What got me started down the path of two element horizontal arrays was finding a design online for a two element horizontal phased dipole array for 40m using the well known Christman feed system with 84 and 71 degree coax delay lines, the info suggests it works well.
I was curious to see how this system looks when modeled and created one based on the given dimensions - two 63ft dipoles spaced 1/4 wavelength apart at a height of 45ft.
VA7ST has a handy Christman phasing calculator which calculates the 84 and 71 degree line lengths with the velocity factor of the coax being used.
With EZNEC the delay lines can be entered using the transmission lines function, and connected to the source using virtual connections - DW_Christman.ez.
Plots:
Primary (black) is 7.15 MHz, blue 7.05 MHz, and green 7.25 MHz.
The forward gain is fairly good at near 8 dBi, but the F/B ratio is low at around 7 dB.
Looking at the current magnitudes and phase in EZNEC the phase shift is about 105 degrees, and the current magnitudes are not equal (or close) at segment 25 (feed-points) which results in low F/B. The info given doesn't indicate if the polarity of one of the coax lines is reversed or not, reversing one further reduced the performance.
As noted in ON4UN's Low Band DXing book (5th edition, chapter 11, section 3.4.2) the 84 and 71 degree lines are calculated for a pair of ground mounted 1/4 wave verticals spaced 1/4 wavelength apart, and is derived from the feed-line impedance (50 ohm) and driving impedance of each element in the array.
How to improve it?
Playing with a 2 source model where the phase shift can be directly entered I found around 115 degrees peaks the F/B with a pair of horizontal dipoles spaced 1/4 wavelength apart.
I then wondered how much this could be improved upon and got reasonable results with F/B 15 dB or better and good SWR bandwidth with an L match using two different feed systems..
Coax Delay Lines and Current Forcing
Coax delay lines can be correctly calculated using a couple different tools to calculate the drive impedances.
Until recently I thought I had a solution using coax delay lines, but it was the result of making a mistake mixing up the leading and lagging drive impedances inputted into Arrayfeed1 or Feed2EL, somewhat surprisingly it almost works when the polarity of one coax line is reversed in the model!
I need to re-write this page with new examples, until then Phased Arrays - Christman Feed System has two examples on how to calculate these:
- Coax Delay Lines, this array type has no solution with 50 or 75 ohm coax, there might be a solution with 25 ohm (50 ohm paralleled) or 100 ohm coax, try it and see.
- Current Forcing, the 20m dipoles example could easily be scaled to 40m.
Opposite Voltage Fed
Opposite Voltage Fed (OVF) arrays were developed by Pekka Ketonen OH1TV. His site contains several examples in different configurations, including details on direction switching and matching networks. OVF uses 1/2 wavelength lines, at a common point where they meet a loading inductor is put in series with one line which makes the array directional, and with a relay electrically reversible. An L match network matches to 50 ohms. The system is simple and can offer broader F/B and SWR performance.
A previous post Phased Arrays - Opposite Voltage Fed (OVF) using a pair of elevated 1/4 wave verticals attempts to explain how the transmission lines, loading and matching networks are "wired up" in the model with virtual connections.
Key details:
- Height 13.75 m / 45 ft.
- Element length 19.65 m / 64.4 ft (#12 uninsulated wire).
- Element spacing 10 m / 32.8 ft.
- Loading inductor 3.2 uH.
- The polarity of one of the 1/2 wave lines must be reversed when using two.
- Gain 8.9 dBi, F/B 15 dB or better (under 60 degrees).
Model file: 40m_OVF_Horz_45ft.ez
Plots:
Primary is 7.15 MHz, blue 7.05 MHz, and green 7.25 MHz.
Don't get too excited about the 30 dB odd of F/B, it's a very specific elevation angle where this occurs. Attention should be given to overall shape of the pattern at the back half below about 60 degrees (3 notches up from 45 degrees). Below this point the F/B is 15 dB or better.
The gain, F/B, and pattern shape is very well controlled across the 40m band.
