Scott McClements wrote:
> On 1/21/08, Jim Lux <jimlux@earthlink.net> wrote:
>
>> First off.. Most of the benefit of directional antennas is not in the
>> forward gain (as you've probably found, almost anything gets you to
>> within 1 dB of the absolute best) but in the management of what the rest
>> of the pattern looks like.
>
> *Exactly*. This is why optimizing for gain, to me, isn't so important.
> Just as you said, at worst its -1dB down from something that is
> optimized for gain (for a given boomlength). However, the "rearward
> rejection" can be anywhere from "great" to "totally useless".
>
> So in most operating situations, which form of rearward rejection is
> best. I can think of two cases where rearward rejection would help:
>
> 1. Interference, QRN or QRM
> 2. Multipath fading
>
> I think its obvious that if the QRM is coming from a certain
> direction, then a some type of deep notch, would be very useful
> because you can use that notch to help reduce interference. Ok, what
> if there is more than one station interfering?
>
> What about multipath fading? Does deep multipath fading predominantly
> come from the direction of the primary station? Or can it equally be
> caused by a reflection coming from the rear??? If so, is there a more
> likely angle that a reflection might come from (say 180 degrees to the
> main signal?)?
>
>> This is especially so
>> if you use a modeling tool that lets you put symbols and equations in
>> the model, like 4nec2. You can make the end points of the elements some
>> function of an angle, and then vary it:
>
> Actually, this is why I started this whole discussion. I spent months
> writing my own version of a tool to optimize a cubical quad antenna. I
> don't use variables in NEC2, my program works directly on the NEC
> input deck and feeds it to a NEC2 engine and reads the output deck.
>
> So now I want to understand my design choices before I start
> automated, month long runs of optimization.
>
So what you're looking for is a way to build a suitable goal function
for the optimizer. I think you might want to use some sort of
"integrated total side lobe" metric along with a "peak sidelobe" term,
perhaps with an elevation angle cutoff. These are pretty standard
metrics for things like radar antennas (ISLR =Integrated Side Lobe
Ratio) You could do some VOACAP style modeling to find elevation angles
of interest (or non-interest).
The other thing, not well accommodated by the classic sidelobe sorts of
metrics, because the typical HF antenna has a very broad main lobe, is
the thing of performance when deliberately pointed off the desired
signal to further suppress an inteferer that is not separated by a huge
angular amount. If you think about a "canonical" sort of pattern it's a
cos^n(theta*k) thing. The null-null beamwidth is pretty much determined
by the physical size of the antenna, but you can probably play with how
broad and fat the main lobe is, and how fast it rolls off. Think in
terms of the difference between a sine wave and a square wave of the
same frequency.
The ability to point a null, and still have decent response off the
null, might be a more valuable capability, at least for receive. Then,
there's the W8JK two elements fed out of phase sort of thing.. very
narrow beam with two elements, but you can take a big hit on efficiency
because of the circulating currents.
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|