Low-distortion Audio-range Oscillator

[davada]

Originally Posted by richiem

Yes, I too doubt the Lynx results. As to C0G/NP0 -- I have been told numerous times by folks I trusted that all of the ceramic cap materials are to a greater or lesser extent piezoelectric. I think this is not a Good Thing. But again, in the absence of facts....

Well the best way to find out is to stick them in one our oscillators and see what happens.
Any volunteers?

[DDB]

If someone has a 1K Victor oscillator it is pretty easy to remove the red film caps and reinstall cogs there for a test.

[Samuel Groner]

There has been discussion that capacitors do create distortion but I'm not clear about the mechanism that causes this.

In fact there are several mechanisms. Consider this equivalent circuit:



At audio frequencies, only the two R and the basic C element are of importance. Any of these may show a voltage coefficient and thus contribute distortion.

The exact physical explanation why the basic capacitance element exhibits a voltage coefficient is not clear to me either. One possible contribution could be that the electric field contained within the plates of the capacitor may slightly bend/displace the plates, and thus cause capacitance modulation with signal voltage.

The leakage resistance is formed by parasitic elements of not well known chemical composition. These not necessarily follow Ohm's law (which in essence applies to metal conductors only) very well, and may show significant voltage coefficient.

The ESR is at least partially formed by end contact resistance. Bruce Hofer points out in his "Building Analog in the 2010s" slides that, particularly for film capacitors, these may be unreliable and cause some distortion contribution.

Which mechanism dominates depends very much on the circuit impedance, i.e. the corresponding R which forms the effective time constant. At medium impedances (1-10 kOhm), the basic voltage coefficient of the C element seems to be the most significant in my experience. Nonetheless, this discussion shows that you can't fully characterize capacitor distortion at a single frequency. Capacitor distortion should be measured at least at three widely spaced frequencies.

Capacitor in series have a tendency to not share equal amount of charge.

You're probably thinking about series connected electrolytic capacitors in HV supplies, where one needs to use parallel resistors to ensure equal voltage distribution (as otherwise the very variable leakage resistance could cause one capacitor to carry most of the burden). In AC applications, the voltage/charge distortion is well definied (simply by the capacitance ratios).

Samuel

[davada]

Originally Posted by Samuel Groner

In fact there are several mechanisms. Consider this equivalent circuit:



At audio frequencies, only the two R and the basic C element are of importance. Any of these may show a voltage coefficient and thus contribute distortion.

The exact physical explanation why the basic capacitance element exhibits a voltage coefficient is not clear to me either. One possible contribution could be that the electric field contained within the plates of the capacitor may slightly bend/displace the plates, and thus cause capacitance modulation with signal voltage.

The leakage resistance is formed by parasitic elements of not well known chemical composition. These not necessarily follow Ohm's law (which in essence applies to metal conductors only) very well, and may show significant voltage coefficient.

The ESR is at least partially formed by end contact resistance. Bruce Hofer points out in his "Building Analog in the 2010s" slides that, particularly for film capacitors, these may be unreliable and cause some distortion contribution.

Which mechanism dominates depends very much on the circuit impedance, i.e. the corresponding R which forms the effective time constant. At medium impedances (1-10 kOhm), the basic voltage coefficient of the C element seems to be the most significant in my experience. Nonetheless, this discussion shows that you can't fully characterize capacitor distortion at a single frequency. Capacitor distortion should be measured at least at three widely spaced frequencies.



You're probably thinking about series connected electrolytic capacitors in HV supplies, where one needs to use parallel resistors to ensure equal voltage distribution (as otherwise the very variable leakage resistance could cause one capacitor to carry most of the burden). In AC applications, the voltage/charge distortion is well definied (simply by the capacitance ratios).

Samuel

Is there a possibility of dielectric swelling. I seem to have a distant memory of some mention of this from a physics class.

[HpW]

Demian,

Originally Posted by 1audio

I have my doubts but perhaps its possible. Does anyone else have experience with the L22? It would be a simple solution.

I own a such L22 (PCI-card) , just ask for a specific measurement, keep in mind that this card is/was not cheap

HpW

[coluke]

On C0G/NP0 caps THD: this is from Bateman - the only one I have at hand at the moment:


L.

[RNMarsh]

Just about anything-including electrolytics- measure well with symetrical waveforms (sine waves). In my 339A I have bipolar types in the signal path. Its a very INsensitive test. But with sine waves to get below -130 takes special care in cap selection and test equipment.
[ BTW- its with asymetrical waveforms (music) where the cap issues show themselves to measure and... sound poor. ]

-Richard

[davada]

Originally Posted by coluke

On C0G/NP0 caps THD: this is from Bateman - the only one I have at hand at the moment:


L.

Hi Coluke,

Do you have one for a PP just for comparison sake.

[scott wurcer]

Originally Posted by RNMarsh

[ BTW- its with asymetrical waveforms (music) where the cap issues show themselves to measure and... sound poor. ]

-Richard

As you know I don't agree with this anymore than in-harmonic distortion. Please describe an I/V characteristic of a capacitor that would show this behavior. The mechanical effect just quoted has a rectifying property (the force vs voltage is always anti-repellant) and will manifest on a simple sine wave. The extreme case would be the condensor microphone analyzed in detail by B&K.

[davada]

Originally Posted by scott wurcer

As you know I don't agree with this anymore than in-harmonic distortion.

Hi Scott,

I,m not clear about your meaning here.

You don't agree in-harmonic distortion in capacitors?