1

I understand what decoupling is needed, but I did not understand how it is applied.

For example, in this scheme that I found, there are many capacitors at the output. (Is a DC-DC converter).

enter image description here

Some cap are unpolarized, and other polarized. The total capacity seems to be 680uF.

However, I did not understand if decoupling occurs only if the capacitors are installed in the way the producer done the scheme, or that is only the minimum capacity, and a greater capacity is always better.

Northumber
  • 127
  • 1
  • 7
  • 3
    As a note, be aware that in many schematics many/all decoupling caps are placed closely together in some corner to get them out of the way, which does absolutely not reflect their placement on the physical circuit and seems to cause confusion again and again ("why are there 15 different caps placed right next to each other?!") – JimmyB Jul 31 '17 at 09:56

3 Answers3

2

High-capacity capacitors typically are polarized (electrolyte) caps. Those have the advantage of being cheap and small per capacity, but at the cost of having a high effective resistance for high-frequency currents.

Thus, although there's potentially a lot of stored energy available in a electrolytic cap, the moment that a device suddenly needs more current, it can't be drawn from the cap.

Thus, a smaller, but lower-ESR, and hence "faster", capacitor is usually used in conjunction with a large polarized one.

Whether or not that is necessary or not of course depends on your device's dynamic properties, on the stability of your power supply, on trace layout and robustness to voltage drops. So, there's no general rule of thumb other than "if you need as much capacity that you'd use a polarized cap, it's usually necessary to also have a ceramic bypass cap". The actual demand must be calculated, or, more realistically, just done by experience or reference schematic.

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
2

Don't confuse power supply output capacitor with decoupling. Although both are on same nets in your schematic design, they are there for different reasons, so they are placed differently and their value is different.

PS output capacitors are there so the power supply can have sane bandwidth, and it will not have to react to fast changes of either input voltage or output current. This is why the capacity here is defined (mostly) by the power supply control performance and the load.

Decoupling capacitors are placed right near your silicon IC. If an IC requires a pulse of current (for example, 30% of flip flops are triggered on clock), this pulse will come from the capacitor, preventing high AC currents in plane or traces of the PCB. Usually those capacitors are relatively small (0.1uF is most used, but not only). They are small so they also have low ESL/ESR, otherwise there would be no effect.

Another case to think about- analog amplifier receives something on input that causes high output current. If you forget the decoupling capacitor, due to PCB trace inductance the voltage will dip, so suddenly the amplifier will not do what it has to.

So those decoupling caps are usually put on each VCC pin of an IC.

Additionally they usually put another big capacitor (maybe 10uF) in parallel with the small ones, so it stores enough energy for longer current pulses.

  • I (partially) disagree with "they are there for different reasons". And in your answer you too write that both the PS caps and decoupling caps are there to handle pulsed currents. – JimmyB Jul 31 '17 at 09:47
  • @JimmyB yes, but if you look at them differently during the design, you get both. Anyway "current pulses" is a slogan, for design you need numbers. –  Jul 31 '17 at 09:50
  • By the way, placement is different in any case... –  Jul 31 '17 at 10:26
1

When placing decoupling caps, there are two main roles they play:

  1. Providing stored charge via capacitance
  2. Low/High frequency filtering

The bulk capacitance is largely provided by polarized electrolytic capacitors (usually the large 'can' capacitors with a high value near a regulator or power supply). These perform the function of providing current to the system when it requires a lot in a short space of time (when switching relays etc.). They also filter some of the low frequency noise from the system by providing it a low impedance return path.

The high frequency bypass caps (low value tantalum/ceramic), when placed as close as possible to an IC power input pin to limit trace inductance, provide a low impedance return path for high frequency noise to leave the system (gets rid of some of the ripple seen on a supply input). They also provide a small amount of current that can be supplied quickly (quicker than an electrolytic) when a transient pulse occurs.

These effects are only apparent when the decoupling capacitors are connected to a low impedance return path (e.g. directly connected to a copper fill ground plane).

I use this app note for reference: http://www.analog.com/media/en/training-seminars/tutorials/MT-101.pdf

And I found this answer for a related question very informative: https://electronics.stackexchange.com/a/15143/141400

D. Coleman
  • 21
  • 1