I done some testing projects with a microcontroller on a bread board, all wires were flying all over the place and no decoupling capacitor, guess what, it worked every single time.
Can you explain this?


Dear Mahmood,

yes, it's mostly working, but if you go to a CE tester your bread board design will fail the radiation test very probably!
If you neglect decoupling capacitors and use flying wires, you get enormous large radiation loops emitting electromagnetic radiation. On homepages of philips and murata there are some nice application notes demonstrating, that with even only one loop excited by 74HCMOS gate emission of electromagnetic noise is so high, that you will fail according CE test. The only remedy is to use decoupling capacitors and solid groundplane.

It's interesting to note, that some manufacturers of mainboards do not locate decoupling capacitors next to each chip. But then they use multilayer PCB!! Then, due to extremely small inductivity of the arrangement of very closely separated VCC and 0V planes, decoupling capacitors may be located at some distance to chips. But nevertheles, decoupling capacitors are there!

I want to focuse that decoupling needs can be very differently. A board with extended digital buses leads to very high capacitive loading. In this case the need of adequate power supply decoupling is extreme. But if you have your micro in a stand-alone application, situation is less critical.

But for the unprofessional it's hard to find out, how much decoupling is necessary. Then it's a good design practice just to use a decoupling capacitor for each chip.

Kai