3
\$\begingroup\$

This question already has an answer here:

Having different types of capacitors (electrolytic, film, silicon, tantalum, etc.), which types would be the optimal choice for polar/non-polar pair in this common configuration:

enter image description here

I understand the reason for having two or more caps of different capacitance values. Though wonder, why does the pair includes a polar and a non-polar capacitor? Is it essential to include a polar capacitor, or it's OK to use two non-polar caps?

The application is for bypassing power supply for opAmps, ADCs, DACs, microcontrollers, and other low power ICs up to 100MHz bandwith. I know that electrolytic ones would be too big for surface mounting apps.

\$\endgroup\$

marked as duplicate by tcrosley, Daniel Grillo, Leon Heller, Chetan Bhargava, PeterJ May 16 '14 at 0:01

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • 3
    \$\begingroup\$ The usual reason to use a polarized capacitor is because you cannot get a non-polarized capacitor with the capictance and WV rating you want. Old designs will show polarized capacitors for 10 uF or even 1 uF because ceramics couldn't historically reach those values without other compromises. \$\endgroup\$ – The Photon May 15 '14 at 18:57
  • \$\begingroup\$ @ThePhoton And now I can easily get 22uF non-polar ceramic cap. However, current designs still make schematics with '+' sign. Is it just an old habit, or I am better to use a polar one? \$\endgroup\$ – Nazar May 15 '14 at 19:24
  • \$\begingroup\$ For a 22 uF ceramic cap be sure to check the temperature range and variation. Also, large ceramics (1210 and higher) can have reliability issues due to board flex or vibrations. So you might still prefer an electrolytic in some situations. Also compare prices, of course. \$\endgroup\$ – The Photon May 15 '14 at 20:17
  • 1
    \$\begingroup\$ @Naz Both caps being non-polar would work too. Usually, a 22uF tantalum would be about the same mechanical size as 22uF ceramic. 22uF ceramic might be somewhat more expensive. Here's another point. It seems like you're decoupling a regulated +5V supply. In such case, a 25V rating for the cap is an overkill. Larger voltage rating increases the mechanical size. Typically, caps with 2x the max voltage (which would be 10V in your case) are used. \$\endgroup\$ – Nick Alexeev May 16 '14 at 21:58
3
\$\begingroup\$

As was explained in the "possible duplicate" link, the two capacitors serve different frequency regimes: the polarized cap, though larger, responds to lower frequencies, while the smaller, nonpolarized cap handles much higher frequencies (or, if you prefer, longer and shorter transient current demands). The polarized capacitor is typically either an aluminum electrolytic or a tantalum capacitor. The non-polarized is usually a ceramic type.

In a previous thread, I recommended tantalums for the polarized cap, and still do. For a given voltage and capacity they are smaller and have better frequency response. My recommendation was criticized on the basis that, if improperly selected or installed, tantalum caps can explode and injure an experimenter. I've burned out my share of tantalum caps, and they smell godawful, but I've never had one explode.

Also,

electrolytic ones would be too big for surface mounting apps.

is simply untrue. For the values shown above, for instance, Digikey reports 178 choices in SMT.

\$\endgroup\$
  • \$\begingroup\$ I understood that large capacitance better handles lower frequencies, whereas smaller capacitance filters higher frequencies. But does one of the caps have to be a polar one, or they both can be non-polar? \$\endgroup\$ – Nazar May 15 '14 at 19:59
  • 2
    \$\begingroup\$ Up to about a decade ago, for medium (1uF - 100uF) caps, tantalum was king, with aluminum electrolytics a decent second. This is probably no longer true, and you can use large non-polarized ceramics instead. There is some evidence that ceramics, by their nature, are less tolerant of mechanical stresses than the alternative, but this should not be a factor for what you're doing. \$\endgroup\$ – WhatRoughBeast May 15 '14 at 20:54

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