Re: NEC-LIST: VHF CFA paper

Brian,

Thank you for your further comments on the CFA. Re delay in replying,
no problem. I was not in the office to read any e-mail messages
anyway, since I participated in the ACES Meeting in Monterey, 15-20
March, and just returned to my office yesterday.

You wrote:

> Dear John
>
> This time I've not been quick in my respone to your last e-mail! So
> please accept my apologies. I didn't have a copy of the VHF article
> by Colin Davis until this morning and now I've read it I feel I can
> respond to answer a few of the issues raised by yourself and
> others. Some of course I cannot answer yet and am always open to
> discussion and experimental experience. Anyway, here we go.
>
> Reading Colin Davis's article there is no way that his CFA will
> work. The reason is that his E X H is always inward. His 90 degrees
> phase shift is the wrong way round for the E and D plates.

The fact that his CFA radiates at all must mean (in my view) that he
had the phases right --- but I must say that until now (see below) I
did not believe the sign of the phase made any difference. Since the
phase reverses every half cycle, it seemed to me that the Poynting
vector associated with the E- and H-fields between the plates (or
beneath the plate) would point in during one half the cycle and out
during the other half cycle.

> There are a few interesting details in the article but I would not
> expect much radiation so -65dBm is probably an acceptable
> value. Another issue involves Colin's assessment of the dummy 50ohm
> impedances attached to E and D feeds. Why he chose 50 ohm real
> impedance in place of something reactive I've no idea. It should be
> reactive and real at the frequency of radiation, i.e. when it is
> radiating. The issue is where the real impedance should be as it is
> a "load" to space. In addition I don't actually feed the E and D
> plates with a 50 ohm transmission line from the phasing unit.

The source impedances for the CFA are a real problem --- see what
follows below. Hatley and co have never said what these impedance
are, and neither have you (see above).

> Now some other issues. In relation to general questions about
> connections to ground, the GP is the other side of the E plate and D
> plate capacitances. I don't use transformers anywhere near the E and
> D plates. I often use one for impedance matching at the input of the
> phasing unit from the transformer but this is away from the CFA
> itself. Certainly, over the past year or so since I've returned to
> CFA work, the phasing circuits have been improved since their early
> days. People keep questioning us about the E and D plate feed wires
> and the magnetic field they produce when feeding the CFA saying
> these are producing the radiation. The answer to that one is quite
> simple. We feed the E and D plates with wires coming in on the
> horizontal not the vertical. This way any H field vectors due to
> currents in the feeders always lie in the vertical plane - easily
> proved from Ampere's law etc. All the broadcast CFAs, and also my
> own CFAs, are fed this way. The H field from the radiating CFA is in
> the horizontal plane as the polarization is vertical - so the
> radiated H source cannot come from the current flowing in the
> wires. The only source of H field vector which is in the horizontal
> plane comes from the H field created from time varying voltage
> across the D plate.

The CFA developed by the Egyptians is fed by a short vertical
conductor coming up from below --- and so the conductor feeding the
cylinder must make a right-angle bend to connect to the bottom edge of
the open cylinder.

> In terms of models I suppose a good start would be the dimensions of
> the CFA published in AntenneX. If NEC can cope with that then it's a
> good starting point.

> I am going to check out the NEC code to see if the codes are
> actually capable of evaluating H field only from time varying
> voltages and not from current. I'll pass on my thoughts at sometime
> in the future - I hope!

I have yet to see a proper analysis of this antenna. To understand
the CFA we have to understand power flow as given by Poynting's vector
(and that is not my forte)

        P = E X H

The power flow/unit area at any point due to a current element is
given by the instantaneous values of E and H. But real power flow
(away from the antenna), and not just surging back and forth (average
power zero over an RF cycle) is found only in the radial Poynting
vector at a distance sufficiently great that the 1/r terms dominate,
since only the 1/r terms contribute to outward flow of power.

The total power radiated by current elements can be computed by
integrating the radial Poynting vector over a sphere (or hemisphere)
located at this far field distance (where the 1/r field dominates).
This has not been done.

In my view, NEC-4D should be able to model any antenna which can be
constructed as a wire grid --- and versions of NEC can model plates
and cylinders. NEC calculates near fields as well as far fields.

My further comments follow.

