A number of propagation programs can tell you what angle you need. One
that's easy to use, reasonably accurate, and free is W6ELProp, available
from http://www.qsl.net/w6elprop/.

The ideal height for the antenna is a fairly easy geometry problem, but
it's easiest to let a program tell you this, too. There are a number of
antenna modeling programs available, some of which are free, and all
will give very accurate results. Because it's my product, I recommend
the free demo version of EZNEC from http://eznec.com, which is entirely
adequate for this purpose. But any number of other programs would be
suitable also.

One bit of advice -- don't get hung up on the "takeoff angle" reported
by some antenna analysis programs (like EZNEC). It simply means the
angle at which the pattern is maximum, and isn't important for this or
most situations. What counts is the field strength or gain at the angle
which you'll be using for communication -- that's the angle reported by
the propagation program. That's the number you want to maximize.

Roy Lewallen, W7EL

Al Lorona wrote:
Hi, Everybody,

How does one determine the correct antenna takeoff angle to communicate over
a certain distance?

I can't find any references anywhere that discuss this, taking into account
all of the variables: height of ionosphere, frequency, etc.

Here's a practical example of when it would be important to know this. Say
you are going to put up a dipole antenna for 20 meters specifically to
communicate to a friend 2000 miles away during the day. The takeoff angle of
the major lobe varies widely as the antenna is raised from ground level to 1
wavelength above ground, so presumably there is a 'best' height for the
antenna to put the major lobe right where it needs to be to skip to your
friend's QTH.

The trouble is, we do not seem to have a good way to determine this 'best'
height ahead of time. One would, I guess, have to vary the height of the
antenna while the other station monitored your signal strength, and one
would have to do this over several days, weeks, or months to get good,
averaged data. But... isn't there a better way to do this?

Regards,

Al W6LX