parallel capacitance

They are for localised decoupling of noise...
The 10uF will act as a reservoir and remove low/medium frequency noise, but electrolytics are not so good at high frequencies, so the smaller value ceramics are there to fiter out high frequency nise e.g clock noise and switching noise from logic and microcontrollers etc. This sort of thing is vital for several reasons.

1. To avoid glitches.
2. To give good stability to any onboard oscillator.
3. Helps improve immunity EMC performance... e.g it reduces susceptibility to conducted and radiated noise and also reduces emission of conducted and radiated noise.
4. The little capacitors look cute (I like little blue Polyester ones myself .)

Del

Hello AH_WA70

The larger electrolytic capacitors are there to filter out 50Hz or 60Hz (Mains frequency) 'hum or ripple' from the Mains, or whatever the Power Supply supplies, because at Audio Frequencies up to 50KHz or thereabouts they have a low impedance.

Please note that at Radio Frequencies, the wire axial 'leads' or connecting wires of electrolytic capacitors have increasing impedance as the frequency increases, thus at 100+ MHz they are sometimes effectively several hundred ohms impedance, but at DC those same 'leads' or connecting wires may measure just a few thousandths of an ohm resistance.

Please note Impedance is AC resistance which varies with frequency, and is always non-DC resistance.

In the design of electrolytic capacitors, they are normally a pair of metallised flexible plates, insulated between, wound in a circular pattern round a mandrel, which mandrel is removed before the capacitor goes into a sealed can.

The much smaller disc ceramic capacitors are there to filter out high-frequency 'noise', 'spikes', 'hash', or 'interference' at 50KHz and above, from the Power Mains, or whatever the Power Supply is connected to, because they have a low impedance at Radio Frequencies.

Being of flat disc construction, the discs do have a much lower impedance at the higher frequencies.

Capacitors are often placed in parallel to help eliminate this 'hum', 'spike', 'hash', or 'interference' from disrupting a power supply which is required to have an extremely smooth waveform, and also very accurate voltage regulation, such as in Computers.

Each capacitor also has a resonant frequency, and by placing several in parallel, the problem of resonance at a particular frequency is overcome, because even though the capacitors may be marked as equal capacitance, there are always variations, even in the same batch of manufacture.

Back in the 1950's, there was not the same need for disc ceramics, (They had yet to be invented), and paper or mica dielectric type capacitors were used at higher frequencies, but in those far-off days there was also not the prolific use of electronic equipment, which caused "irregular waveforms" to be transmitted via mains power wiring, also Frequencies in common use were very limited to the lower portion of the Radio Spectrum.

These days, with Microwave Ovens and Variable Speed Drive equipment such as Washing Machines etc, in homes, the usage of such parallel capacitors is a very cheap remedy.

In many of today's Commercial or Industrial installations, the electrical 'noise'. hash', spike' or 'interference' problem can be very large, and requires careful design of filters to prevent one equipment item, or a part thereof, from interfering with proper operation of other equipment items or parts thereof.

Trust you now feel assisted.

It appears that Del the cat has been busy and Posted above, while I was composing my answer to your question.

Kind Regards....

Just a bit more (possible) clarification . I'm assuming you're looking at the circuit diagram - showing these caps directly on the incoming power supply rails (or after the regulator, if there is one).

On the physical lay-out, they'll be scattered around. The ceramics will probably be located adjacent to switching components (e.g. logic ICs) to cope with the current surge when they switch, while the electrolytics will be spaced - probably one close to the incoming supply, and another near any large load (connector to motor etc.) to provide a reservoir when the load is turned on.

This is my best guess, given that we don't know what the circuit is doing.

If you're talking just power supply, then the purpose is to filter the noise from the supply - especially if it's a switching power supply. They can also add to the stability of the power supply. The frequency response of ceramic caps differs from electrolytics, as mentioned in the other posts. But also, among ceramics, the frequency response differs according to the value of the capacitor, among other things. To ensure the best possible frequency coverage it's a good idea to mix values of ceramics.

I prefer to think of it in terms of switching current rather than noise. When a logic gate flips, or the output of a high speed amp changes, that requires a little bit of current. The wires or etch that connects the device to the power supply are inductors. When you try to instantaneously pull that bit of current through an inductor, it responds by creating a voltage spike. It's these voltage spikes that cause noise. They can result in ground bounce, where the local potential of the ground increases to a point where transistors that were biased on now turn off. They can also appear on the output of the component, or cause oscillation in an analog circuit such as a comparator.

Typically you'll want to sprinkle the ceramic capacitors around the board, mounting them as close to each devices power/ground pins as possible, and vary the values. For instance, on a large device with 10 power pins, you might use one 3 0.1uF, 4 0.01uF and 3 0.001uF, with a single 10uF nearby.

It seems counterintuitive to use lower values rather than higher values for decoupling caps, but larger value caps come with larger parasitic inductances, so a .01uF can be better at sourcing current at higher frequencies than a 0.1uF.

Also, package style makes a big difference. Leaded components have much more inductance than surface mount components, and so are much less effective at either filtering noise or supplying current.

As for having 10 0.1uF capacitors on you design, I suspect it is either one of two things. It may be they needed 1uF and couldn't find it in the package / dielectric / voltage rating that was required, so they used 10x 0.1uF. If you're looking at a schematic of a board with both supply and circuitry, sometimes all the decoupling capacitors will be drawn in one place - at the power supply, even though they are actually scattered around the board at individual components.

For a thorough understanding of bypass capacitors and board layout go to Howard Johnson's superb website:-

http://www.sigcon.com/

About High Speed Digital Design and Signal Integrity. Although he specialises in digital design most of the information about decoupling capacitors applies equally to analogue.