Teledyne LeCroy VirtualProbe - Manual (Page 15)

Operator's Manual
l
All frequencies are assumed to go from the first frequency listed to the last frequency listed with
constant frequency spacing.
S-parameter File Interpretation and Usage
S-parameters provided to the virtual probe will tend to have various frequency scales, differing in both
resolution (frequency spacing) and range (start frequency and end frequency). The VirtualProbe com-
ponent will resample all S-parameter sets onto a new frequency scale dictated by the system sample rate
(half the sample rate will be the last frequency) and the time length (see Time Length) – where the time
length dictates the new frequency resolution. In performing this resampling, it is important to under-
stand that the resolution may be higher or lower than the original resolution and that the frequency
range may require extrapolation of points. When the frequency range is higher than that provided in the
S-parameter data (i.e. the S-parameters are not provided to half the system sample rate), then the S-
parameter data will be set to zero at these extra required data points. Furthermore, most S-parameter
data does not go to DC, because VNAs don't measure data to DC, and therefore the DC response is also
extrapolated by setting the low frequency points equal to the first frequency provided in the S-parameter
data.
What this means is that it is helpful to follow some simple rules that will maximize the effectiveness of the
virtual probe.
l
Take S-parameter data as low in frequency as the VNA allows to minimize DC extrapolation errors.
l
Take S-parameter data up to one half of the system sample rate. In other words, if you are going
to supply waveforms to the virtual probe at 40 GS/s, it is good to provide data up to 20 GHz. Other-
wise, if the frequency content of the signal provided to the system is much lower, then it is unnec-
essary to sample at a high sample rate anyway.
l
Take S-parameter data with sufficient frequency resolution. The resolution that needs to be pro-
vided depends on the electrical lengths of the circuit elements involved. In other words, if you
have a cable that is 5 ns long, and there are reflections going on in the system, then taking data at
maybe one tenth the minimum frequency resolution of 1/5 ns = 200 MHz (like 20 MHz) is a good
idea. Said differently, in a more practical manner, if there are lots of bumps and wiggles in the S-
parameters that are narrow in frequency, then you will need to measure S-parameter data with an
appropriate frequency resolution to capture these bumps and wiggles. Usually the narrowness of
the bumps and wiggles becomes smaller when the electrical lengths of circuit elements becomes
longer.
921148 Rev A 13