You did not mention how the S11 data was generated and what it was
supposed to represent.
From your question I infer that it is supposed to be data for a uniform
interconnect (trace or cable) and that is why you expected a straight
line with the characteristic impedance value.
In practice, there are multiple possibilities to get non-flat TDR
response when otherwise we would expect a flat line.
The most typical reason is losses, but it can tilt the response both
ways: series conductive losses tilt it upwards, dielectric losses will
tilt it downward. In a real situation we have the two simultaneously
present. If the data is measured on a pcb trace, manufacturing
tolerances of dielectric heights and trace cross section and material
property variations along the interconnect can create tilts in either
way. But it is true that in these days the most typical tilt is positive...
Regards,
Istvan Novak
Oracle
On 9/23/2015 9:10 AM, jun zhang wrote:
Hi experts,
I observe a phenominon about TDR obtained by fourier transform of S11. I
notice that it gradually increases along the time axis. What does this mean?
It seems that the impedance at each position is related to, but not
indepedant of that at other position. Characteristic impedance is unique.
so can TDR not reflect characteristic impedance? Or early time TDR still
can reflect characteristic impedance? A designcon paper mentions the
phenominon and find the cause is frequency-depend loss. So I think this is
a intrinstic drawback of TDR. It can't show accurately the impedance at
each position of the link path.
Do you agree with me?
