Endfed 1/2 wave matching system end feed 
End Fed Half Wave AntennasEndfed halfwaves have become increasingly popular. They also have become bathed in misinformation, mystique, and contradiction. Some people swear by endfed halfwaves, while others swear at them. The reason for that is simple, endfed half waves have problems that affect consistency of user results. The purpose of this article is to show why opinions vary widely by including analysis's of typical antennas and measurements of components. The analysis of a particular style of end fed antenna should include best and worse case scenarios using realworld component values. It is imperative for any analysis to include data of common mode currents on feed lines and how changes might affect performance. Styles or Types of EFHW AntennasThere are four basic styles or types of end fed HW antennas:
The four styles above share one common trait. Absent a counterpoise of some type and proper care in construction, they all induce common mode currents on the feed system. The order of feed problem severity may surprise many of us. The list above is actually ranked in order of construction care required or construction difficulty! The two configurations thought of as "dipoles" are actually the worse of the systems. We'll see why as we analyze each type. I'll start below with the simplest system. Direct end feed or traditional longwire style with matching networkThe direct end feed antenna brings the end of the halfwave to a transformer or lumped component network of some type. The network is connected to something it "pushes" against so current can be forced into the antenna. The system looks like this: We can calculate endimpedance of an antenna in the clear with EZnec. The model is: The model for this antenna consists of the following wires: The feedpoint impedance (the little circle on the antenna view) is: EZNEC+ ver. 4.0
6/2/05 6:37:43 PM

SOURCE DATA

Frequency = 7.08 MHz The current distribution of the antenna is: CURRENT DATA
Frequency
= 7.08 MHz mean
value of current in
segment Wire No. 3
counterpoise 5 ft
long 4 ft high .25
ft per segment
One recently proposed theory is this; "if the antenna is not exactly resonant, ground currents will flow." We already know ground currents flow with a resonant endfed because endimpedance is not infinite. Let's change frequency 10%, since this would be equivalent to a 10% error in length of all sections, and see how much ground currents increase...EFHW standard
6/3/05 5:43:47 PM

CURRENT DATA
Frequency
= 7.788 MHz Before at resonance we had: lowest vertical antenna segment .17408a counterpoise first segment .16418a Being 10% off resonance just about doubles current. We see it is a definite current increase, but not one from zero to problematic currents! The proposal exact resonance eliminates all current in a counterpoise isn't correct. A 10% length error only doubles current. Moving 5% in frequency:EFHW standard
6/3/05 6:20:14 PM

CURRENT DATA
Frequency
= 7.434 MHz
With a 5% length error from resonance, we now see only an 18% increase in current. That's negligible since many other things we might do (like moving the antenna a few feet in height) would make a much larger change. 
SOURCE DATA
Frequency
= 7.434 MHz We can see there is some merit to maintaining resonance because current is at a minimum value, but we only need to worry when resonance errors are somewhat large. When length errors are modest (under 5%) the error has virtually no effect on ground current. The reason for this is very simple. The reactance or lack of resonance isn't what determines current, the resistive part of the impedance does. We are looking for a resistance peak in the endimpedance of the antenna...not necessarily resonance. The source resistive part at resonance was 3300 ohms. At 5% error it was 2382 ohms. With the nonresonant antenna, we have increased electric fields around the feedpoint. Voltage is 925v instead of 575 volts. RF voltages (the electric field) might be an issue with endfed antennas. What else affects Antenna Impedance?From above we see higher antenna resistance is a good thing for current, and length is not overly critical. We also see lower reactance is a good thing for voltage, and length can affect voltages (and the electric field) surrounding the antenna and counterpoise near the feedpoint. What about a thicker antenna? With a 1" thick antenna 7 MHz impedance becomes 1684  J 716.3 ohms and resonance is well below 6 MHz. The reactance problem is because the counterpoise is too short. The drastic resistance reduction at peak resistance occurs because the wire is thicker. Obviously a thicker antenna has higher ground currents! Halfwave broadcast towers often have impedances under 800 ohms at resonance. What about antenna surroundings? As the area around the antenna becomes more cluttered and/or has more power loss, antenna endresistance is reduced! Over perfect ground the antenna endimpedance almost doubles from that over average ground. Over lossy ground, especially when the antenna is low in height, feed resistance decreases even more. With a small counterpoise and endfeed we:
ImprovementsThe best solution I can think of to common mode or "RF in the shack" problems with this form of antenna is to isolate the counterpoise or antenna ground from the station feed. At low power levels a simple link coupled matching network is a good solution, provided the secondary has no RF path to station equipment. One way to accomplish this is by using two output terminals that float. In the circuit above:
Assume we have a 5000 ohm load and 50 ohm rig. The turns ratio of T1 is sqrt of 5000/50 or a 10:1 ratio. The reactance of C1 at 3/4 mesh (so you have adjustment range) should be 5000/10, or 500 ohms. (This is a loaded Q of ten, you need LESS Q with lower transformation ratios and more Q with higher ratios or the circuit becomes too sharp or too "mushy" to tune.) The reactance of L1 secondary should be 500 ohms in this example. U1 should connect to the antenna, U2 to the counterpoise or ground which should NOT connect to the station equipment. The counterpoise should be as long and straight as possible, and directly under the antenna if possible. Ideally the counterpoise, if less than 8 wires 1/4 wl long, should be elevated and insulated from earth. Do NOT make the counterpoise longer than 1/4 wl, especially if it is only a single wire! Most of us, since this is a temporary or compromise antenna, will use a very small ground system. I've found connecting a counterpoise to earth, say a ground rod, actually reduces antenna efficiency. If you run low power and don't have a ground or counterpoise, you might just connect U2 right back to the coax shield from the radio. This way you can use the capacitance of the radio and station wiring as a ground system.
To
be continued soon!

