Power Budget

>From Ben Hadj:

>I'm using NEC2 to calculate the Electricfield radiated by a PCB card.
>in the output file, the POWER BUDGET is :
.................

When I ran the model through a checker it found a number of apparent
errors or "strange segment configurations", including segments that
intersect at midpoints (first occurrence: segments 4 and 25) and
parallel, overlapping segments (first occurrence: segments 43 and 52).
Segments can only connect at segment ends.

Even with such errors eliminated it is possible to get negative efficiency
from a model of this type due to other accuracy problems. For an
electrically small problem the radiated power will be very small, so
there will be little difference between input power (.5*V*Conjg(I) at
the source) and loss in the loads. Radiated power in the Power Budget
is computed as (input power) - (loss in loads), so a small error in
either of these quantities may result in the difference going negative.
Clearly the result is meaningless. Accurately modeling the source is
usually the most difficult part of constructing a model (lengths of
adjacent segment should equal source segment length). A load is just
a voltage source proportional to current, so the source modeling rules
apply. Electrically small structures, and especially small loops, can
result in accuracy problems in the model.

If you cannot get accurate results from the Power Budget, you can get
the radiated power by integrating the radiated field using

RP0,181,361,1002,0.,0.,1.,1.,

The resulting "average gain" is radiated power dived by NEC's computed
input power. Over PEC ground use

RP0,91,361,1002,0.,0.,1.,1.,

and divide average gain by 2.0 to get efficiency. It is good to compute
average gain anyway as a check. Larger angle increments can be used if
the radiated field does not have too many lobes. This integral is a
variational form for the radiated power, so the result should be accurate
for the given input current. Relating the input current to input voltage
depends on the accuracy of the source model, but the input power
(.5*V*conjg(I)) may be reasonably accurate, even if P_in - P_loss is not.

Jerry Burke
LLNL
Received on Tue Apr 16 1996 - 17:32:00 EDT