Ahh, your post didn't say anything about a vibration pulse. That's almost always down to the winding tightness, not direction- I thought you were looking at noise pickup.
Too hot to sleep and I'm thinking out loud here, so tell me if this is right-
With no voltage on the cap, any signal generated by the cap has to come from some slight piezoelectric phenomenon. That's why high K ceramics are so terrible on microphonics. Since it's a characteristic of the dielectric, construction changes probably won't help much.
With some bias on the cap, any change in plate spacing will cause the voltage to change. This would be true of any cap, and stuff like winding tension, end treatment (epoxy) and outer sleeve could affect it.
There are plastics like PVF that are strongly piezoelectric, but I didn't think the usual decent cap films had any issues with this. Definitely something to look at. IMO, the residuals and non-linearities I'm looking at are so small, microphonics could easily be on the same scale or larger.
With no voltage on the cap, any signal generated by the cap has to come from some slight piezoelectric phenomenon. That's why high K ceramics are so terrible on microphonics. Since it's a characteristic of the dielectric, construction changes probably won't help much.
With some bias on the cap, any change in plate spacing will cause the voltage to change. This would be true of any cap, and stuff like winding tension, end treatment (epoxy) and outer sleeve could affect it.
There are plastics like PVF that are strongly piezoelectric, but I didn't think the usual decent cap films had any issues with this. Definitely something to look at. IMO, the residuals and non-linearities I'm looking at are so small, microphonics could easily be on the same scale or larger.
SY, how would YOU suggest that I make the comparison? Personally, I think that your original comment about the difference between caps was arbitrary and overstated. However, I could be overlooking something, so I made a simple comparison test. Had I not seen a difference, with the same general quality Mylar cap showing a potential problem with zero volts across it, I would have applied a DC voltage as well.
Actually, it's not unless you pole it- that's why it's a good insulator for wire-wrap wire. teflon, OTOH, can be strongly triboelectric. Further, it's very difficult to get enough winding tension on it and it's rather soft, so microphonics are often an issue. PVDF is a lot tougher.
What are you trying to compare? Microphonics? Or something else?
Pease showed how to make a stepped R/C ladder network to approximate a real capacitor with DA. There is no true delay just the phase/frequency response of an ordinary R/C circuit. No source of distortion either.
Pease's work on film caps was to make precision integrators, with far better linearity than Bateman's numbers, Pease never mentioned a distortion mechanism. Any capacitor that has a voltage coefficient of capacitance will produce seconds with a DC bias and thirds at zero bias. Voltage coefficicient of C and DA are possibly related but not the same. Bateman does not offer a compelling and complete description/model of what is going on.
NP0/C0G does give some very low measurable 2nd order nonlinearity
I've seen a little below -100 dB 1 KHz IMD product from a ~40 KHz Sallen-Key using NP0 with 19+20 KHz excitation - that close to the corner there is some dV across the caps
the IMD product disappeared when the NP0 caps were replaced with polystyrene (indirect measurement setup had ~ -160 dB resolution for the 1 KHz IMD product)
Pease' DA plot shows the better of the 2 Mylar as lower DA than the Silver Mica he measured
Understand Capacitor Soakage to Optimize Analog Systems
the semi's junction C modulation will be a much bigger effect than the caps under discussion - cascoding and buffering VAS likely required for any compensation C "quality" discussion to be relevant
a 2-pole "T" network is a option to reduce Miller compensation cap audio frequency "quality" effects on VAS linearity - particularly if the compensation network can be driven by a buffer
for the nested loop Hawksford I/V feedback C I've simmed a passive composite feedback C circuit to be able to use smt caps for low Z @ MHz DAC switching edge/glitch frequencies and wound foil polystyrene at lower frequencies: http://www.diyaudio.com/forums/digi...sford-iv-nested-loop-op-amps.html#post2235820 ( see the "C2" discussion )
I don't know if such extremes are really necessary - but it shows that there are topological work-arounds to some component limitations
I've seen a little below -100 dB 1 KHz IMD product from a ~40 KHz Sallen-Key using NP0 with 19+20 KHz excitation - that close to the corner there is some dV across the caps
the IMD product disappeared when the NP0 caps were replaced with polystyrene (indirect measurement setup had ~ -160 dB resolution for the 1 KHz IMD product)
Pease' DA plot shows the better of the 2 Mylar as lower DA than the Silver Mica he measured
Understand Capacitor Soakage to Optimize Analog Systems
the semi's junction C modulation will be a much bigger effect than the caps under discussion - cascoding and buffering VAS likely required for any compensation C "quality" discussion to be relevant
a 2-pole "T" network is a option to reduce Miller compensation cap audio frequency "quality" effects on VAS linearity - particularly if the compensation network can be driven by a buffer
for the nested loop Hawksford I/V feedback C I've simmed a passive composite feedback C circuit to be able to use smt caps for low Z @ MHz DAC switching edge/glitch frequencies and wound foil polystyrene at lower frequencies: http://www.diyaudio.com/forums/digi...sford-iv-nested-loop-op-amps.html#post2235820 ( see the "C2" discussion )
I don't know if such extremes are really necessary - but it shows that there are topological work-arounds to some component limitations
Last edited:
It seems to me, that hi K caps being made with polar molecules tend to be more microphonic than low K caps, even though the low K caps are necessarily larger. This should be considered an important property of audio caps. So far, Teflon looks pretty good, and Polystyrene as well. Mylar is problematic. Well, what about NPO ceramics? Are they really that good?