ON THE CROSS FIELD ANTENNA
______________________________

At the recent 15th Annual Review of Progress in Applied Computational
Electromagnetics Conference, Monterey, 15-20 March 1999 I spoke in
open forum (Session 7 on Wire Models) on the CFA. I decided to do
this, since Broadcast consultants were asking me about the CFA.

I have noted before that in the February 15, 1999 National Association
of Broadcasters weekly newsfax for Radio broadcast engineers, in the
Radio TeckCheck Column, it is announced (headline to column Please
Forward to the Engineering Department) that:

"AM broadcasters may be on the verge of an antenna revolution,
according to a paper to be presented at the 1999 NAB engineering
conference, with the advent of a new design known as the Crossed Field
Antenna. Currently in limited operation for the Egyptian Radio and
Television Union (ERTU), the CFA represents a fundamentally different
approach to medium wave (MF) antenna design."

It is further stated that one version of the antenna, operated at a
power level of 100 kW, frequency 603 kHz has a bandwidth of 48 kHz,
and while gain is not mentioned, the "flared version" is said to
produce a significant increase in ground wave radiation and an
accompanying decrease in sky-wave radiation.

You can see why broadcasters are excited about the possibilities for
this antenna.

The statement about the effect of the flare creates an additional
problem for anyone to believe. In accord with conventional thinking
there is no way such a small flare can change the vertical plane
pattern, since the contour surface of the flare is such a small
fraction of a wavelength.

The theme of my talk at the ACES Meeting was that there is a real need
to CEM model the antenna; and to measure the performance of the
antenna by someone other than its inventors.

So, on my return to my office I modelled the (original) monopole
version for this antenna --- the antenna described in the 1991 IEE
ICAP paper by Hately, Kabbary and Khattab. I should have done this
prior to my ACES talk.

The frequency is 1161 kHz.

The principle dimensions are:

        E plate Hollow Metal Cylinder 2 m diameter, 2.5 m high,
supported 0.6 m above:

        D plate Flat metal disc, 4 m diameter, supported 0.6 m above
ground plane (wire mesh mat) 10 m square and thoroughly bonded to
earth. The ground-plane grid was on the roof of a building.

I have assumed that the antenna is sitting on a PEC ground --- lifting
the antenna off ground, and connecting it to ground by a wire or wires
will chnage the picture, since current will flow on this wire, or
wires. And in writing this note I discover I made a mistake --- my
plate has a diameter of 5 m not 4 m.

The only program I have for creating wire grids to model solid
surfaces, models the surface by many-many square or rectangular grids,
all wires connected. I chose to model both surfaces, the cylinder and
the plate as a wire grid side length about 0.001 wavelengths (exact
size chosen so that the grid models the dimensions). My cylinder is a
polygon with 20-sides, which looks quite round; but my "circular" disc
is a bit ragged.

The impedance of the cylinder monopole alone Za = 0.11 - j 706 ohms.
This seems about right.

Clearly (according to NEC 4D) there is a complicated interaction if I
feed both the cylinder and the metal plate; and, I have a problem. I
cannot feed these elements by equal powers, since the impedances are
so different, and the impedance of the cylinder has a negative
resistance.

In what follows I have used equal current sources, phases + 45 degrees
(E-Plate); and - 45 degrees (D-Plate).

        Z (cylinder) = - 158 - j 702 ohms

        Z(plate) = + 170 - j 327 ohms

        Gain = - 15 dBi.

If I tune out the inductive reactance of these elements (inductor
Q-factors 300) this reduces the gain a bit, to - 16 dBi.

If I reverse the phases the gain is: - 99.99 dBi --- and so finally we
have found point of agreement. AND, curiously the resistive component
of the impedances change --- it is now the plate that has the negative
resistance. VERY INTERESTING.

This is not a very efficient antenna, considering I have a PEC ground
beneath it!!! But it is a complex antenna.

If I use equal voltage sources I compute quite different impedances
(but the resistive component of the cylinder is still negative) and a
slightly different gain --- even a bit worse --- and so clearly the
source impedances depend on the method of feed.

John S. Belrose
26 March 1999

_____________________________________________
John S. (Jack) Belrose, PhD Cantab, VE2CV
Senior Radioscientist
Radio Sciences Branch
Communications Research Centre
PO Box 11490 Stn. H
OTTAWA ON K2H 8S2
CANADA
TEL 613-998-2779
FAX 613-998-4077
e-mail <john.belrose_at_crc.ca>
_____________________________________________
Received on Sun Mar 28 1999 - 15:13:00 EST