One question i need to ask, why don't chip manufacturers include a decoupling capacitor between VCC and ground on the chip itself saving us from using a decoupling capacitor?


That's a really good question, Mahmood. I think it's not easy to implement 100nF directly on chip. Some hundreds of picofarad is possible, but 100nF? It's a guess...

But, on the other hand, I know of some GaAs chips working in the GHz range, where such decoupling capacitors were implemented; not on die, but on a common substrate, like a mini mini SMD circuit. Decoupling capacitors were less than 1nF, which is quite acceptable in the GHz range. I know this very well, because these chips were fabricated at my university in cooperation with 'Frauenhofer' institute for the ATLAS project (large hadron collider) at CERN.

I think pentium and 'friends' do also have some decoupling capacitors directly on chip. When I lately mounted an ATHLON processor I believe to remember that there were some decoupling capcitors mounted directly on package.

Did you ever work with video amplifiers and operational amplifiers with bandwidths of several 100MHz? If you want to observe output signal with an oscilloscope and you connect 10X probe to the output even unavoidable probe capacitance of only 10pF will immediately cause unstability, unless you have adequate power supply decoupling capacitors connected. In these applications perfect decoupling is an absolute must.

Kai