Radiation from grids or plates
Updated 20170912

Leading on from the 2017-1 page, I’m researching a small antenna with performance comparable to a full size one. This page begins with a measurement to back that up. To emphasise - I’m NOT working on EH antennas, having concluded their performance is 7-10dB worse than a conventional antenna.

Conventional small loaded antenna comparison
For a frequency near 28MHz, a
Boomerang 27W was mounted in the same position, just slightly higher to accommodate it’s long bottom dipole element. The boomerang is 1.75m x2 = 3.5m long, which is 0.67 of a half wave dipole. It was compared with horizontal, tilted and vertical polarisation to a rebuilt E-field antenna of 0.7m height. The MMI-dipole used for horizontal and vertical, the inverted-V for tilted polarisation.

Boomerang signal readings off KiwiSDR:
Horizontal = -86dBm
Tilted = -74dBm
Vertical = -75dBm

E-field signal readings off KiwiSDR:
Horizontal = -96dBm
Tilted = -75dBm
Vertical = -72dBm

EH antenna signal readings off KiwiSDR:
Horizontal = -96dBm
Tilted = -80dBm
Vertical = -80dBm

It’s difficult to really “average” these results, but it’s clear the E-field is very close to the Boomerang. Taking half the Boomerang length as a fair comparison, it’s 1.75/0.7 = 2.5x longer than the E-field antenna. As said on the previous page, you should assume some uncertainty in these measurements.
Boomerang photo.
27W

Bandwidth of Boomerang 27W, 2:1 = 1.2MHz.
Boomerang back at Merrivale

Bandwidth of E-field antenna, 2:1 = 600kHz, albeit centre frequency 500kHz higher.
0.7m high VSWR

Exactly half the bandwidth. A photo of the antenna that gave the above VSWR plot.
Mini mesh

Chu-Wheeler Limit Comparison
Any suggestion such a short radiator can match a far longer one is controversial. Does it contravene the Chu-Wheeler criterion? It measures as linearly polarised, so the McLean mixed polarisation limit doesn't apply. Using algebra terms defined by Chu-Wheeler...

Radiansphere 0.75 high, 0.4 wide = 0.84m (by Pythagoras)
Frequency = 28MHz
ka = 2xPI/λ x 0.84 = 0.50
Efficiency (eta η) estimated by measurement at 0.5 (-3.0dB down on half-wave vertical)

There are several documents transposing the basic Chu formula to give VSWR bandwidth, here is a link to one. Prof. S.R. Best gives equation (6) with a handy calculation including SWR (s). Working out on paper gives 0.28MHz. Measured 2:1 SWR bandwidth = 0.6MHz, so about twice the conventional theory. The document also gives bandwidth comparisons for many proven small antenna designs.

Of course this doesn't include the size of the ground-grid (or ground plane). The grid may not be acting purely as a reflector. If anyone likes to correct my arithmetic above, go ahead. With all uncertainties involved, I cannot confirm the antenna breaks conventional theory. But, it’s definitely a good antenna for its size when comparing to the accepted limitation.

H-Field Probe Test
H-field probe test… where is the H component generated? To get some idea, I made a simple H-field probe. Consisting of a shielded loop, diode detector, and moving coil meter. The coax braid stops the electric field, but the magnetic field penetrates the braid. Shielding is important to minimise pickup from the strong high voltage electric field present in antenna testing.
H-field probe
Showing the results on paper is difficult. I made a YouTube video. The largest magnetic field seems to be the side of the grid over the edge of the ground-grid. The side near the middle of the ground-grid has hardly any magnetic reading. This is significant, somehow?

This experiment disproves what some say about Maxwell’s displacement current. They say there’s a magnetic field between two plates of a capacitor, because there’s an electric field. But, a magnetic field can only be made by current flow in wires. As you can see I measure no magnetic field when the meter is moved away from the plates.

The antenna obviously generates a high electric field (E). But, a corresponding magnetic field (H) is needed for generation of far field radiation, with the E x H vector product. An H-field is generated across areas of the metal high voltage grid, and from the transformer coil. Improving the H-field characteristics is probably the key to improving the signal further.


Looking for improvements - Further tests
Some experiments, with a single paragraph for each with its result.