Capacitor  Cap RMS Voltage  Power at load  Power lost  % eff  Loss 
Apollo Series  98.5  2.07 W  2.93 W  41.4 %  3.8 dB 
OM series  109.6  2.56 W  2.44 W  51.2 %  2.9 dB 
Air Variable  132.8  3.75 W  1.25 W  75 %  1.2 dB 
Of particular concern is the low Q of compact capacitors as well as upper impedance limits of compact powdered iron cores wound with large numbers of turns. It would be much better to use a small air variable rather than notoriously troublesome "transistor radio" tuning capacitors.
The inductor could also be improved with a taller core or stack of cores. That would minimize the number of turns required for resonance! Also a different mix might increase Q.
On the previous page, I mentioned the impedance ratio of a tightly coupled transformer of high quality was equal to the square of the turns ratio between primary and secondary. If we wanted to match a 4700 ohm load to a 50 ohm source, the turns ratio should be sqrt of 4700/50, or 9.7 : 1.
4700 in parallel with 3300 is 1939 ohms. 1939 ohms over 50 ohms is a resistance ratio of 38.78. The square root of 38.78 is 6.23. 28/6.23 is 4.5 turns. I calculate a 4.5 turn primary based on perfect mutual coupling, the 3.3k secondary loss equivalent parallel resistance, 4700 ohm load, and a 28turn secondary.
4.5 turns isn't possible, and won't be exact anyway because of flux leakage, lead lengths, and other small imperfections in the system. The plot below is with a 5 turn primary.
Let's work the problem backwards. The 28 turn secondary / 5 turn primary transformer I measured above has a turns ratio of 5.6:1 5.6 squared is 31.36 The measured primary impedance was 62 j0, times the impedance ratio of 31.36, for a net secondary impedance of 31.36*62 = 1944 ohms. That isn't terribly far from the 1939 ohms calculated as 3300 ohms of tank parallel equivalent loss resistance in parallel with 4700 ohms of load resistance!
Measuring things two different ways and comparing results is very useful. It warns us of any errors in measurements, logic, or misapplications of theory.
When I measured the secondary impedance, I calculated I'd need about 4.5 turns on the primary to match a 4700 ohm load in parallel with a loss resistance of 3300 ohms, or 1939 ohms. When I optimized the transformer, I found 5 turns was close enough.
So there we have it. Matching loss in the worse transformer system I measured is around 4 dB. If the antenna was 4700 ohms j0, more power would be consumed in the tank circuit feeding the antenna than is actually applied to the radiating part of the antenna system! Of the power applied to the antenna, a significant portion will be distributed in the antenna feed cables, rig, and everything connected to the rig including the operator and station wiring.
On the other hand we could build a dipole, and feed it with RG174 cable. We would eliminate about 4 dB of matching circuit losses, and a few dB of power lost in common mode losses. How much RG174 could we use to equal the endfed losses? At 50 MHz we could use about 100feet of RG174 and break even with end feed using the system I constructed, a small T502 toroid with link coupling resonated by a typical polyethylene insulated "broadcast tuning" capacitor. At 7 MHz it's a nobrainer. I'll take the RG174 dipole feed every time! My reasons are mostly the convenience and repeatability of the feed system.
Another possible alternative is a 1/4 wl 450ohm Q section or stub. TLA estimates loss as .205dB on 7 MHz with that system.
In my experience it's far easier to have repeatable results with a stub, rather than small inconsistently manufactured lumped networks. Even large tuners using transmitting components become inconsistent in loss measurements at impedance extremes. Losses that would cause big smoke at high power can go unnoticed at low power levels..
Endfed half waves are sometimes inefficient or troublesome feed systems. At low power we might never notice efficiency problems or common mode current problems. Of course at high power very few shortcuts can be tolerated.
as of 5/27/2005
This page copyright W8JI 2005. All drawings and text may be used when discussing specifics of this article (that's fair use), but may not be used in any other form.