Active Wideband Directional Antenna with Vertically Polarized Small Loop and Small Dipole

Published on: 2020/01/20, Rev. 1.0 Jun 2019, Rev. 1.1 Jan 2020

Author: Chavdar Levkov LZ1AQ

A combination of a small magnetic loop and small electrical dipole or ground plane (GP), both vertically placed in a single point is widely used as a simple direction finding antenna. This method is very old and well known and goes back to the early days of radio. In amateur fox hunting it is a preferred directional antenna for lower frequencies (whip + ferrite rod). This combination is still used in some less expensive modern direction finding equipment, airplane devices and also in some sea surface radar systems [1] Fig.1a . Further we will use terms dipole for short dipole or short ground plain (GP) and loop for small loop. The combined array will be denoted as SALAD (Small Active Loop Active Dipole).

Fig.1a Sea surface radar receiving antenna system, bandwidth from 10 to 100 m wavelength.

1. How it works?

Let us have two transmitters radiating vertically polarized waves with equal phase placed at the left and right side of the loop along the X axis as shown on Fig.1. The currents induced in the loop from both transmitters are with opposite phases (180 deg.) This is a direct consequence from the Faraday law – the magnetic field vectors are opposite. From the other hand the currents induced into the dipole are in phase with the wave field from both transmitters. If we add the signals from the loop and dipole we will obtain a deep null in one direction and double amplitude in the other. We assume that the voltages we pick up from the dipole and the loop are with equal amplitudes and with 0 or 180 deg. phase difference. So if we have a wideband active magnetic loop and wideband active electric dipole, both with constant versus frequency gain and similar phase response, equalizing gain and phase and adding both signals will give a wideband uni-directional antenna.

Fig.1 The 3D radiation pattern of SALAD. A small square magnetic loop and a short electric vertical are modeled in MMANA.

At first glance there is one problem. According to Faraday law the electromotive force created in the loop by the wave field is 90 deg. shifted versus the phase of the field itself. But if we make the loop to work in a short circuit mode – e.g. load it with amplifier with very low input resistance, the current in the loop will be in phase with the wave field. The reason is that we have another 90 deg. shift from the serial inductance of the loop so the overall shift of the loop current becomes 0 or 180 deg. The essential requirement is that the input impedance of the amplifier must be much lower than the inductive impedance of the loop. This requirement makes the loop also wideband [2].

2. Models

Vertically polarized SALAD

The radiation diagram of SALAD is a cardioid. I have built several MMANA models (as in Fig.1) with two sources in transmitting mode. All these models can be downloaded here.

(SALAD-S.maa) This model is for single frequency. There the gain of the loop and dipole are trimmed to obtain equal field strength. Additional 90 deg. shift is set to one of the sources. In this model we have voltage sources and the current in the small loop is 90 deg. shifted (the loop is almost pure inductance) so we need another 90 deg. shift to obtain 0 or 180 deg. current phase - to be the same as the current in the dipole.

(SALAD-W.maa) This is wideband model. Here we feed the loop with a current source which is in phase with voltage source in the dipole antenna. In MMANA there is no current source available. In the feeding point of the loop, a 10 Kohms resistor is added, the voltage is set to 100 V and now the source current is 0.1 A and is independent from the frequency. Again the voltage amplitudes can be trimmed to obtain the best F/B ratio. In this model there is no need to introduce 90 deg. phase shift since now the current in loop is in phase to the voltage in the dipole. This model is wideband and we have almost perfect cardioid patterns from KHz to 10 MHz without any change in model parameters. Above 10 MHz the pattern begins to degrade since the loop and the dipole are not “small” anymore. This model is the transmitting equivalent of the receiving wideband SALAD where the loop is loaded with amplifier with very low input impedance and dipole with high input Z amplifier.

(SALAD-CC-W.maa) This is also wideband model but there the loop element consists of two crossed coplanar loops. The CC loops have much higher frequency limit up to where they exhibit small loop behavior [2]. This model is exceptionally wideband – the unidirectional pattern is preserved from KHz up to 30 MHz. Modeling shows that the dipole element length must be shorter than 0.2 wavelengths. ( for dipole it is 2 x 0.2 wl and for GP is 0.2 wl ). Above this length the dipole is no longer small.

Fig.2 Radiation pattern of SALAD with height 2 m over a medium conductivity soil.

