George, I understand that leakage current, especially in a coupling cap, is undesired. Yet, how would it have any effect on audibility? Would for instance increased output offset due to cap leakage have any change to be audible unless it is pathological?
Jan
Hi Jan
Leakage allows dc voltage pass through the capacitor and alter dc bias point of a circuit.
Higher leakage current develops a higher voltage on it’s load resistor, therefore a higher current means that a higher dc voltage has the potential to alter the dc bias point of a circuit.
After initial excitation of the coupling cap by the dc voltage which is to block, there will be a settling time during which the leakage current falls off exponentially approaching asymptotically to a settling point .
The settling time seems to be much longer and settling point is higher for the Al electrolytic compared to the non polar film caps.
When the electrolytic cap is in a high impedance circuit, the settling time can be very long (quarter to half an hour), and when first charged (new) or after prolonged period of non use, quite longer.
My understanding is that it is the shift of bias point that will cause the audible effect.
George
Yes. Depending on a lot of things in each case then. Best to use a low leakage to be sure.
Next question: what is a reasonable data point for low leakage? We could relate it to the DC bias input current of the circuit. Like: the leakage should be less than 1/10 of the circuit input bias? Just thinking out loud here.
Jan
This long controversy about caps and servo is amusing, and reveling of the shortcomings of this forum, the pleasure that a little bunch of people, always the sames, seems to feel in attacking and contradicting.
Are caps in the signal path perfectly transparent for Audio ? No.
Can we measure they flaws ? Yes.
Can-you hear differences in the way they sound ? Yes. more or less.
Are the colorations they can bring obnoxious ? No. But it depend of your level of exigence.
Are some brands better than others ? Yes. (The worse are electrolytic, followed by ceramic and tantalum).
Is-it necessary to spend a lot of money in buying "audiophile" caps to get good enough performance ? Better to offer flowers to your wife or add new albums of music to your collection.
But there is a very good way to get rid of those caps, and it is Servos. It is not so complicated, in a new design, the price of an appropriate integrated circuit will not kill-you, and the caps you will use in it will be lot less expensive and of a better quality, because you will reduce its value, his volume, so, its price.
Are servos totally transparent ? Not necessary, depend of a lot of factor, but, well designed, a lot more than the caps they replace. If I compare mine in/out, I cannot hear any difference in the texture of the sound. So transparent enough on my opinion.
What is the flaw of a servo ? It is a little complicated, you need to power-it with all the extra work it brings (appropriate voltages, filtration, Printed circuit design etc.)
Now, obviously, if you want to improve an existing amplifier that use electrolytic caps in the signal path, it will be more simple to exchange them with a good film cap. When I say good, I dont mean golden plated with unoptainium inside, but industrial brand on the shelf.
About electrolytic caps, their is no way to get rid of them,when you need big farads and voltages, in filtering as an example. They will wear, means that you need to recap all the 10/15 years. And they need, for best results to be paralleled with better caps at HF. Taking care to do not create a resonant circuit with their inductances.
In the signal path ? I try everything to avoid them, free to everybody to do as he feels..
Are caps in the signal path perfectly transparent for Audio ? No.
Can we measure they flaws ? Yes.
Can-you hear differences in the way they sound ? Yes. more or less.
Are the colorations they can bring obnoxious ? No. But it depend of your level of exigence.
Are some brands better than others ? Yes. (The worse are electrolytic, followed by ceramic and tantalum).
Is-it necessary to spend a lot of money in buying "audiophile" caps to get good enough performance ? Better to offer flowers to your wife or add new albums of music to your collection.
But there is a very good way to get rid of those caps, and it is Servos. It is not so complicated, in a new design, the price of an appropriate integrated circuit will not kill-you, and the caps you will use in it will be lot less expensive and of a better quality, because you will reduce its value, his volume, so, its price.
Are servos totally transparent ? Not necessary, depend of a lot of factor, but, well designed, a lot more than the caps they replace. If I compare mine in/out, I cannot hear any difference in the texture of the sound. So transparent enough on my opinion.
What is the flaw of a servo ? It is a little complicated, you need to power-it with all the extra work it brings (appropriate voltages, filtration, Printed circuit design etc.)
Now, obviously, if you want to improve an existing amplifier that use electrolytic caps in the signal path, it will be more simple to exchange them with a good film cap. When I say good, I dont mean golden plated with unoptainium inside, but industrial brand on the shelf.
About electrolytic caps, their is no way to get rid of them,when you need big farads and voltages, in filtering as an example. They will wear, means that you need to recap all the 10/15 years. And they need, for best results to be paralleled with better caps at HF. Taking care to do not create a resonant circuit with their inductances.
In the signal path ? I try everything to avoid them, free to everybody to do as he feels..
How do you know about the number of beliefs?
At least in the case of coupling caps between MC head amps and equalization stage we found corrobation for the "caps might sound different" hypothesis, as we compared metallized polyester film to polypropylen film cap in controlled listening tests (as described earlier).
