My audio "bible" states that the use of polarized electrolytic caps should be avoided for audio filters due to the fact that capacitance, ESR and inductance can change with temperature and frequency. The 50 year old pre-amp I'm working on happens to have a 100uF, 15V polarized electrolytic in a filter circuit. I'm going through it and checking all of the caps with my Sencore LC-101. This cap is marginal (ESR) and am considering replacing it with an MLCC. That is, until I got the sticker shock (~$13.00 for 1 cap). I'm used to paying hefty sums for large electrolytics, but this seems high for a 100uF cap. Questions: is it worth it or would you just replace it with another polarized electrolytic?
Here's the schematic. The cap in question is C223 which is a bypass cap. The neg side of the cap is tied to ground. Thanks for any tips.
Attachments
It's still hard to see the full ramifications of device choice without the full schematic. Even then the layout of the circuit will impact on the actual performance of the device. Schematics imply that grounds are equipotential, which they are not. This capacitor is bypassing ripple on the bias for Q205. The capacitor can only do this if ripple current flows through it, and in turn current can only flow if there is a voltage drop, therefore the (-) leg of C223 cannot be a noiseless ground unless directly connected to a star earthing layout
In any case, the capacitor in question is not just a bypass; it is there to bypass the supply operating point of Q205, but also so that the junction of R239 and R241 is an audio signal ground. The quality of the audio signal presented to the base of Q205 will be effected by the audio quality of C223, as well as the integrity of the ground end of this capacitor.
Given that the rest of the circuit is likely riddled with similar short-comings I suspect that spending too much on this one component is overkill. I'd replace C223 with an Elna Tonerex audio grade electrolytic as first choice, Nichicon Fine Gold audio grade electrolytic as second choice. Tonerex have significantly lower microphonic (mechanical) noise in my experience, therefore are less subject to environmental noise coupling into the circuit. I can't comment objectively on the sound quality of either; subjectively I like to think that Tonerex has the edge over Fine Gold.
Elna Miniature Standard Capacitors for Audio
Nichicon High Grade Standard Type, For Audio Equipment
In any case, the capacitor in question is not just a bypass; it is there to bypass the supply operating point of Q205, but also so that the junction of R239 and R241 is an audio signal ground. The quality of the audio signal presented to the base of Q205 will be effected by the audio quality of C223, as well as the integrity of the ground end of this capacitor.
Given that the rest of the circuit is likely riddled with similar short-comings I suspect that spending too much on this one component is overkill. I'd replace C223 with an Elna Tonerex audio grade electrolytic as first choice, Nichicon Fine Gold audio grade electrolytic as second choice. Tonerex have significantly lower microphonic (mechanical) noise in my experience, therefore are less subject to environmental noise coupling into the circuit. I can't comment objectively on the sound quality of either; subjectively I like to think that Tonerex has the edge over Fine Gold.
Elna Miniature Standard Capacitors for Audio
Nichicon High Grade Standard Type, For Audio Equipment
Thank you for the insight. I really appreciate the time that you took to explain your suggestion.
Regards....
From what I can see of that partial schematic that cap just separates the DC feedback path from the signal by shunting it to AC ground - no significant signal voltages appear across it. Its such a large value that all the time-constants its contributing to are long (several seconds) AFAICT.
There's nothing wrong with using electrolytics like this as its not affecting the audio signal, any ESL/ESR is seen through a 220k resistor by the signal path so is basically irrelevant.
The key thing is to ensure the roll-off frequency is well below the audio band so the cap value doesn't affect the response curve. Hence the time constants measured in seconds. At 20Hz 100µF has about 80 ohms reactance, so any imperfection in its response is being diluted by the ratio of that to the 220k bias resistor, ie about 70dB down. And at higher frequencies the effect is even less. So even if the cap had 5% distortion it would only give 0.0018% distortion at 20Hz (0.000036% at 1kHz) to the circuit, and at the low signal levels seen across it 5% is an extreme over-estimate of electrolytic non-linearities - if the signal level was as high 5Vrms at 20Hz the capacitor sees only about 2mV and 22µA of signal - a capacitor would have to be as non-linear as a diode to give anything like 5% distortion at 2mV signal levels, which clearly isn't remotely the case even with a cheapo-brand cap.
There's nothing wrong with using electrolytics like this as its not affecting the audio signal, any ESL/ESR is seen through a 220k resistor by the signal path so is basically irrelevant.
The key thing is to ensure the roll-off frequency is well below the audio band so the cap value doesn't affect the response curve. Hence the time constants measured in seconds. At 20Hz 100µF has about 80 ohms reactance, so any imperfection in its response is being diluted by the ratio of that to the 220k bias resistor, ie about 70dB down. And at higher frequencies the effect is even less. So even if the cap had 5% distortion it would only give 0.0018% distortion at 20Hz (0.000036% at 1kHz) to the circuit, and at the low signal levels seen across it 5% is an extreme over-estimate of electrolytic non-linearities - if the signal level was as high 5Vrms at 20Hz the capacitor sees only about 2mV and 22µA of signal - a capacitor would have to be as non-linear as a diode to give anything like 5% distortion at 2mV signal levels, which clearly isn't remotely the case even with a cheapo-brand cap.
That's a fair call Mark, the 220kΩ resistor is acting as a potential divider of any distortion in the capacitor.
The C223 does need to be sized relative to R239 to reduce power supply ripple on the base of Q205 to acceptable levels, which demands a larger value than that necessary for low frequency rolloff in this circuit.
Would I still use a Tonerex or similar? Yes, if only because I keep on hand a large range of values in Elna Tonerex, Cerafine and Silmic II, and Nichicon Fine Gold and Muse capacitors for audio applications, which are all premium capacitors of known quality, performance and life.
Last edited: