thanks for reply. one question still remains unanswered. if multiple starpoints (or mecca) are used what are the other things to do to nullify the effect.

i used to do a heavy decoupling (1UF) when ground line becomes long or VCC/VDD is long, or .1 uf decoupling when lines are within the limits.

Hallo Abhishek,

the methode of local ground points means following:

Assume you have a big enclosure with much circuitry in it. Assume further, that you cannot implement one big PCB with solid ground plane. May be that you must use the volume more intelligent, by using lots of subboards, mounted in such a way, that total board area can be extremely increased. Then methode of local ground points assumes, that ground potential of all those subboards are not expected to be identical, although they should, of course. Just think about routing a digital signal from one subboard to another. Due to voltage drop accross inductivity of cable, there is a certain potential difference between these two subboards now. It's called common mode noise or simply ground noise.
Each subboard is now treated, as if it was totally alone. Means, each subboard has it's own solid ground plane with standard power supply decoupling etc. But connections between subboards are fabricated in such a way, that (1.) common mode noise is reduced and (2.) susceptibility to common mode noise is also reduced. This can be done by symmetrical signal routing with additional slew rate limiting and perhaps some common mode filtering (common mode ferrite choke) and by the use of LRC filtering for DC connections from board to board. Aim of these measures is to reduce common mode currents along ground connections from subboard to subboard.
Symmetrical routing for digital signals can be done by RS485 transcievers with slew rate limiting, like MAX487. We just discussed this topic in another thread. Symmetrical routing for analog signals uses concept of instrumentation amplifier at recieving side for suppressing common mode noise
How to design LRC filters? Assume your subboard has some of these 100nF X7R capacitors shared over the PCB. Following a rule of thumb you will also have a big electrolytic capacitor at power input of subboard, something like 47µF/25V or so. If you now insert a ferrite bead like BL02RN2 from MURATA into Vcc connection comming from power supply unit subboard, then, if your subboard tries to draw high frequent supply current from PSU subboard, it's blocked! High frequency supply current then can only be drawn from decoupling capacitors of subboard, but not any longer through Vcc connection from subboard to power supply subboard. The reason for this is, that ferrite bead becomes high ohmic for high frequencies: 15Ohm at 1MHz, 30Ohm at 2MHz, 60Ohm at 5MHz, 100Ohm at 20MHz, 130Ohm at 100MHz and 160Ohm at 1GHz.
Of course, if high frequency supply current through Vcc connection from PSU subboard is reduced, from Kirchhoff's law, ground return current is also reduced:





BL02RN2 looks like a 2.2µH inductor at low frequencies. BL02RN2 acts in combination with 47µF electrolytic capacitor like a two pole low pass filter with corner frequency of about:

fc = 1 / 2 / PI / SQRT(L x C) = 15.7KHz

Ringing is suppressed when:

R >= SQRT(2 x L / C) = 0.3Ohm

Well, there can no R be seen in drawing. It's hidden in electrolytic capacitor. Equivalent series resistance of 47µF/25V electrolytic capacitor is in this range. So, an additional R component is not neccessary.

Bye,
Kai