Could someone explain, how small ceramic caps are worse than corresponding electrolytic ones?

Hallo Bartosz,

if you see, that in an application a certain part is used, although there could have been used a cheaper and better alternative, then this need not have to do with good design practice, but it can also be the result of 'tradition', or even worse, 'philosophy'.

If today we want to choose a ceramic capacitor for a certain application, then we have three options: NP0, X7R and Z5U. Of course, there still some other flavours, but these three types is the standard. If we go a bit back in history, let's say 20 years or so, situation was different. People could choose between dozens over dozens of different ceramic materials:

Class I ceramics, made of TiO2:
P100, NP0, N033, N047, N075, N150, N220, N330, N470, N750, N1500, N2200, N3300, N4700, N5600 and others.

Class II ceramics, made of BaTiO3: 2A4, 2B4, 2C4, 2D4, 2E4, 2F4, 2A2, 2B2, 2C2, 2D2, 2E2, 2F2 and many others.

Class III ceramics, made of semiconducting ceramics, using Ba and Ti, showing so dramatical bad performance, that these capacitors should never be used in any high reliability application!

Within each class there are better and worse variants, depending on permitivity 'Er' of according ceramic material: The lower 'Er', the more stable (better) the ceramic material. Means NP0 has better stability than N1500 for instance, or 2A4 is more stable than 2F4, etc.

Nevertheless, these three classes drastically differ in performance:
Class I ceramics do not show any dependency of capacitance with DC voltage superimposed to signal or with frequency of signal. But due to relatively small 'Er' only small capacitances can be built with them.
Class II ceramics do show a certain dependency of capacitance with DC voltage superimposed to signal and with frequency of signal.
Class III ceramics are so terrible, that their performance is nearly unpredictable. They can only be used for decoupling purposes, where quality of capacitor does not play any important role...

Todays NP0 type is identical with older NP0 class I ceramics, so that performance is also outstandingly good. X7R is a rather high quality derivative of older Class II ceramics and Z5U is a rather low quality derivative of older Class II ceramics. Means, while X7R shows an acceptable dependency of capacitance with superimosed DC voltage and frequency for many applications, Z5U is much less stable than X7R and should only be used for decoupling purposes at room temperature. Z5U capacitors are smaller than X7R ones of same capacitance, because Z5U ceramics shows a much higher 'Er'.

Back to your question: Is a ceramic capacitor worse than an electrolytic one?
Well, this mainly depends on your application! If you want to use a cap to provide proper power supply decoupling, then I would never choose a 0.47µF aluminmium electrolytic instead of 100nF X7R. And if I want to filter out some high frequency noise in a DC application, I would never use a small aluminium electrolytic capacitor instead of some X7R type. Here the outstanding low equivalent series resistance of X7R capacitor in the range of 0.01Ohm beats the small aluminium electrolytic cap by about three orders of magnitude!!

But if I had to remove DC voltage from an audio signal, then I would not use any ceramic capacitor. Because the ones providing enough capacitance for that job, would show too much dependency of capacitance with DC voltage and with frequency of signal. In such a case I would use a metallized polyester foil capacitor (MKS, MKT, etc.) or just an aluminium electrolytic capacitor of suited capacitance. If capacitance is choosen rather high, let's say >10µF, then voltage drop of audio signal across cap would be so small, that unlinearity of cap would not play any relevant role. No relevant increase of harmonic distortion would occur.
Providing 10µF with a ceramic capacitor is impossible. Even an MKT type would be too big.

Kai