The quality of the capacitors, specially when they are in the signal-line, have a very big impact on the sound-quality of an audio-circuit.There are several factors that determine the quality of a capacitor, some of them are very important for audio-applications:
- The tolerance and the real capacitance (this is important for usage in filters)
- The dependence of the capacitance on the frequency (1µF at 1000 Hz does not mean 1µF at 20 KHz!)
- The internal resistance (ESR)
- The leakage-current
- The ageing (how will those factors evoluate over time)
The best capacitor-choice depends on the application within the circuit and the capacitance we need.
1pF to 1nF range: HF management and feedback circuits
This range is mainly used for HF-elimination in audio-circuits or for feedback purposes like in the bridge of a Quad 606 amplifier.
The best choice in this range are the Silvered Mica capacitors. They have a very good tolerance (up to 1%) and very low distortion- and noise-figures but they are rather expensive.
MKS or MKP are a good alternative. Ceramic capacitors should be avoided in the signal-line as they can cause an extra non-linear distortion of up to 1%.
1nF to 1µF: Coupling, decoupling and supression of oscillation
These are most used in audio-circuits, ao. between stages when there is a DC-level difference, to eliminate oscillation or HF-parasites and in feedback-circuits.
Typically film-capacitors will be used in this range, and even up to 4,7µF. The best choice is Polystyrol/Styrene (MKS) like Wima or Polypropylene (MKP) like Vishay. Polyethylene (MKT) is an alternative at a lower cost (but we are talking cents here).
1µF and above: Power supplies, output capacitors, filters, decoupling
In this range Electrolytic capacitors will be used. The advantage is the very high capacitance (up to 1 Farad). But there are several disadvantages:
- Electrolytic capacitors are subject to ageing and drying out. After 10 years or more the oil is drying and the important factors like the ESR are changeing. They should be replaced every 10 years otherwise they will have a negative impact on the sound.
- They are polarised. There should be at least 1 volt DC over an electrolytic capacitor otherwise non-linear distortion will increase to 1% or higher.
When designing a coupling-circuit electrolytes in the signal-line can often be avoided by recalculating the time-constant (RxC) for a lower capacitance below 1µF so we can use film-capacitors. If this is not possible it is important that there is at least 1 Volt DC over the electrolyte and that a good quality capacitor is used (BHC Aerovox, Nichicon, Epcos, Panasonic ...).
Bipolar capacitors should be avoided, they are in fact two electrolytes in series with opposite polarity and twice the capacitance. This is against the "Less is more" High-end philosophy. In many cases the polarisation can be calculated or measured.
This is often a problem in Opamp-circuits with a symmetric power-supply where only the (low) offset-voltage is available for polarisation. In this case a lower capacitance is often possible because of the high input-resistance of the opamp and we can use film-capacitors. Or we can replace the capacitor with a wire-link.
By choosing the best solution for every application the best sound-quality can be reached. And investing in quality capacitors will have a more positive effect on the sound quality than with any other component.
Comment by Joost
Within the company we had some discussion about Tantalum and other polar capacitors.
I found two very good documents on this subject. One by C. Bateman and one by W. Jung. Both documents are in the download section of our website (Theoretical and practical Electronics papers).
Although the documents are “old”, I did not find any recent quality documents (comments welcome!). Most opinions on this subject are based on those two studies.
In a recent article in the Dutch magazine Elektor, they measured an Electrolyte capacitor intended for the input stage of their Op Amp based power amplifier (yes they did! 32 NE5532 per channel). They rejected it because of the distortion compared to a film capacitor. The distortion can be reduced by connecting two capacitors back to back, as Jung and Bateman also found.
Bateman find high levels of distortion in Tantalum capacitors and rejected them for audio usage. Jung also did notice the high distortion, but with some special techniques they can perform reasonable. Will be continued.
Interesting links about capacitors: