http://www.lazygranch.com/images/radio/lv_3.gif

I've been experimenting with some homebrew direction finding gear,
namely loops and a Wellbrook amp. I have two loops, which for this
discussion will be called large and small. The large loop is about 2ft
on a side, and the small loop is about 1 ft on a side. The loops have a
platform to hold a compass, plus a handle to hold the loop. The handles
and compass platform are plastic. I use a magnifier over the compass to
minimize error. The compass is your basic Silva (about $10 at REI).

The beacon is 374Khz, located in Livermore, Ca. It backs right up to
hills on its east, which I'm sure adds nasty reflections.

The first measurement, point "1", is 2.4 miles from the beacon. I only
had the large loop at the time I did that measurement. The null was
very sharp.

I removed point "2" from the digram as the error was huge. I could not
get a decent null. At the time, I wasn't sure if the loop was the
problem, or the location. That was when I thought building a smaller
loop for comparison would be a good idea.

I used both the large and small loops at point "3". The null was good
with both loops, but clearly the smaller loop had less error.

At point 4, the large loop had a poor null, so I skipped the reading.
The small loop had 1 degree error.

I found that using the radio in CW mode with a narrow CW filter made
finding the null a bit easier.

Any ideas on how to improve the technique would be appreciated. The
goal of this exercise is to locate a beacon that is in a location where
deadly force is authorized, so I'll need accuracy from at least 10
miles away and probably not line of sight in all locations.