RE: NEC-LIST: Definition of Gain

Robert S. Dixon wrote:

> > In my anechoic chamber, I measure the performance of my horn using
> > a network analyzer. That network analyzer's impedance is fixed at
> > 50 ohms.
>
> But is it really? How can one know how the power output of any
> signal source varies with load impedance?

The simplest, most dependable way is to use a good-sized attenuator
(10 dB or so) in the line as close as possible to the antenna under
test. The same pad is in line while testing the standard-gain
reference antenna. This masks the impedance effects of the AUT, as
well as keeps LO mixer current out of the antenna. This is probably
not necessary with a newer network analyzer system, but it is
necessary with the older remote-detector SA systems, however. Yes,
without a good match at at least one end of the line, standing waves
can distort gain measurements quite a bit.

Different signal generators and transmitters may vary widely in this
regard. If the definition of gain includes these effects, then the
gain of the antenna will depend on the kind of transmitter connected
to it.

There is no question that realized gain definitely will depend on the
transmitter (or receiver) impedance. The goal is to make measurements
that allow us to accurately predict system performance from component
data. Antenna theory tends to be expressed in terms of Z parameters,
but gain test equipment tends to measure S parameters most directly.

> > I cannot complex conjugate match the antenna for each frequency
> > point. I could measure VSWR, and then back out the impedance
> > mismatch loss from the gain mathematically, but the fact is that no
> > one I know does this.
>
> Nevertheless, I feel that publishing VSWR curves vs frequency for
> any antenna is important.

Sometimes it is difficult to measure the impedance of an AUT in the
exact gain test conditions. A sharp resonance may shift a bit between
VSWR testing on the ground and when the antenna is up on the pole. A
chamber pushed beyond its limits can effect the impedance of a large
antenna as a function of height and even angle. These effects are
recognizable when looking at the raw S-parameter data from a range
using good fixed-impedance ports. Processing gain data with flawed
VSWR data would lead to errors that might not be so easily found. A
resonance shift would be immediately recognizable from the raw
S-parameter data, but if it were processed before viewing into
IEEE-standard gain format the results could show a peak gain much
higher than that which really exists.

If the end-system performance (transmit power or receive sensitivity)
is stable when operating into the expected range of antenna impedance
values, e.g. if the system includes a circulator, then the importance
of antenna VSWR is primarily as a diagnostic tool to flag unit-to-unit
variation. Antenna VSWR has no direct effect on system performance in
this case. Even if there is some transmitter or receiver performance
variation as a function of input impedance, it is a usually a
secondary effect in a reasonably well defined system. Cable losses
tend to mask antenna VSWR effects even further. Link analysis is not
quite as simple as adding the gain, cable loss, and transmit/receive
S12 values, but it's pretty close when the system is decently matched
at at least one end. More importantly, the S-parameter model reduces
the sensitivity of overall system performance predictions to the most
likely measurement errors.

> > They report what is currently defined as "realized gain", i.e. the
> > gain including mismatch loss. Current industry practice reports
> > realized gain, but they call it "gain". It is not in synch with
> > the current IEEE definition.
>
> I am not aware of that industry practice, but if it is true, then it
> seems that people are trying to oversimplify the situation.

This is probably true from an academic viewpoint, but it gets the job
done with a managable level of effort and risk.

                David de Schweinitz
Received on Fri Mar 10 2000 - 06:18:54 EST