Fig.3 Radiation pattern for V and H (red) components. It can be seen that there are still two sidelobes with horizontal polarization but their relative weight is significantly reduced compared to the single loop case

Models show that the coupling between loop and dipole can be ignored since both antennas are aperiodic and small. The null depth does not depend on the mutual space positions between the loop and dipole if they are placed in homogenous field. A model where the dipole is moved with 1 m off the loop center asymmetrically does not change the radiation pattern.

Horizontally polarized SALAD

The SALAD can be used in horizontal position . The usual requirements for sufficient height are valid as for any other horizontally polarized antenna. Fig.4 shows SALAD radiation pattern for 0.25 wl and 0.5 wl height. The SALAD is small and compact and can be placed on the top of a tower without any boom . Its performance is expected to be slightly better than a single flag antenna with much higher antenna factor than the flag itself.

Fig.4 Horizontal SALAD at 0.25 wl and 0.5 wl height.

Two elemen SALAD

If we consider SALAD as a single antenna element, two or more SALADs can be used in phased arrays as described in [3].

Fig.5 Two phased end fire elements SALAD.

As an example, a wideband MMANA model is given in SALAD-2el.maa This is a subtractive (end fire) array (Fig.5). The distance between SALAD elements is 12 m . Each element is set as a separate SALAD and then they are phased with delay line as described in [3]. Distance between 10 and 15 m is a good compromise for 1.8 to 10 MHz bandwidth.

Comparison with other receiving antennas

I have used spreadsheet created by AC6LA [4] to calculate the receiving antenna parameters of SALAD and compare them with some others RX antennas using MMANA models for 80m band. The reader can read the presentation of Juka, OH6LI [5] and also [17], W8WWV, [18] K7TJR, to get information of the meanings of the calculated parameters and detailed discussion about receiving antennas. The gain is not given since it is meaningless for active antennas case. The F/B ratio is also omitted since it is a single point metrics and in reality the model values cannot be reached.

Table 1 Comparison of different receiving antennas. RDF is Receiving Directivity Factor ,W8JI[16], DMF is Directivity Merit Figure, ON4UN [8]. Leaking index, OH6LI [5].

As can be seen SALAD is slightly better than K9AY and much more wideband. K9AY is very similar to SALAD since it can be assumed also as a combination of a loop and vertical electric element [6,7] where the summation takes place at the loading resistor. It must be pointed out that two element phased SALAD has low leaking index [5] and will be very efficient in reducing QRM and noise outside the desired direction ( at least in modeling, not tested practically).

3. The setup

The setup is simple. The signals from both amplifiers are added in combiner. The transformer T1 is needed to change the phase with180 deg. in the loop chain in order to switch the direction to opposite. The attenuator is placed in dipole chain since the antenna used has higher antenna factor than the loop.

Fig.6 The signals from control boards of both amplifiers are added in combiner. The transformer T1 is needed to change the phase with180 deg. in the loop chain in order to switch the direction to opposite. T3 is 25:50 ohms transformer.

Two active antennas type AAA-1 are used - one in the loop mode, the other in the dipole mode [9]. The loop amplifier has very low input impedance (6 ohms) and the dipole amplifier is with high Z input. The loop and dipole amplifiers of this equipment fulfill the requirements for flat gain and phase response. The phase delay between the loop and dipole amplifiers was measured with a single source and it was estimated that the signal delay difference between both amplifiers is small. For frequencies between 1 and 15 MHz the phase difference is within several degrees. The reason is that both amplifiers in AAA-1 have common output stage with the same transformers and we can expect to have similar phase response. The spice models also confirm this proposition (Fig.7) . The frequency responses of the antenna factors in both amplifiers are flat and there is only fixed gain difference between them [9].

Fig.7 Phase response of the loop amplifier (black) and the dipole amplifier (red) curves (AAA-1C type) from a spice model. Equivalent antenna elements are modeled so this is a response to electromagnetic field source. The dipole source impedance is with capacitance of 20 pF and loop source has inductance of 2.5 uH. At 1.8 MHz the phase difference is 4 deg. Above 3 MHz the difference is less than 2 deg. Ignore the curves at the right side of the graphics which are part of the amplitude response. The increased phase difference at the lower frequency edge is due to the low loop reactance which becomes comparable to the loop amplifier input impedance. If the loop inductance is increased (larger loop) the frequencies below 1 MHz can be used.

