Thanks Bob, I ran across this problem just last week. I worked out a really cute phono buffer for my article based on using the 48V phantom powered mike inputs available on a lot of external sound cards. But it needs 4-10uF input coupling caps.
Hi Andrew.
That's a whole bunch of questions, I'm just looking for quick easy answers
.Can you give relevant references, or could you give more explanations to the questions in your post.
Dan.
I only asked 5 and answered the first myself. The 4 others were about the silence in the test reports re signal coupling duty.
Read my first sentence again.
Last edited:
If You are looking for coupling caps in line level aplication (so only few mV AC voltage across cap in aplication, in whole audioband), simply use "big enough" capacity value(e.g for 1kohm load use 47uF or more) and You can forget distortion (in order 0,00005% across audioband) . You can use cheapest polarized electrolytic capacitors..
Last edited:
The only place where i was obliged to use electolytics in my passive filter was the RLC Motional (resonance) compensation network of my bass speaker. But i found a problem with those: Their values tend to decrease with years. I'm still looking for a solution here.
(I can use here caps with a big internal and inductive impedance)
AES Settles on SOFA for 3D HRTF Datasets | InnerFidelity If this happens then there is hope yet for a real step forward in music reproduction.
See how it goes? A question about caps (or something else) is raised, then people give their advice, some good some bad, then some people say that it isn't important in the first place, and to use whatever you can get.
The best answer for coupling caps is: NO CAP is better than any cap. Use servos or better IC's to remove the need for coupling caps.
When it is absolutely necessary to use a cap, like in RIAA EQ, some coupling applications, or loudspeaker crossovers, use polypropylene film if you possibly can, preferably a well known brand with serious audio manufacturers.
Only use electrolytic caps when you absolutely have to, for size reasons. Then try for a Silmic cap, that serious designers use, like OPPO, on that occasion.
Besides IM or harmonic distortion measurement, there is always dielectric absorption and mechanical resonance. I know we started to talk about and measure this more than 35 years ago, and many here probably missed it, but our published results, Richard Marsh's and mine, for example, are still valid. Check out Walt Jung's website for further input.
The best answer for coupling caps is: NO CAP is better than any cap. Use servos or better IC's to remove the need for coupling caps.
When it is absolutely necessary to use a cap, like in RIAA EQ, some coupling applications, or loudspeaker crossovers, use polypropylene film if you possibly can, preferably a well known brand with serious audio manufacturers.
Only use electrolytic caps when you absolutely have to, for size reasons. Then try for a Silmic cap, that serious designers use, like OPPO, on that occasion.
Besides IM or harmonic distortion measurement, there is always dielectric absorption and mechanical resonance. I know we started to talk about and measure this more than 35 years ago, and many here probably missed it, but our published results, Richard Marsh's and mine, for example, are still valid. Check out Walt Jung's website for further input.
Last edited:
servos and caps
It is odd that many see the servo as somehow not being in the audio path because it is not specifically in series between one stage and the next. But agree with JC that the best capacitor is no capacitor. And you can use a somewhat smaller value in a servo, permitting it to be a decent film or even a stack of NP0s.
The other trick with servos is to not need them to work too hard. Then the noise associated with them can be minimized. There is also the option of making them more of a sampled-data system, so that the convergence to approximately zero d.c. can occur under system control and not during music. A multiplying DAC as the source of the correction can be used. A friend for whom I was doing some work cringed when I mentioned the DAC, but for him a DAC was always something churning away with the attendant noise. A sampled-data loop can be static after arriving at a correctional setting.
Another servo thing I see applied a lot that is questionable: the servo is a differential one, with two capacitors. There is almost always a place to apply the correction so that a simple inverting integrator will suffice.
Brad
It is odd that many see the servo as somehow not being in the audio path because it is not specifically in series between one stage and the next. But agree with JC that the best capacitor is no capacitor. And you can use a somewhat smaller value in a servo, permitting it to be a decent film or even a stack of NP0s.
The other trick with servos is to not need them to work too hard. Then the noise associated with them can be minimized. There is also the option of making them more of a sampled-data system, so that the convergence to approximately zero d.c. can occur under system control and not during music. A multiplying DAC as the source of the correction can be used. A friend for whom I was doing some work cringed when I mentioned the DAC, but for him a DAC was always something churning away with the attendant noise. A sampled-data loop can be static after arriving at a correctional setting.
Another servo thing I see applied a lot that is questionable: the servo is a differential one, with two capacitors. There is almost always a place to apply the correction so that a simple inverting integrator will suffice.
Brad
All Your published measurements results were in circuit, where is significant AC voltage across capacitor. And this is not coupling , but frequency filtration, so quite different application. Try to connect ordinary electrolytic capacitor e.g 47 uF in series with 10kohm load resistor and make distortion measurement across 10k load resistor..Very simple.
So do I. Stupidly trying to use caps in Speakers passive crossovers.
But, i confirm, you cannot be a serious designer, because only one exists: The great John Curl himself.
But don't even try to approach him: he is the human equivalent of the black hole in the center of the universe.
Last edited:
True, although some of the imperfections, like limited Q due to series resistance, can be accommodated in some applications. But of course there is the susceptibility to external B fields.
I got a surprise when I specified an air-core coil for a sensor. I wanted the inductance to be a simple function of the shape and size and anticipated a temperature dependence of Q based on the copper resistance tempco. Surprisingly, the insulation had an effect of increasing Q with temperature for a while, and it was an effect larger than the opposing one due to the Cu. I tried to get additional coils with different insulation, but dismal sales of the non-audio product ended that investigation.
- Status
- Not open for further replies.