Generally, how are capacitors picked? (Applies to radio, MCU's, digital, analog)

Hi Thomas,

They store a certain amount of charge when you stick a voltage across them, but that is much like describing a glass window as "made of silicon dioxide", which while true at a high level, doesn't explain glass nor windows in much detail.

In electronics, there are certain fundamental circuits that can be created using capacitors; these fundamental building-blocks become parts of larger circuits like computers, radios, etc.

Here are some building blocks which can all be created with capacitors and other components:

1. Decoupling

2. Filter

3. DC blocking (related to filter)

4. Tank circuit

5. Timing (often used for resets)

A computer will make extremely extensive use of the first block and #5, whereas a radio will make use of #1,2,4 and an amplifier will make use of #1,2,3.

There are formulas which are used to calculate the preferred value for the above building blocks. The wikipedia pages for them will hopefully contain the formulae.

Note that since electronics relies on the physics of materials, capacitors can behave in a non-ideal way limited by physics and therefore some designs will use certain types or even certain shapes of capacitors, and made of certain materials (cost-performance tradeoff). It can get very advanced; for example, here is a PDF document just dedicated to decoupling. (AVX is a manufacturer of capacitors, so expect them to have lots of useful information).

Some of the other topics require using complex math, depending on how much detail you want to go into. At a high level, it is possible to state things like "a filter will allow certain frequencies to pass". One nice way to think of a capacitor (again, high level) is as a frequency-dependent resistance (actually impedance, but that requires complex math). Again, from that high level, you would not really try to convert that into the building blocks listed above; those building blocks are like well-known templates of designs, so you can directly use the appropriate formulae for calculating the component values.

If you want to be enlightened without any deep math, and want practical guidance, a very good book is 'The Art of Electronics'. If you want the math too, then I can recommend some university-level books, but they are definitely not easy-going and I don't want to scare you ; ) Personally, I'd suggest The Art of Electronics, it is very easy to read.

If you want to experiment, you could try a simulator. Here is a simulation of a filter. You can alter the component values if desired. Notice when you run it, the input signal frequency slowly increases. You can observe the input signal, and the output graphs at the bottom, dynamically.

EDIT: here is a practical example of building block #2 (filter).

My expertise is digital, where capacitors are mostly used for bypass.  The purpose of a bypass capacitor is to keep the supply voltage for an IC constant.  The problem with most digital ICs is that the amount of current they need can change very rapidly, especially if multiple inputs change at the same time.  For example, if 32 IC outputs all switch from 0V to 3.3V simultaneously, the IC needs to get a bunch of charge from somewhere quickly.  Now it could try to get the charge from the power supply, but it's far enough away that it will take a while to get the current to the IC.  Meanwhile, the IC will pull charge from whatever is near, which causes the supply voltage of the IC to dip.  If it dips too much, it will malfunction.

So what the bypass capacitor does is provide a reservoir of charge right next to the IC power pin so that it provides the charge and minimizes the dip in supply voltage.  To calculate the amount of capacitance you need, you add up the amounts of charge needed to change the outputs.  Usually nobody bothers with the calculations and instead uses 0.1 uF or 0.01 uF capacitors everywhere.  Parts with lots of voltage supply pins (usually called Vcc or Vdd) often tell you in the data sheet what values are recommended, e.g., 0.1 uF ceramic cap for each Vdd plus an additional 10 uF bulk capacitor.  Also, many manufacturers provide a reference design board or schematic and encourage you to copy the capacitor values from that board.

One thing you have to be careful about with capacitors is that they don't just have capacitance: they also have resistance and inductance.  Typically those little 0603 SMT 0.01 uF ceramic caps have very little inductance so they make the best bypass.  However, they can't store much charge so you add larger 10 uF multi-layer ceramic caps to provide a second reservoir.

There are also large electrolytic capacitors.  They're slow, but can store a lot of charge... unless they get too warm and the electrolyte boils off making them useless.  This happens a lot with cheap PC motherboards.

Capacitors also have a working voltage.  You usually want to have twice the highest voltage you expect to get good margin.  Higher voltage capacitors are physically larger, and may have more inductance.

Analog capacitors are a huge subject which I know very little about.  If I need to do something analog, I try to copy the reference design exactly, including the same types and values of capacitors.  I leave the differences between polyester and mica and paper to others.

In his explanation John Beeten wrote about changing the output Voltage of an integrated circuit. The inputs of the circuits connected  to those outputs are the real need for a decoupling capacitor. The output of a C-mos digital circuit acts as a switch, connecting the output pin to the supply pin or to ground. The load is mostly capacitive. With fast components,  the inductance of the trace to the supply cannot be neglected. A capacitor close to the supply pins of the driving IC will charge the load when the output goes from low ho high. Without this capacitor, the supply Voltage at the driving IC will drop when the polarity of the output changes. If this IC is a gate, this will cause a delay. If this IC is a counter or a memory, the drop in supply Voltage can cause loss of the content stored in the IC. The value of the decoupling capacitor must be more than 10 times the input capacity of the inputs, connected to the outputs of the IC. In general ceramic capacitors with a value of 0.1 μF are used for this purpose. Capacitors also have inductance.  Larger capacitors have larger inductance, as a result if their mechanical size and construction. Properties of a capacitor are not only value and maximum Voltage. Inductance, current rating, ESR(Equivalent Series Resistance), accuracy and stability are sometimes important. For critical decoupling over a wide frequency area, often several capacitors with different type and value are used in parallel. For example  a 0.1 μF ceramic capacitor, a 1 μF polyester capacitor and a 1000 μF electrolytic capacitor. For decoupling purposes, accuracy and stability are not important. In tuned circuits and timing systems they are. In switched mode power supplies some capacitors must handle high currents at the switching frequency. Here a low ESR is important. J. van den Berg