Audio coupling: 100nF in parallel with 100uF and a load of lets say 10k. Any improvement seen (THD, freq. resp.)? What do the scientists say?
Last edited:
100uF becomes 100.1uF
Does a change of 0.1% mean much?
If it really is a coupling capacitor then it will have no AC voltage across it.
No AC voltage means no contribution to the AC signal passing through.
Does a change of 0.1% mean much?
If it really is a coupling capacitor then it will have no AC voltage across it.
No AC voltage means no contribution to the AC signal passing through.
At best it will make no difference. At worst it might introduce a notch when the 100nF resonates in parallel with the inductance of the 100uF. Fortunately this notch will be heavily damped by the ESR of the electrolytic, so if you are going to 'bypass' caps best to use high ESR electrolytics. The notch, if present, may or may not be audible and may or may not affect the phase of HF audio. If audible, bypass fans will always interpret the notch as an improvement even though it is in fact a degradation.
And your engineering point is?diyralf said:
Be aware that it is possible to acquire a Masters in Electronic Engineering without doing very much in the way of circuit theory or electromagnetism. For example, some EEs can be almost indistinguishable from IT graduates or ergonomes.AndrewT said:
Almost likely to be placebo. A typical 100µ/25V Al electrolytic has about half an ohm of ESR, that's 0.05% of 10k. Even if it were rather badly nonlinear and generated 1% of THD at these very low levels (0.1 mV of drop), we'd still be talking 0.0005%. The THD spec of a typical CD player is 0.0025-0.0035%.
What happens when the electrolytic is old and leaky (for lack of any meaningful voltage across it) is anyone's guess, but the 100n couldn't do anything about that anyway. Its impedance is only at 60 ohms by 20 kHz.
What happens when the electrolytic is old and leaky (for lack of any meaningful voltage across it) is anyone's guess, but the 100n couldn't do anything about that anyway. Its impedance is only at 60 ohms by 20 kHz.
Last edited:
Thank you for this theoretical explanation. But I would be interested in practical measurements, from people who have expensive and accurate measuring devices.
You asked what the scientists say. Scientists expect components to behave according to the laws of physics, so are happy to use appropriate theory to predict results of experiments on trivial systems. People who have expensive and accurate measuring devices do not necessarily make good measurements, as all measurements require the correct application of theory to turn raw data (what the instruments say) into measurements (what reality is) and the mere possession of an instrument does not guarantee possession of the necessary knowledge to use it correctly.diyralf said:
If you want a voodoo answer then I am sure some people can provide it. They will probably not use any instruments at all, and they probably will not have a clue about science.
Agree and add: expensive and accurate instruments *confirm* that Physics Rules still apply.
Back to your preamp example: 100uF in series with 10k is ridiculous high, because :
1) crossover frequency is 0.16 Hz .
Looks impressive on a brochure, not much music that low.
Definitely NO speaker will reproduce it.
2) using 100.1uF instead will now shift crossover frequency down by 0.1%
A very expensive meter *might* measure some change, most will not, even Lab types.
And a Real World fact is that the main 100uF cap will probably shift capacitance by more than 0.1% simply because of room temperature variation, applied voltage or other factors.
3) adding a ceramic capacitor in parallel with a (presumed) electrolytic is normally done to bypass any internal inductance it might show.
Since 0.1uF cap impedance is 1000 times *higher* than the 100uF one, for it to "improve" things electrolytic internal impedance, be it ESR or inductive must rise *at least* 100X its normal value, just to *start* seeing an improvement offered by the add-on.
I think we will be talking 100`s of kHz to reach that point, if at all.
By the way, all this will be confirmed by expensive accurate instruments 😉
Back to your preamp example: 100uF in series with 10k is ridiculous high, because :
1) crossover frequency is 0.16 Hz .
Looks impressive on a brochure, not much music that low.
Definitely NO speaker will reproduce it.
2) using 100.1uF instead will now shift crossover frequency down by 0.1%
A very expensive meter *might* measure some change, most will not, even Lab types.
And a Real World fact is that the main 100uF cap will probably shift capacitance by more than 0.1% simply because of room temperature variation, applied voltage or other factors.
3) adding a ceramic capacitor in parallel with a (presumed) electrolytic is normally done to bypass any internal inductance it might show.