The SWR plot is in 100 kHz steps. Matched SWR is 1.5:1 or better across nearly the whole band except for the top 20 kHz.
There is also a relationship between the element lengths and the loading indicator value that results in better F/B ratio stability across the band which I discovered by trial and error.
Note that the polarity of one 1/2 wave line must be reversed.
Some thoughts on building this array:- The model would need to be updated if elements using insulated wire or tubing are used.
- Remove all but one dipole from the model, note its resonant frequency and impedance. Put a dipole up in each position separately adjust to match the single dipole in the model.
- Coax VF/loss is for LMR-400. 1/2 wavelength at 7.15 MHz with VF 0.84 = 17.6 m, I found increasing to 18m length and tweaking the loading inductor a little in the model helped keep SWR under 1.5:1 at the bottom end of the band. The VF of the coax being used to build the array would need be measured and entered into the model with adjustments as needed to account for it.
- At a height of 45ft the lines from each dipole will conveniently reach each other at ground level according to a triangle calculator.
- 1:1 baluns/current chokes should be used at the dipole feed points with any additional electrical length they add factored in.
- The direction of the array can be reverse simply by switching which line the loading inductor is inserted into where they meet.
- The loading inductor value in the model is critical as changes can drastically change F/B. This might be a sticking point when building the array is adjusting this and validating its right?
- OH1TV notes in at least one of the OVF articles that keeping the leads and lengths as short as possible in the switching and matching network etc is important as they add inductance.
With OVF arrays a range of element spacings work from near 1/8 out to at least 0.35 wavelengths. The closer the spacing the narrower the operational bandwidth.
1/4 wave spacing was chosen as this results in a matched SWR under about 1.5:1 across the band.
The weight issue with end supported dipoles and the balun and coaxing hanging could possibly be solved if the system can be/is reworked with using 450 ohm ladder line for the half wave lines.
The K5UA 40 meter Phased Array
The downside to using coax delay lines, or loading networks (OVF) is the complexity and having the test equipment, skill, time, trial and error to validate what you built matches what the model predicts.
The K5UA 40 meter Phased Array bypasses all of that and simply runs two equal lengths of coax from each dipole to the shack and into an adjustable phasing network, then an antenna tuner and into the the amp/rig. Varying the phase shift changes the elevation angle of the null to the rear allowing the operator to optimize the SNR (signal to noise ratio) of the desired signal.
The weight issue with end supported dipoles and the balun and coaxing hanging could be solved with equal lengths or 1/2 wave lengths of ladder line hanging to or near ground level then 1:1 balun/choke into equal lengths of coax back to the shack.
Conclusions, and which one?
Good question.
The K5UA system makes it simple, just build the phasing network and get an ATU.
I like the OVF array, it's an elegant system.
As with any of these a means of validating the tuning and performance using current probes and far field testing with a signal source to validate the F/B performance.
-----
I recommend reading the Phased Arrays chapter in ON4UN's Low Band DXing book (which now appears to be out of print!), it is a deep dive into the subject, and a lot of it is way over my head.
There are a couple important details about phased arrays that should be known:
The current magnitudes at the element feed points need to be equal, and with the right amount of phase shift. If either of these are not right, things fall apart.
With coax delay lines the phase shift of a transmission line is only equal to its electrical length when the line is terminated into its characteristic impedance. Which is pretty much never in a phased array, so we cant cut a line for x degrees between two antennas and expect it to work.
Delay lines calculated for one type of phased array can not be transplanted into another type as the driving impedances of the elements in the array will be different, even changing the height of an array can/will alter the driving impedances.
-----
Models are good starting point, and a way to investigate and better understand antenna systems. These tools can also help guide us to and validate the final result, if a good correlation is observed in the real world then we can have confidence the patterns and other information are accurate.
The models I have created and made available may contain errors, or overlook something someone more experienced can see. I don't claim to be an expert or authority on the subject of antenna modeling or phased arrays. I simply want to further my own knowledge and understanding of antennas which I find fascinating. Comments, suggestions, discussion are welcome - lonney@gmail.com.
This post is one of several on Phased Arrays.