused by high ripple current or high temps.
My capacitor uses are:
Polarized electrolytic: power supply filter.
Non-polarized electrolytic: DC feedback.
Ceramic C0G: frequency compensation, RF filter.
Ceramic X7R: power supply decoupling, RF bypass for electrolytics.
Tantalum: signal coupling, power supply decoupling.
Polypropylene: equalization.
X-capacitor: AC line filter.
Ed
Polarized electrolytic: power supply filter.
Non-polarized electrolytic: DC feedback.
Ceramic C0G: frequency compensation, RF filter.
Ceramic X7R: power supply decoupling, RF bypass for electrolytics.
Tantalum: signal coupling, power supply decoupling.
Polypropylene: equalization.
X-capacitor: AC line filter.
Ed
In the world of 'Audio', you can use absolutely every type of capacitor ever made.
So by 'Audio' you mean;
am tuner, fm tuner, phonograph riaa stage, pre pre-amp stage, pre-amp, power amp, active speakers, passive speakers, headphone amps, active crossovers, passive crossovers, eq stages, cd players, dacs, power supply stages for or in any of the before mentioned items.
There's other 'audio' items I will have missed, but 'Audio' is not one thing, so obviously what parts go into any one item, and also within the various sections of that one item, renders your simple question rather impossible to answer.
So by 'Audio' you mean;
am tuner, fm tuner, phonograph riaa stage, pre pre-amp stage, pre-amp, power amp, active speakers, passive speakers, headphone amps, active crossovers, passive crossovers, eq stages, cd players, dacs, power supply stages for or in any of the before mentioned items.
There's other 'audio' items I will have missed, but 'Audio' is not one thing, so obviously what parts go into any one item, and also within the various sections of that one item, renders your simple question rather impossible to answer.
When you bypass bigger poly xover caps with a much smaller quality mkp caps by all sense it should not really change sound in any way. After all what can a small change in capacitance do to sound? But we live in a less than ideal world. Some caps can be not so good. As noticed there was a positive change even to the suspecting in my experiment. Since then I just bypass small mkp/mkt in xovers. They are cheap and should not hurt if anything.
Here's the classic work on distortion in different capacitor types:
https://www.collinsaudio.com/Prosound_Workshop/Capacitor-Sound.pdf
In short ceramic NP0 (aka C0G) are fine, other ceramics definitely not(*). Polystyrene (PS) and polypropylene (PP) are excellent for film caps, but if you are using SMT neither are available (they melt). Polyphenylene sulfide (PPS) is a good contender for SMT. Teflon (PTFE) is also excellent but hard to source.
Electrolytics don't have stable values of capacitance so only get used for DC-blocking and decoupling, no use in an active filter. They are fairly distorty, but bipolar electrolytics are somewhat better.
(*) they even have cross over distortion, not just massive voltage & temperature coefficients, and are microphonic.
https://www.collinsaudio.com/Prosound_Workshop/Capacitor-Sound.pdf
In short ceramic NP0 (aka C0G) are fine, other ceramics definitely not(*). Polystyrene (PS) and polypropylene (PP) are excellent for film caps, but if you are using SMT neither are available (they melt). Polyphenylene sulfide (PPS) is a good contender for SMT. Teflon (PTFE) is also excellent but hard to source.
Electrolytics don't have stable values of capacitance so only get used for DC-blocking and decoupling, no use in an active filter. They are fairly distorty, but bipolar electrolytics are somewhat better.
(*) they even have cross over distortion, not just massive voltage & temperature coefficients, and are microphonic.
Last edited:
I suspect that electrolytic capacitors may be more noisy due to having an electrolyte in which ions move. Ions are much heavier than electrons which may affect their ability to cancel voltages formed by their random motion. I am not certain of the last statement, that is why I created this thread. In other words are normal electrolytic capacitors suitable for audio signal processing?
When there is 0 V across them, they have to generate thermal noise in accordance with their effective series resistance, otherwise they would violate the second law of thermodynamics. You could then connect one to a normal resistor and have the electrolytic capacitor supply more or less noise power to the resistor than it gets back.
Their effective series resistance is definitely higher than that of film capacitors of the same value. See the tan(delta) numbers and graphs in their datasheets, the ESR numbers are usually only given at one or two frequencies.
Their effective series resistance is definitely higher than that of film capacitors of the same value. See the tan(delta) numbers and graphs in their datasheets, the ESR numbers are usually only given at one or two frequencies.
There is something escaping my understanding. If thermally generated noise voltages are present in an electrolytic capacitor, why shouldn't they be able to supply current to a resistor? Thermally generated noise affects signals, in fact we have signal to noise ratios, which indicate noise plays an important role in signal processing.
It should be in balance with the noise power it gets back from the resistor, if the resistor and the electrolytic capacitor have equal temperatures. Otherwise you could make a fridge (albeit not a very good one) that doesn't require any power source by thermally insulating the device that generates most noise.
A fundamental principle of statistical mechanics is equipartition - each and every degree of freedom has thermal energy of kT/2. So it shouldn't make a difference what the charge carrier is, electron, ion or a charged bucky-ball... The devil in the details is figuring out what the active degrees of freedom are in a particular situation, especially if there's quantization.