Why Caps in Parallel?

[AndrewCE]

already posted this on ProjectGuitar, sorry for the redundancy, but...

I've seen a few schematics that use caps from a DC power supply to ground, to smooth out the power supply, but on some schematics there are 2 or 3 caps in parallel. They would have a 47uF cap in parallel with a 22pF cap for example. I know that this gives a slight boost in capacitance (parallel caps are added) but is there any other reason to do this? It seems like all the AC ripple would go through the 47uF and then the 22pF would be useless.

On some schems, the 22pF would be a different material, like metal film as opposed to electrolytic.

Why 2 caps instead of one?

[Johan]

sometimes a big cap is parrallelled with a small value cap becouse the smaller cap is better at capturing transients... in a powersupply, that could mean cleaner power.
in high end audio where better than 20Hz - 20kHz-or more might be necesery, it can mean better high end responce ( remember, Rupert Neve build his mixers to be flat up to 50kHz even thou we can only hear op to 20kHz..in best cases)
for guitar, it probably doesnt matter...
j

[andrew_k]

In the PSU, little cap cleans up RF, big cap smooths ripple... as is my understanding anyway

[zyxwyvu]

The goal of the capacitors in a power supply is to have a very low impedance at all non-zero frequencies, so only DC will be present at the output. Electrolytic capacitors typically have pretty high internal resistance and inductance, which makes them not as effective at higher frequencies. To counteract this, you parallel multiple capacitors. Film or Ceramic capacitors work much better at high frequencies than electrolytics, so they are often used. You can also parallel multiple electrolytic capacitors to simultaneously increase capacitance, and reduce impedance (the resistances and inductances associated with the capacitors are in parallel, and so the result is smaller).

You can find some info on which capacitors are good for different things here:
http://en.wikipedia.org/wiki/Types_of_capacitor

Also, these may be helpful depending on your previous knowledge:
http://en.wikipedia.org/wiki/Electrical_impedance
http://en.wikipedia.org/wiki/Capacitor#Equivalent_circuit

[doitle]

It was my understanding you always want decoupling caps RIGHT ON the power pins of any ICs. So there are multiple caps in paralell yes but they have some real world distance between them in the actual circuit implementation. I.E. It doesn't look right on a schematic but on a PCB it will make sense.

A transient load decoupling capacitor should usually be placed as close as possible to the device requiring the decoupled signal. The goal is to minimize the amount of line inductance and series resistance between the decoupling capacitor and that device, and the longer the conductor between the capacitor and the device, the more inductance there is.
A power supply decoupling bypass capacitor should be placed as close to the voltage/current source as possible. The idea is to minimize the line inductance and series resistance between the capacitor and the supplied devices.

[AndrewCE]

The goal of the capacitors in a power supply is to have a very low impedance at all non-zero frequencies, so only DC will be present at the output. Electrolytic capacitors typically have pretty high internal resistance and inductance, which makes them not as effective at higher frequencies. To counteract this, you parallel multiple capacitors. Film or Ceramic capacitors work much better at high frequencies than electrolytics, so they are often used. You can also parallel multiple electrolytic capacitors to simultaneously increase capacitance, and reduce impedance (the resistances and inductances associated with the capacitors are in parallel, and so the result is smaller).

You can find some info on which capacitors are good for different things here:
http://en.wikipedia.org/wiki/Types_of_capacitor

Also, these may be helpful depending on your previous knowledge:
http://en.wikipedia.org/wiki/Electrical_impedance
http://en.wikipedia.org/wiki/Capacitor#Equivalent_circuit

thanks for the answer, it makes sense now. one site had said something about lowering the impedance for high freq's, but i didnt get it. guess i didnt take into consideration the internal resistance/losses of electrolytic.

[R.G.]

OK, so it's a little more complicated than that.

All real parts have non-perfect natures that prevent them acting like a textbook part. For instance, straight wires have inductance as well as resistance. As frequency rises, the resistance starts at the DC value, then starts rising as the skin effect forces charge carriers to the outside of the wire. Quite apart from the skin effect, the raw inductance starts to predominate at some high frequency and the impedance becomes the rising resistance caused by skin effect and the rising inductance of the wire.

With that as a background, you can well imagine that something as complicated as an electro cap has even more imperfections in the real world. Electrolytic caps have Equivalent Series Resistance (ESR), which is the resistance of the parts getting charge into and out of the cap. Since they are made from thin (and resistive!) foils of aluminum wound up in a roll, there is a lot of thin metal to get through to get charge evenly distributed. They also leak a little, varying from one to the next. This is the Equivalent Parallel Resistance (EPR) and it becomes a problem with power supply caps as they get old. The winding of those foils in a bundle also adds inductance, the ESL. The bigger the capacitor, the bigger the ESL, in general.

The capacitance causes a declining impedance of 1/(2*pi*F*C) until the size of the capacitive impedance equals the ESR. At thiat point, the impedance CAN'T get any lower, and the capacitor now looks to the outside world like a small resistance. Also, the impedance of the ESL is rising. At some frequency, it becomes bigger than the ESR. From that frequency on out to infinity, you now have an inductor, not a capacitor. The curve of impedance magnitude versus frequency is a V or a bathtub, depending on construction. Any one capacitor can only get so low in impedance.

So how do you get low impedances at high frequencies? By using smaller capacitors in parallel. A smaller capacitor won't be as good as a bigger capacitance at low frequencies, but at some point, the smaller cap's lower ESL will let it go lower in impedance for higher frequencies than the bigger, but slower caps.

Electro caps store a lot of charge down around DC, but they are getting inductive somewhere around the top of the audio range. If you either know what you're doing and know you need them, or have just read the buzz words from the cork sniffing elite, you will use a paralleled low ESR cap to help the poor fat electros out at high frequencies.

There is one other instance where paralleled caps is slick and fairly elegant design work. Many active filters need frequency setting caps in the ratio of 1:2. It happens that the EIA standard set of capacitor values does not have many 2:1 ratio values. It is simpler and usually cheaper where you need two of one value and one of twice that value, to buy four caps of the same value, and parallel two of them for the twice-as-big cap.

[andrew_k]

Another gem from R.G.
Thank you!

[Mark Hammer]

It is not at all uncommon to see power supplies with a big cap before the regulator, a medium-sized cap after the regulator, and a .1uf or similar right alongside or even soldered to the supply pins of chips.

[doitle]

It is not at all uncommon to see power supplies with a big cap before the regulator, a medium-sized cap after the regulator, and a .1uf or similar right alongside or even soldered to the supply pins of chips.

This is more what I was getting at. At least in digital design you see a lot of ICs with capacitors soldered either directly to or practically on top of the supply pins of an IC.

[R O Tiree]

Have a read of these 3 articles:

http://en.wikipedia.org/wiki/Decoupling_capacitor - scroll down to "Transient Load De-Coupling" and then "Placement"

http://www.hottconsultants.com/techtips/decoupling.html - a bit more in-depth...

Lastly, http://www.ultracad.com/mentor/esr%20and%20bypass%20caps.pdf for a more rigorous approach with some interesting conclusions.