1. Mesh gives almost the same top-grid/ground grid capacitance as a solid plate,
as confirmed by NASA. 0.5” square stainless steel welded mesh of approx. 10% larger than a perforated sheet gives the same resonant frequency, and far field strength. The advantage is greatly reduced weight and windage. TEST COMPLETE.

2. The shape of the top-grid was changed from a flat rectangle by adding mesh pieces of half the size of the main piece. By this method, triangular and X-shaped top grids were tested. The triangular grid produced a 10% reduction in frequency, the X-shape only 5%. This counter-intuitive result needs further investigation. Accurate far-field measurements were not taken, however it was noted the background sky noise was similar. TEST INCOMPLETE.

3. Balancing by moving the tapping point, to see if its position can minimise common mode. Measurement of common mode was by the simple current meter shown elsewhere on this site. Moving the tapping point 1 or 2 turns up the coil didn’t make a measurable difference to the common mode current. Moving it further made things worse. To help improve magnetic linkage, a tap 1 or 2 turns up the coil is thus preferred. TEST COMPLETE.

4. Temperature tests. Running 12W through for 5 minutes produced a temperature rise above ambient of: 10 degrees for the (small) choke balun; 5 degrees for the coil. This is some evidence the antenna can support 100W SSB, but a larger piece of ferrite is needed for higher power. The power dissipation of the choke is estimated as about 0.5W, or 5% of the total power. So, blocking the common mode current like this doesn’t impact significantly on overall efficiency. TEST COMPLETE.

5. Previous results showed the ground grid must be larger than the top grid. How much larger to be optimal is unknown. The resonant frequency rises by about 2% by folding over the ends by a quarter. This indicates the ground-grid is not the main source of radiation, or the shift would be bigger. However, no far field tests were taken from this, and they must be for a definitive result. TEST INCOMPLETE.

6. Small “solenoid” coil vs. a larger 1.5 turn loop with pickup coil. It’s easy to use a primary coil and get a good VSWR. Unlike a small magnetic loop, it’s not directional. In terms of wire length, about 30% more is needed with a large coil/loop to get the same frequency, but the wire can be much thicker.

Interestingly the primary loop is half the diameter of the main 1.5 turn coil. This suggests it may be possible to directly couple across 1-turn with a 2.5 turn coil. Far-field tests on this gave 2dB down on the solenoid coil with autotransformer connection. The antenna is shown in the following picture, lifted off the ground grid to more clearly show the size. TEST INCOMPLETE.

Using a transparent support rod, the antenna is almost invisible at a distance!
Loopy E-field

Height of the plate above the ground-grid for maximum far-field radiation? All previous tests indicate the spacing should be 4-6% λ otherwise radiation is reduced by 6dB or more.

Position of the coil relative to the grid? Rotating the top grid makes no measurable difference to the far field signal. Moving the top grid physically off to the side may make a difference.

A link coupled tap reduces common mode current, but does efficiency suffer? Previously I found efficiency was reduced by 1dB which is not much, but worth doing the test again.

Effect of concentrating flux by putting an iron-dust toroid in the coil. Does this restrict the H-field and reduce radiation?

Quarter wave fibreglass pole comparison. Put a conventional quarter wave vertical on the ground-grid and measure.


Theoretical Speculation
Electromagnetic theory applied to antennas is often counter-intuitive. So what I’m about to say maybe completely wrong!

In conventional antennas electromagnetic radiation is generated by charge acceleration in long wires. The balance of E and H field magnitudes suggests conventional antennas (dipole) have medium near field wave impedance. Small loop antennas concentrate the current (H), and have a low near-field wave impedance. They have high voltages across the area at the tuning capacitor. In their case the E-field is constrained between the plates.

E-field antennas increase the electric (E) component and have high near-field wave impedance. But to radiate, there must still be a displacement current. Exactly how the varying magnetic field is formed by predominantly the electric field is unclear. Enhancing the magnetic component which seems to be made without current flow in a wire might be the key to improving the antenna.

The antenna is 0.06λ high, and appears to defeat the Chu-Wheeler limit. But the different coils used and their alignment didn’t produce much improvement. This again indicates there is no Poynting vector synthesis happening.

It’s fun to poke around in the dark when improvements are made. But demoralising when things go wrong and the performance of the antenna gets worse. Some more theory must be forthcoming soon…