1
\$\begingroup\$

Most signal path/timing capacitors are of stable types such as film or C0G ceramic. However, size and cost force an upper limit on this -- C0G caps out at 10-100nF, while film caps can manage up to 1-10µF albeit at a severe size penalty -- this isn't enough to cover things like audio coupling capacitors and large timing capacitors, which may be 47µF or more. In the "bad old days", the only choices for these tasks were wet aluminum (bulky and with poor tolerances) or MnO2 tantalum (intolerant of reverse polarity but more precise) electrolytics. However, modern technology has given us more options for this -- conductive polymer electrolytes are now in common use with both aluminum oxide and tantalum oxide grown dielectrics, and niobium oxide based capacitors are a niche product as well.

Which of these technologies for large capacitance/high-CV capacitors, old or new, provides the best performance in terms of:

  1. stability (for timing applications)
  2. not distorting coupled signals or introducing excessive noise (for signal chain apps, such as audio or other low frequency signals)?
\$\endgroup\$
  • \$\begingroup\$ and your question is? large value signal,path? or current or frequency? Essentially Z(f) must span a broad range of f and wide range of currents. So Zc(min) at SRF ~ ESR is most important. This can be done with high k X7V on constant voltage supplies, then add bulk storage and compute ESR*C for figure of merit in microseconds. For RF , COG is best where values are small and series R attenuates ripple of Rf for decoupling. \$\endgroup\$ – Sunnyskyguy EE75 Sep 12 '16 at 22:57
  • \$\begingroup\$ @TonyStewart -- you can't stick a Class II dielectric in a signal path app as voltco will cause C and thus Xc to vary with V across the cap = distortion. Same problem with timing caps, too -- the voltco is tolerable when voltages are constant, but not so much when the V across the cap is varying wildly. (You're referring to decoupling/supply bypassing, which is a whole another ball of wax.) \$\endgroup\$ – ThreePhaseEel Sep 13 '16 at 0:30
  • \$\begingroup\$ oops yes plastic caps are still the best here. PU, PE or PS \$\endgroup\$ – Sunnyskyguy EE75 Sep 13 '16 at 0:57
  • \$\begingroup\$ @TonyStewart -- agreed that film and C0G are the top choices here, but I'm talking about cases where you need so much capacitance for coupling, filtering, or timing that a film cap becomes impractically large and costly -- then what? In the past, electros of various types were used, but I'm not sure how the more modern electrolytic technologies (Nb, Ta-poly, Al-poly) fare in signal path/timing apps, either compared to films or old-tech (wet Al, MnO2/Ta) electrolytics.... \$\endgroup\$ – ThreePhaseEel Sep 13 '16 at 3:14
  • \$\begingroup\$ I rarely need a large plastic cap except for 3 phase PFC using rack mounted PU cap banks for large kVAR correction. how about this el34world.com/Forum/… \$\endgroup\$ – Sunnyskyguy EE75 Sep 13 '16 at 3:20
4
\$\begingroup\$

There are actually two cases here, for COUPLING caps specifically, like a mic pre input cap, just make them have really high values so the signal voltage across the cap becomes negligible and so does the distortion, you want the pole way down below the audio band anyway to avoid in band phase shifts (This is not the place to be doing your HPF).

Timing and filter caps on the other hand are a different problem, for timing my usual instinct if I need more then is convenient with the caps at hand is to stick a little toy micro in there and use that, it is generally cheaper then faffing with tants and multi meg resistors (A toy 8 pin micro with a kB of code space and 64 bytes or ram is cheap, cheap, cheap).!

That leaves filters, where scaling the resistor values (and eating the noise penalty) is IMHO usually the way to go, then use C0G or film.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.