It consists of a crystal oscillator (Q1) and a class A buffer stage
(Q2), which feeds a bridge formed by C7, C8, R8, C10 and an unknown impedance
ZX. A meter with a rectifier is connected in other diagonal
of the bridge. L1 and L4 are short wave band local oscillator coils from
a transistor radio. They have 25 turns of 0.25 mm enamelled wire. Their
diameter is 6 mm (1/4 inch), length is 10 mm (0.4 inch). L2 has 5 turns
over the L1, and L3 is 3 turns over the L4. These coils have adjustable
cores made of ferrite (mu=100). R8 is a cermet trimmer from Bourns, and
C10 is an air dielectric trimmer. Since balancing the bridge actually changes
a capacitance connected across the L4, it is necessary to adjust this coil
to keep the power supplied into the bridge (and hence its sensitivity)
at maximum.
Due to some stray capacitance a real bridge is balanced when the unknown
impedance is not exactly equal to its equivalent formed by the R8 and C10.
This can be compensated by adjusting a C7/C8 ratio, but actually a substitution
method was used. When a balance is achieved, measure the R8 value, mark
the position of the C10, then use a bank of known value components instead
of the unknown impedance to get a balance at the same R8 and C10 settings.
This technique provides an accuracy figure like +/- 4 Ohm for a real part
of the unknown impedance from 15 to 30 Ohm and +/- 6 Ohm for higher values.
A problem with a small real part value of the impedance is a very small
deflection of a pointer. More power supplied to the bridge may help this.
The EH dipole used in this test is shown below.
It is made of a 2 inches plastic pipe from a HP plotter 24" roll paper. Both radiators are approximately 6 inches long and spaced by 2 inches gap. The dipole was installed on a tripod during measurements, the bridge was adjusted using a dielectric screwdriver made of a discarded toothbrush. Below is a close view at the bridge installed on the dipole.
Eight various configurations were tested:
1. Dipole alone. Nothing else was connected to the dipole during this
test.
2. MFJ in my left hand. This test emulated an in-shack adjustment procedure
recommended by Ted: hold you MFJ (or other signal source in a metal case)
in your left hand during adjustment. This setup was emulated by a 0.82
m (2.7 ft) test lead connected by an alligator clip to a metal bangle of
my wristwatch. Other end of the lead was connected to the lower radiator
of the dipole, like a coax shield for an L+T network.
3. Tripod. The same lead was connected to the aluminium tripod. The
midpoint of the dipole was 1.5 m (5 ft) above the floor.
4. Mains neutral. The bottom radiator was connected by two test leads
in series (total 1.6 m or 5.25 ft) to the mains extension cord. The cord
then ran 4 m (13 ft) on the floor (steel enforced concrete) before going
inside a wall (the same steel enforced concrete).
5. Central heating radiator. The same 1.6 m (5.25 ft) link, but now
from the bottom radiator of the dipole to the central heating radiator
on a wall nearby.
6. 1/4 wavelength ++. A wire of a total length slightly more than 0.25
wavelength was connected to the bottom radiator, then put horizontally
at a height varying from 0.9 m (3 ft) to 1.8 m (6 ft), hooked to a wooden
furniture at two points. The length of the wire was 6.3 m (20.7 ft).
7. 1/4 wavelength. Same as above, but the wire shortened to 4.4 m (14.4
ft), which is supposed to be a quarter of wavelength at 16 MHz used in
this test.
8. 1/4 wavelength --. The wire shortened even more. Now 3.5 m (11.5
ft).
The results are summarized in the table below.
As expected, an asymmetric dipole with one its half close to the quarter wavelength has a high radiation resistance. On other hand, mains connection provides a higher value of the real part of input impedance, but we may suspect mostly looses in this figure. The same is true for the "MFJ in my left hand" test: a short cable to the antenna analyser plus losses in one's hand can make an illusion of a high radiation resistance.