A ground plane with 5 m length was used for electrical antenna. A small metal rod with 80 cm length was inserted into the soil for ground connection. No radials are used. The dipole amplifier has high-Z input and ground losses can be ignored. The loop antenna has 2 crossed coplanar loops [2] in the same plane each with 96 cm diameter made from PE/Alum heating tube with 16 mm diam. The GP is made from a wire fixed on a fishing rod and is placed at the center of the loops 10 to 20 cm away from the tubes. The GP antenna has higher antenna factor than the loop and a step attenuator (11 dB with 1 dB step) was placed in the dipole chain. Both active antennas are connected to control boards with equal length FTP cables. Identical choke baluns were inserted in both FTP cables [10,11] in order to avoid any common mode influence on the antenna performance. The shields of the FTP cables were grounded between the common mode baluns and amplifier.

A phase shifter ( variable delay line) was inserted in preliminary experiments in the signal path either in loop or in dipole chain in order to compensate possible phase difference. The delay line has a negligible attenuation and does not influence the amplitude balance. The idea was to compensate the phase difference at certain frequencies (e.g. in ham bands). But practical measurements show that there is no need for compensation at all since the real performance does not need such a precise tune. The phase dispersion of the real signals is so high, that small phase corrections are useless.

The summation of both signals is performed in wideband magic-T combiner. The loop output polarity is changed by swapping the secondary terminals of transformer T1 in order to switch the direction to opposite.

How to tine the array

I have used GP antenna with slightly larger gain than the loop. The tuning procedure is to set the attenuation in the dipole chain to proper value so that the gains of loop and dipole are equalized. We must bear in mind that a deep null can be reached only in controlled environment with a small test transmitter placed several wavelengths away with vertical polarized antenna. Set the attenuator at position which gives maximal F/B ratio.

Another method is to use a nearby MW BC station for test signal. During the day time there is only ground wave and there is no fading. By switching between loop and dipole we can find attenuation level where both signals are equal. The requirement is that we must know where the transmitter is located, then to turn the loop to this direction and just then to compare the amplitudes from the dipole and the loop.

Other method is to periodically switch loop and dipole to the receiver input as described in [12] with real skywave signals on short wave frequencies. With trial and errors an optimal attenuation can be set where the gain of the loop and dipole are almost equal and F/B ratio is maximal. Again we must know the direction of the transmitter and turn the loop. There are world lists of transmitter locations and their schedules so these BC stations can be used for equalizing the gain in both channels. There will be some dependence from the working frequency but a compromise can be found which serves all bands of interest. Fading makes this measurement a bit uncertain . Plus minus 1 to 2 dB are not so important and within this range the tuning might be considered complete. There is no need to use phase equalization – a small phase mismatch is acceptable and hardly will be noticed with real signals. When the setup is tuned the variable attenuator can be changed with a fixed one.

I have used also another setup. The signals from both antennas are fed to the inputs of a two channel coherent receiver. I have used a simple dual channel direct conversion receiver for this purpose with common oscillator [21]. Then the AF outputs were fed to a two channel oscilloscope. In XY mode the Lissajous curves are quite informative. There the amplitude and phase of both signals can be observed. For me this is the best method.

Sometimes I have observed a small phase difference between loop and dipole voltages in controlled environment with a small transmitter placed several wavelengths away. This effect is usual when the experiment is made in environment where there are near objects and a reflections takes place. Both antennas are at one point but they have different patterns which leads to different sensitivity to reflected waves thus giving a phase difference in the output voltages. This can be avoided if the measurement is performed in an open field without any obstacles and reflecting objects. Any ground discontinuities or vertical conductive structures (poles, buildings, power lines), especially those in resonance can cause significant pattern distortions.

Only the F/B ratio was measured at frequencies from 1 to 30 MHz. The method is described in [12] and is quite accurate. A rotary loop system was used and the loop was turned until best F/B ratio was reached. The null depth usually was between 10 and 20 dB. The system definitely has good directivity from 1.5 to 15 MHz with the described setup.

It must be pointed out that the presumption that there must be fixed phase and amplitude difference between loop and dipole elements is simply not true for skywave signals. Propagation changes combined with reflections from local objects makes those signals quite different at time. Substantial and floating differences in amplitude and phase shift were observed. I have used a coherent 2-channel receiver and oscilloscope to observe the signals from both antennas. A phase difference up to 90 degrees were sometimes observed and amplitude differences up to 10 dB. This is one of the reasons why this phased array does not work sometimes. Measuring the instant F/B ratio of SALAD for sky wave signals might give figures between 0 and 30 dB in the period of several minutes . Even if the presumption of homogeneous filed around both antennas is true the difference in radiation patterns and the horizontal polarization sensitivity of the loop element results in differences of the induced voltages during propagation changes.

