This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Decoupling

HI,

From most of the TI reference design for MCUs, I found numerous decoupling capacitors next to power pins.

There are two questions about decoupling capacitors.

1) Value of the capacitor

Most of the decoupling capacitors have value of 100nF.

Is there any specific reason that most of MCU use 100nF as decouping capacitors?

2) Replace several capacitors with one large value capacitor

In case of extremely size-constraint design, it would be great to replace several 100nF decoupling capacitors with one large value capacitor to save space.

If the replaced capacitor had large enough value, would it be OK for stability of entire circuitry?

If so, could ten 100nF capacitors simply be replaced by one 1uF capacitor?

Thanks in advance

  • It is the high frequency response that makes it necessary to put by-pass capacitors near the Vcc and Vss pins.

  • Woosuk Lee said:
    If so, could ten 100nF capacitors simply be replaced by one 1uF capacitor?

    No. Because single 1uF capacitor most probably will have higher ESR at high frequencies than ten parallel 0.1uF capacitors and you can't place single capacitor close to multiple supply pins. PCB traces have some resistance too. It's all about low ESR power storage as close as possible to power supply pins.

    There's many resources about bypass capacitors around on internet, I find this one good enough:

    http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf

    Well, depending on PCB design you can omit some decoupling capacitors - if you know what you are doing.

  • Woosuk Lee said:
    Is there any specific reason that most of MCU use 100nF as decouping capacitors?

    Yes. It is an optimum value for capacitance (the higher the better), ESR (the lower the better), size (the smaller the better) and price (the cheaper the better).

    Since everyone uses this value, the 100nF are usually the cheapest. Also, larger capacitance usually requires a larger package (thicker while same footprint or larger footprint while same thickness) while the smaller capacitances usually use the same package sizes.

    Woosuk Lee said:
    In case of extremely size-constraint design, it would be great to replace several 100nF decoupling capacitors with one large value capacitor to save space.

    Once we had someone external makign the layout. Since in the schematics all decoupling capacitors were palced in one corner with the supplies of the chips, he placed all caps on one space of the board. Well,m thsi was really easy to route, but if this would have been the plan, we could jsu thave picked one large capacitor.

    The reason for decoupling capacitors is to decouple the local current consumption from the supply. So that no (or reduced and lower frequency) current ripples appear on the supply voltage lines. Decoupling capacitors shall provide a local current source that is not limited by the impedance of the PCB traces. This avoids local breakdown of supply voltage and also prevents that the supply traces become antennae.

    Woosuk Lee said:
    If the replaced capacitor had large enough value, would it be OK for stability of entire circuitry?

    No. Current (electrons) is slow. Even slower if they have to crawl through lengthy PCB traces or wires. So a local storage right at the supply pins is required to keep the voltage stable. Or else supply voltage will break down locally even though the big capacitor at the supply is full to the rim.

**Attention** This is a public forum