On Mon, 15 Oct 2007 08:47:16 -0700, Tim Shoppa
wrote:

I think I can put down at
least quarter-wavelength radials from each base in all directions and
half-wavelength or longer in some other directions.

Hi Tom,

Radials, on/in the ground, do not need to be fractional wavelength
multiples (eighth, quarter, half...). Length = radiator height is
fine, shorter and more is roughly equivalent.

Is there
something I can do to prove to myself that having my house on one side
of the array and the edge of my lot on the other side of the array,
resulting in a radial plane that's not radial but more like a
rectangle, won't mess up the whole concept?

No, probably not. And if it proved anything, what could you do about
it anyway?

And if I draw things out it looks like an 80M foursquare array would
be possible but my house would be in the middle of it :-). Is this a
recipe for disaster or would it actually work?

No, not a disaster. It probably might/might-not work.

Radials are for matching and efficiency. They do impact lobe shape,
but in the scheme of things (a four-square) you are already managing
that detail and the lobe offsets probably average out anyway.

What I mean is that if you located each element in the corner of your
lot (hypothetical, as set-backs for neighbors and guying will probably
force another configuration); then you would have a radial field at
each one that could only fill 90 degrees of the circle expected at
each (for a pristine, factory guaranteed installation). The lack of
3/4ths of those radials would bring loss. The lack of 3/4ths of those
radials would reduce signal strength in those un-radialed directions.

However, you have three others that do point in those directions
filling in all 4/4ths - as an average. You still suffer the ground
loss, true, but the ground coverage you do have probably stiffens the
system sufficiently to exhibit the expected beam forming - I presume
this is the feature you are after.

73's
Richard Clark, KB7QHC