As a matter of fact any classic array has similar problems. The essential requirement for the receiving phased array to work as the modeling predicts is : a/ elements with the same radiation pattern and b/ homogeneous surrounding field to all elements. I have made a lot of tests with the same two channel coherent receiver comparing sky wave signals from two identical elements (loops or dipoles) placed at a distance from each other.( from 0.05 to 0.5 wavelengths). The presumption that there must be only phase difference (equal to time delay of the wave propagation between elements) is simply not true most of the time. The amplitudes and phases float with time in the case of real signals. We are not receiving the incoming wave near the ground – the field around the antenna is a result of the interference from the incoming ionospheric wave and all the reflections from ground and other objects. Even at very close points the filed pattern might be different. When the incoming wave changes its angle of arrival and polarizations, the interference field picture is also changed and the voltages obtained from antennas at these two points are not changed in a similar way leading to changes in F/B ratio during the time.

The author has used SALAD for several months in semi-urban environment to listen from 1.8 to 14 MHz. It was made with two crossed coplanar loops in one plane each 1.3 m diam. made from 16 mm PE/Alu heating tubes and wire GP 5 m long on fishing rod, two AAA-1C amplifiers. Up to 10 MHz this SALAD is not limited by internal noise and can be assumed as a serious extremely compact uni-directional DX antenna. Above 10 MHz it is still directional. At 14 MHz it is comparable or better than my inverted V at 10 m height. Tubes between 10 to 16 mm diameter can be used.

Trapezoidal shaped crossed coplanar wire loops were tested also (Fig.8). This wire antenna was made with four directions by switching two orthogonal sets of crossed coplanar loops. A single wire GP , 4 m length was used. The performance was similar to the tube loop variant but its low frequency limit was extended down to 0.8 MHz. Using shorter GP (4m) extends the HF bandwidth up to 21 MHz.

Fig.8 Wire version of SALAD . Two crossed coplanar loops have 4.8 uH resultant inductance. Low frequency limit is 0.8 MHz. It has high F/B ratio up to 18 MHz and is still useful at 21 MHz (models) . The conductor is 1.5 mm2 (d=1.4mm) with PVC insulation both for loop and GP. 4 meter fishing rod is used for mast. This antenna was used as a 4 –directional SALAD where a second identical loop was used rotated at 90 deg. The four ropes fixing the vertical mast are used as wire guides. With AAA-1 amplifiers the GP has higher gain and attenuator settings must be somewhere between 3 and 7 dB. Distances between loop and GP conductors are 3 – 4 cm. A distance scale is drawn at the right side of the picture.

4. Practical consideration


The gain of the loop antenna is very stable and is weakly influenced by the weather and humidity. We can use crossed coplanar loops in order to reduce the inductance and increase the sensitivity and bandwidth [2]. But there is some controversy – the loop reactance must be higher at least 4 times than the amplifier input impedance. So the low frequency margin of the array will be limited by the value of the loop inductance.


Changes in humidity , ground moisture content etc. might change the dipole gain noticeably. In order to stabilize the performance, ground plane (GP) antenna is better to be used. For GP type antenna a ground rod 0.5 to 1m long is sufficient since we have high-Z amplifier. For places where grounding is not possible a vertical dipole or GP with short elevated radials are the other choice.

The high frequncy limit

The high frequency limit is determined both by the small loop and dipole. The loop is no more “small” above frequencies where the loop perimeter is longer than 0.1 wavelength. Transmission line effects are taking place and the loop impedance is no pure reactive and high. Crossed coplanar connection extends this limit to much higher frequencies . Modeling shows, that for the dipole, the element length must be shorter than 0.2 wavelengths. ( for dipole it is 2 x 0.2 wl and for GP is 0.2 wl ). Above this length the dipole is no longer ” small ”. There is some difference in results between MMANA and NEC2. MMANA predicts patterns with good F/B ratio to higher frequencies. For example, MMANA gives 22 MHz and NEC2 - 15 MHz for the SALAD shown on(Fig.8).

The low frequency limit

The low frequency limit of SALAD is determined by Xloop/Rin ratio. If we set a boundary of 15 deg error (somewhat arbitrary) then the lowest frequency will be when Xl/Rin = 4 (Fig. 9). For 2.5 uH loop inductance and 6 ohms input impedance this limit is F=1.5 MHz. It is evident that the gain difference is important and phase difference up to 15 deg can be tolerated.