"caps might sound different" and "Can-you hear differences in the way they sound ? Yes. more or less." are very different statements. For sure we can all come up with a scenario where the difference is significant but with any sort of thought in the design the levels should be at the point where they are well below the room noise floor (or vinyl noise floor, which is much higher).
Having looked at Cyril's results for non-polar electrolytic, H2 at -117dB is for me, the least of the problems in my system. I also remain unconvinced anyone can detect H2 at this level, but have an open mind that a bat/human hybrid exists somewhere
(confession, of course my phono stage is fully balanced and servo'd with only filter caps. I do not claim any audible benefits for this, I just wanted to build one as miniDSP has balanced inputs so seemed a good idea at the time. )
Having looked at Cyril's results for non-polar electrolytic, H2 at -117dB is for me, the least of the problems in my system. I also remain unconvinced anyone can detect H2 at this level, but have an open mind that a bat/human hybrid exists somewhere
(confession, of course my phono stage is fully balanced and servo'd with only filter caps. I do not claim any audible benefits for this, I just wanted to build one as miniDSP has balanced inputs so seemed a good idea at the time. )
And now?
My pet pleasure is taking ages to research something and then finding the benefit of actually doing it is not there. It makes me feel like I've achieved something really worthwhile and done no harm along the way.
Now I know that matching resistors to keep CMRR at a point that balanced is worthwhile is a pain. Only outweighed by the knowledge that both John Curl and Doug Self think it is a bad idea. And now I have a balanced MM stage I have to build a balanced MC so I've really shot myself in the foot. But the joy of doing this for a hobby rather than a living is that you can do silly things
May be it is very different for a native english speaker but not so much for us?!
Considering the comparable german phrase i wouldn´t interpret it as "very" different just as meaning "might be important or not" ....
In our application it wasn´t imo not a scenario where a sonic difference was likely or predicted by the usual capacitor model understanding.
Having looked at Cyril's results for non-polar electrolytic, H2 at -117dB is for me, the least of the problems in my system. I also remain unconvinced anyone can detect H2 at this level, but have an open mind that a bat/human hybrid exists somewhere
I´d agree to the H2 audibility assumption, but all boils down to the point where you have to express what model theory you´re relying on.
You have to _believe_ (totally trust in your model correctness) that the H2 is the only difference possible or at least the only influencial variable to conclude about the audibility.
Using the "my model covers everything" premise can be dangerous because in most cases you simply don´t know if it´s true.
Hello Bill,
please let me bring one real example. The best measuring amp I ever built is the one with class A output stage and this output stage even has local feedback to minimize large signal distortion. In cases like that, I am not able to measure distortion with my toys and have to visit a colleague who owns AP SYS 2712. So we measured this amp on SYS 2712 and had to use a notch filter option to find something. Distortion, with the base frequency notched, was -129dBr (0.000035%), purely third harmonic, all other distortion components below noise bottom. And, this amp has mylar 1uF coupling input capacitor loaded with 47k resistor to ground. Yes mylar, which is believed to have "distortion", and is doing absolutely nothing in this circuit concerning distortion.
BTW, what was the load when CB measured -117dB H2 for a non-polar electrolytic? Anything that would be normally used, or, again, there was a significant ac voltage across the cap?
I'm working on tube amps, so Al electrolytic capacitors as coupling caps are right out.
I would like to compare MKT, MKP, and PIO (Russian surplus) to PTFE (also Russian surplus).
So far, any difference I have seen for THD have been test errors with my test setup.
Test setup is a sound card running 24bit 192K input and output with a limit of 1Vrms before the sound card distortion shows up.
I've thought about building a test jig with a NE5532 or NE5534 configured as a gain of five amp to drive the cap, followed by a high resistance resistor divider driving a second op amp to buffer the signal and drive the sound card input.
The problem I see is I push the signal down by near 14dB by dividing it back down to drive the sound card.
A better solution would be to build two twin T notch filters to remove the fundamentals, and retain the full harmonics. but that gets a bit more complicated than I wished to do. And I'm concerned about THD/IM contribution from the op amps relative to what I am trying to measure
I would like to compare MKT, MKP, and PIO (Russian surplus) to PTFE (also Russian surplus).
So far, any difference I have seen for THD have been test errors with my test setup.
Test setup is a sound card running 24bit 192K input and output with a limit of 1Vrms before the sound card distortion shows up.
I've thought about building a test jig with a NE5532 or NE5534 configured as a gain of five amp to drive the cap, followed by a high resistance resistor divider driving a second op amp to buffer the signal and drive the sound card input.
The problem I see is I push the signal down by near 14dB by dividing it back down to drive the sound card.
A better solution would be to build two twin T notch filters to remove the fundamentals, and retain the full harmonics. but that gets a bit more complicated than I wished to do. And I'm concerned about THD/IM contribution from the op amps relative to what I am trying to measure
Not sure if I understand, the main reason for me to use input cap is to prevent possible problems in case of eventual failure of the preamp or another signal source. Under normal conditions, there is no DC (at least in audio chain that I use).
- Status
- Not open for further replies.