In your hypothetical, but not quite practical, case you are correct: the
time and frequency domain noise numbers would be the same. The kind
of frequency dependency you suggest, however, is very extreme.
It would require infinite selectivity, which does not really exist in
practical circuits.
You can come close to it by using high-Q elements.?? If, on the other
hand, your
question was about uniform,?? distributed interconnects, like two traces
next
to each other,?? the situation is different, because the frequency-domain
coupling
in that case is not constant, it is frequency dependent.
Regards,
Istvan Novak
Samtec
On 5/27/2019 12:55 PM, Tramp wrote:
Dear listers,------------------------------------------------------------------
I'm wondering if time domain coupling coeff. must be equal or lower than
the freq. domain coupling? Assume in FD (Freq. Domain) at 100MHz the
coupling coeff. is x%, and zero at all other freq.s. Now if in TD (Time
Domain) the aggressor present a sine wave of 1Vpp@100MHz, then the TD
coupled noise will be 0.0xV@100MHz, right? If the aggressor contains any
freq. components other than 100MHz, then those additional freq.s will not
be coupled to the victim, hence the coupling coeff cannot be greater than
x%. Is above inference correct?