Fig.9 Phase shift as a function of X loop /Rin ratio in the loop amplifier. X/R = 4 gives a 15 deg error (from desired 90 deg.).

The dipole amplifier also has a low frequency limit since the input impedance is high but at low frequency the capacitive impedance of the whip antenna becomes very high and a phase shift between the field and output voltage from amplifier increases. If the whip capacitance is 20 pF and the Rinput of dipole amplifier is e.g. 100 Kohms then we will have 15 deg. phase shift at 300 KHz. These limitations for loop and dipole shift the phase in opposite directions and the SALAD uni-directional diagram degrades quickly.


Of course the user can use other amplifiers than AAA-1 but measurement of their gain and phase response must be performed. Probably the popular Miniwhip [13] antenna can be used for electric antenna but some experimental work must be performed to test its antenna factor and phase response. The antenna factor of the Miniwhip depends much on the feeder length and height [14,15]. The good news are that the wideband amplifier designs have intrinsically a smooth phase response and if we use similar and equal number of stages in both amplifiers we can achieve low phase difference. We must be careful with lossless feedback amplifier designs which might exhibit frequency dependant gain and phase shift due to load impedance variations. How much gain and phase difference is acceptable? It can be seen on Fig.10 and Fig.11 that a phase difference up to 15 deg. will be not of much influence to the SALAD performance.

Fig.10 F/B ratio as a function of phase difference between signals from loop and dipole. The amplitudes of both signals are assumed equal.

Fig.11 F/B ratio as a function of gain difference between signals from loop and dipole. The phase difference between both signals is assumed to be zero.

The amplitude variations have stronger impact but bearing in mind that in the experiments with real signals the average F/B ratio with precisely equalized antenna gains is somewhere at 15 dB it becomes obvious that a high precision is not needed. The variation of the real induced signals both in amplitude and phase limits the requirements for precise gain and phase response.


The gain equalization can be performed in the shack as shown on Fig.6 but that means we must have two cables (feeders). The other possibility is to equalize the gain and add the signals at the antenna side and to use a single cable to the shack. If AAA-1 amplifiers are used a free pair in the FTP cable can be used for RF signal from the other amplifier so a single FTP cable can be used.

Dipole antenna noise problem

One of the problems with SALAD is the local noise level of the dipole (GP) element. Usually the electric field antenna is more sensitive to the local noise probably due to strong electric field component from the near field noise sources. Also the conducted noise along the feeder is much more difficult to be attenuated with baluns since the dipole amplifier is a high –Z input and its common mode impedance is also with high Z [10]. Before making any plans to built SALAD in a particular location, a measurement of the dipole noise level compared to the loop noise level must be performed (both must be with equal antenna factors). If the local dipole noise level is much higher than that of the loop ( more than 5 – 6 dB ) the benefit of the SALAD compared to a single loop is questionable. Merely the local noise from the dipole will mask all the benefits of the SALAD system.

To reduce the conductive noise I will suggest to use very good common mode baluns both for loop and dipole amplifiers [10] . On Fig.12 is shown Z plot of a very good balun which I used in my setup. The core used is Feroxcube 16 x 9.6 x 6.3 mm, u=2000 , 3F3 material, 14 turns of FTP cable pair. Low frequency inductance is 700 uH. Each pair of the FTP cable is wound on separate core. [11]. The articles of K9YC in [18,19] are excellent guide how to build a good common mode balun. It is advisable to ground the feeder shield at a point between the antenna and the balun. Another grounding near the shack will do no harm. Important note - the user must used identical baluns in loop and dipole chain as well as equal length cables to the combiner otherwise a frequency dependant phase shift between loop and dipole will be introduced.

Fig.12 Impedance as a function of frequency for RF balun . This balun is very good for low bands and has acceptable impedance at 30 MHz.

4 – directional SALAD

Two orthogonal (rotated to 90 deg.) loops and a single dipole(GP) will enable to cover the circle with 4 switched directions. Changing the loop signal polarity will switch to opposite direction and changing the loops will rotate the direction to 90 deg. Switching schematics is quite simple. Only two amplifiers for dipole and loop are needed.

5. Conclusions



6. Links








[8] Low-Band DXing. ON4UN. ISBN-13: 978-0872598560