Re: S-parameters of coax-to-waveguide transition

[Kevin Thomas <kjt@cray.com>]

The problem may lie in the interpretation of the results,
rather than the results themselves.  You mention that the
frequency domain result is unusually large, but do not mention
the time domain result.  I assume then that the time domain
result looks fine.

The frequency domain plots of voltage, current, and power produced
by LC are not normalized, and thus don't relect normal units
(volts, etc.).  To normalize the plots, divide by the input
source in the frequency domain.  This can be done via the
LC "Plot Calculator" or with another tool, such as Matlab.
You may want to express this result in dB (10*log10 or 20*log10,
which can also be done in the Plot Calculator).

Of course these plots can be fed directly into the S-parameter
calculation as well, which automatically normalizes the
results to the incident source.  Since each plot includes the
same scale factor, then the resulting S-parameter plot will
have the scale factor removed.

-- 
Kevin Thomas    kjt@cray.com   tel 1-651-605-9072
http://lc.cray.com/~kjt/        or 1-800-284-2729 x6059072


Brian Frank wrote:
> 
> Hi,
> 
> I've been trying to simulate a coax-to-waveguide transition that uses a
> half-waveguide-diameter probe that sticks directly into a 26 mm radius
> circular waveguide. I've been having a problem with frequency-domain plots
> of the reflected wave, in that it "spikes" to rather large values that are
> much greater than the magnitude of the incident wave (which is a nice
> Gaussian shape in the frequency domain). This occurs whether I look at the
> voltage, current, or power.
> 
> I've exciting the coax section using a current source (modulated pulse
> with center frequency of 6 GHz and maximum frequency of 9 GHz), and
> measuring the voltage and current further up the coax (closer to the
> waveguide). It's intended for 6 GHz operation, and I've tried different
> cell sizes (0.25 and 0.50 mm) with different time steps (4e-13 and 8e-13
> s). I've tried a "reference" simulation (using only the coax terminated in
> absorbing boundary conditions) as the incident wave, similar to the
> approach used in the "Analysis of a Coaxially-Fed Patch Antenna" example
> on the LC web site, but it gives the same results.
> 
> Any ideas what I'm doing wrong?
> 
> Thanks,
> 
> Brian
> 
> --
> Brian Frank
> ECE Department
> Queen's University