Conrad, I read through your paper 'Measured Differences Between Capacitors for Audio Applications', and found it excellent. As you have shown, caps are NOT the ideal components normally used in simulations or implied on the data sheet.
In my life, I went from Tantalum, to ceramic (hi Q), to Mylar or polycarbonate, to polypropylene, and finally to polystyrene and Teflon for, filters, bypasses, and coupling. It took me decades to learn and effectively measure important differences. This is why have thousands of caps that I will NEVER now use for audio design, but still keep around to fix a piece of test equipment or whatever.
In my life, I went from Tantalum, to ceramic (hi Q), to Mylar or polycarbonate, to polypropylene, and finally to polystyrene and Teflon for, filters, bypasses, and coupling. It took me decades to learn and effectively measure important differences. This is why have thousands of caps that I will NEVER now use for audio design, but still keep around to fix a piece of test equipment or whatever.
Maybe not the best you can get, but good enough, even for sub ppm distortion amps.
edit: the higher the voltage rating, the less distortion etc.
Last edited:
Edmond is absolutely correct - use HIGH-VOLTAGE NPO ceramics, and your design will be just fine.
Balaboo, you are talking to an expert in audio design. You would have to PROVE to me that they are as good as polystyrene, before I will readily accept them unconditionally. Perhaps, they really are better in many ways than Mica, that can be shown. Perhaps, with SMT, we find the use for NPO ceramic most compelling.
Last edited:
Given the controversy, and disagreements about cap problems, I would like to show a few insights about capacitor DA. First, IF you only differentially compare one cap vs another with a pure sine wave, all you really seem to see and measure is a few degrees of phase shift, in different parts of the spectrum. However, IF you use an asymmetrical pulse, you get plenty of error. Now, this 'error' shows up both as a delayed change in the original signal, and a reduction in pulse height in the 'high DA' cap. This can be shown in a typical hookup, such as a high DA 10uf electrolytic cap and a 50K load. I measured this 25 years ago and it was published in 'The Audio Amateur' and 'Hi Fi News'. You just have to note the DIRECTION of the error signal, to see that the high DA cap does not get quite as high in level as the quality cap.
Also, the DA effect of even one high DA cap can give you a clearly measurable error compared to a quality cap of up to 100mS. Now, this might not be an 'echo' but it has the time displacement of one. Is it really worth the extra dollar or two saved, to live with this 'error'?
Also, the DA effect of even one high DA cap can give you a clearly measurable error compared to a quality cap of up to 100mS. Now, this might not be an 'echo' but it has the time displacement of one. Is it really worth the extra dollar or two saved, to live with this 'error'?
Mr.Curl, Edmond Stuart, Scott Wurcer;
You are all abolutely correct - I have found that IMD is the distinguishing factor in selecting critical (signal-path) capacitors. Also, it is desirable that the dielectric absorbtion and capacitance variation with-respect-to applied voltage AND current are within (say) +- 25%.
Teflon is fantastic, mica not-so-good, and polystyrene is pretty good - funnily enough, the oil and paper capacitors from the "halcyon" toob days actually are pretty good too !
Definitely foil capacitors, as opposed to metallized.
You are all abolutely correct - I have found that IMD is the distinguishing factor in selecting critical (signal-path) capacitors. Also, it is desirable that the dielectric absorbtion and capacitance variation with-respect-to applied voltage AND current are within (say) +- 25%.
Teflon is fantastic, mica not-so-good, and polystyrene is pretty good - funnily enough, the oil and paper capacitors from the "halcyon" toob days actually are pretty good too !
Definitely foil capacitors, as opposed to metallized.
- Status
- Not open for further replies.