Since 0.1uF cap impedance is 1000 times *higher* than the 100uF one, for it to "improve" things electrolytic internal impedance, be it ESR or inductive must rise *at least* 100X its normal value, just to *start* seeing an improvement offered by the add-on.
I think we will be talking 100`s of kHz to reach that point, if at all.
By the way, all this will be confirmed by expensive accurate instruments 😉
No change. 100nF is a thousand times smaller in capacitance than 100uF, so it cannot make any meaningful difference in the audio frequency region. If you were designing a radio then the answer might be different, but then someone designing a radio would not use 100uF as an RF coupling cap. In RF design it is sometimes useful to bypass a decoupling cap, but then there are usually several bypasses and the aim is to get a low impedance for RF with audio being almost irrelevant.
The 100 mfd will do fine for the low freq and the .1mfd will do Ok for the high freq (actually might be effective above the hearing range). I would by pass it with a large enough value to affect the most important midrange freq. Maybe .47-1.0mfd and let This cap do the majority of the audible work.
THx-RNMarsh
THx-RNMarsh
I'm worried now(should I be?) about what's happening(or not) to all the high(er) frequencies going through(or not) the large(ish) electrolytics in my preamp...🙄
Mid-frequencies get lost in an electrolytic and wander around the cavities produced by the foil etching. They eventually emerge (the mid-frequencies, that is, not the cavities) but the sound is smeared. The simple solution is to bypass the electrolytic with a mid-value cap so that the mid-frequencies have an alternate path to take. They do this even though the mid-value cap is likely to have a higher impedance than the electrolytic at mid-frequencies, because mid-frequencies are sufficiently intelligent to know what was in the mind of the designer.
See, I can answer my own prayers!
People who really understand audio are not convinced by numbers of course, but for the boring engineers here are the numbers:
100uF has a capacitive reactance of j1.6R at 1kHz. Let us assume it also has ESR of 1R. This gives a net impedance of 1.9R in magnitude. Let us assume that the foil cavities raise the mid-band impedance by a factor of 10 (although it is unlikely to be as bad as that). So our electrolytic has an impedance of 19R at 1kHz.
The proposed 1uF mid-band bypass will have a reactance at 1kHz of j160R. Therefore it will carry about 10% of the signal current. Whatever nasty effect the electrolytic will have on the signal if left alone will be reduced to 90% of that nasty effect - clearly a huge improvement. (Note that this is 90% of nasty, not reduced by 90% to 10% of nasty.)
Now let us assume that the 100uF was correctly chosen as a coupling capacitor so that it produces an LF rolloff at around 2Hz or less. This means it will be feeding not less than 800 ohms resistance; round it up to 1k. Our electrolytic would have had about 2% of the signal across it at 1kHz. Let us assume that it is a really bad electrolytic with 10% distortion (I don't know if anyone makes an electrolytic that bad, but go with the flow); 2% of 10% is 0.2%. Pretty bad, eh? Adding the 1uF bypass will reduce this to 0.18% - a huge improvement.
See, I can answer my own prayers!
People who really understand audio are not convinced by numbers of course, but for the boring engineers here are the numbers:
100uF has a capacitive reactance of j1.6R at 1kHz. Let us assume it also has ESR of 1R. This gives a net impedance of 1.9R in magnitude. Let us assume that the foil cavities raise the mid-band impedance by a factor of 10 (although it is unlikely to be as bad as that). So our electrolytic has an impedance of 19R at 1kHz.
The proposed 1uF mid-band bypass will have a reactance at 1kHz of j160R. Therefore it will carry about 10% of the signal current. Whatever nasty effect the electrolytic will have on the signal if left alone will be reduced to 90% of that nasty effect - clearly a huge improvement. (Note that this is 90% of nasty, not reduced by 90% to 10% of nasty.)
Now let us assume that the 100uF was correctly chosen as a coupling capacitor so that it produces an LF rolloff at around 2Hz or less. This means it will be feeding not less than 800 ohms resistance; round it up to 1k. Our electrolytic would have had about 2% of the signal across it at 1kHz. Let us assume that it is a really bad electrolytic with 10% distortion (I don't know if anyone makes an electrolytic that bad, but go with the flow); 2% of 10% is 0.2%. Pretty bad, eh? Adding the 1uF bypass will reduce this to 0.18% - a huge improvement.
- Status
- Not open for further replies.