An incoming electromagnetic wave illuminates both an antenna and its feedline. Current induced on the outer surface of coaxial cable may enter the cable at the antenna feedpoint. Common-mode current induced on parallel line may become differential at the receiver. Either makes the feedline an unintended antenna conductor that can degrade the directive pattern and couple noise from nearby sources. A 75Ω current balun can attenuate unwanted coax shield current. A 300Ω balun can attenuate common-mode parallel line current.
You can make a simple current balun by coiling coaxial feedline in a particular way. At self-resonance the coil acts as a parallel trap whose high impedance inhibits unwanted shield current.
To construct the balun, mark RG-6 coax with tape at two spots 27″ apart (26″ for RG-59). Coil the coax into three turns with the marks aligned. Secure the coil with dark, UV-resistant tie wraps in several places so that adjacent turns everywhere touch. Route the coax leads straight away from the coil.
A current balun is most effective when placed at the feedpoint where shield current is highest. When that's not possible, form the feedline into a second balun a quarter wavelength (30″) away from the first. Multiple spaced baluns are particularly effective at reducing shield current induced by asymmetrical coupling to the antenna, as may occur when the antenna is vertically polarized.
The coil may not meet the cable's minimum bend radius specification. Bending coax too sharply may cause an impedance change, or with certain dielectrics and enough time, an internal short. The spec varies among cables and manufacturers. It is 2˝″ for Belden 1505A RG-59. For 1530A RG-6, it is 3″.
Ken Wetzel uses Belden 1855A coax because it is small, flexible, and has a very accurate characteristic impedance. He uses three turns with an outside diameter of 223⁄32″. He bonds the turns together with superglue and uses two tie-wraps to secure the exit leads. The minimum bend radius for 1855A is 1˝″. Loss is 3.33 dB/100′ at 100 MHz.
Times Microwave LMR-300-75 should make an excellent coiled-coax balun. Its minimum bend radius is ⅞″ and loss is just 1.75 dB/100′ at 100 MHz.
Hans-Peter Dohman, DL9EBA, uses this balun on a 75Ω Yagi. Mounting it perpendicular to the elements minimizes adverse coupling.
This shows shield attenuation from 88 to 108 MHz in a 50Ω system. The attenuation range corresponds to a common-mode impedance of 1.9kΩ to 63kΩ. Coil self-resonant frequency may vary with coax diameter, jacket dielectric constant, winding tightness, and lead dress. This particular balun resonated at 97 MHz.
Test setup. HP 8443A tracking generator driving the balun shield to an HP 141T/8553B/8552B spectrum analyzer. When making a measurement, I move the balun away from the conductive surfaces with an insulated tool and keep the leads away from the coil.
Shield current suppression degrades as shield reactance approaches the negative of choke reactance. Modeling a 0.15″ diameter conductor a few feet long (antiresonant at 98 MHz) fed against a split Yagi driven element yielded a 781Ω reactance peak within the FM band. Peak reactance was much smaller for longer conductor lengths typical of feedline. The model suggests that the 1.9kΩ minimum measured choke impedance is adequate.
Passing coax through ferrite material increases its common-mode impedance without affecting its differential impedance. A Laird 28A0807-0A2 snap-on, split-ferrite core, stocked by Arrow, Mouser, and Digi-Key, is simple to install. Rated impedance is 348Ω at 100 MHz. It will accomodate quad-shield RG-6.
The plastic closures of a split-ferrite core may become brittle and fail when flexed after long outdoor exposure. Tape or tie-wrap a broken housing to ensure that the ferrite halves remain firmly joined. The ferrite material does not seem to degrade outdoors.
A nonsplit ferrite core, installed before terminating the cable, is more robust than a split core. A Fair-Rite 2643625202 from Arrow, Mouser, and Digi-Key has a rated impedance of 361Ω at 100 MHz (Ken Wetzel measured 330Ω). It will accomodate quad-shield RG-6.
A coiled-coax balun yields much higher impedance than a ferrite choke balun and this will better attenuate unwanted shield current. Ferrite chokes provide moderate, reliable, broadband attenuation in a lightweight, compact package. You may prefer them when size or appearance matters, or when you have doubts about the self-resonant frequency of a coiled-coax balun. To increase the impedance and unwanted-signal attenuation of a ferrite choke balun, use more than one core.
To evaluate the effectiveness of coiled-coax and ferrite choke baluns, I modeled a highly directive narrowband Yagi in free space with AO 9.52. Due to its extremely small backlobes, the antenna is very sensitive to stray signal pickup. I added a conductor to one side of the feedpoint to represent the coax shield. In practice, the shield impedance and resulting current depend on the length of the coax, what it couples to, and what it connects to. Since these parameters are unknown, I modeled a traveling wave on the shield as a general, nonresonant example. As you lengthen any conductor, it develops a traveling wave as the incident power gradually radiates away. I created a traveling wave on a relatively short wire by placing a 350Ω load a quarter wavelength from the far end. I adjusted the load impedance and position for the most uniform wire current. Analysis at 88.1 MHz tests the worst-case shield current suppression of a coiled-coax balun self-resonant at midband.
This is the model geometry. The red dot is the feedpoint and the green dot is the traveling-wave termination. The yellow traces represent current magnitude. A traveling wave on a vertical wire radiates mostly downward. To examine a worse case, I bent the shield wire horizontal six feet below the Yagi. The horizontal section is 20 feet long. It is orthogonal to the elements, bisects them, and does not couple to them. Note the discontinuous driven-element current and the nonsinusoidal shield current.
With a circuit analysis program I modeled a coiled-coax balun as a parallel trap in a 50Ω system. I adjusted the component values to obtain the response shape I had measured for the test balun and centered it at 98 MHz. Then I used the values for an RLC load that I placed in series with the shield conductor at the Yagi feedpoint.
This shows impedance magnitude (red) and the resistive part (blue) for the RLC trap model.
This magnified view sights down the horizontal section of the shield conductor coincident with the green dot. Here the current traces are phasors. The distance from the wire to the trace is magnitude, while the angle with respect to the wire is phase. A slowly decaying spiral is characteristic of a dissipating traveling wave.
Azimuth pattern with and without the coax shield. Shield current at the feedpoint was 6.6% of maximum model current. A linear-dB scale with the center at −50 dB reveals low-level detail. To account for all polarization components, the curves plot the total field.
These patterns are for the Yagi plus shield with a coiled-coax balun or two adjacent Fair-Rite 2643625202 cores modeled as a 660Ω resistance. The coiled-coax balun is more effective even far from self-resonance.
To try a resonant shield, I removed the termination and adjusted the horizontal wire length to maximize its current, which occurred at 231″ and was 19% of model maximum. The balun is more effective than for the traveling wave case since its impedance is a higher percentage of the total.
Here I adjusted the horizontal wire length to 194″ to minimize its current, which was 2.9% of model maximum. The higher common-mode circuit impedance reduces balun effectiveness since its impedance is a smaller percentage of the total.
Yagi/Balun Model ; use 27 segments/halfwave Free Space 88.1 MHz 9 6063-T832 wires, inches x1 = 0 ; element positions x2 = 21.9375 x3 = 37.25 x4 = 76.8125 y1 = 67.375/2 ; element half-lengths y2 = 67.1875/2 y3 = 61.6875/2 y4 = 53.6875/2 a = -2 ; position of balun below antenna b = -72 ; position of shield bend c = x2 - 209.5 ; position of traveling-wave termination d = c - 30.5 ; position of shield endpoint 1 x1 -y1 0 x1 y1 0 0.375 ; reflector 1 x2 -y2 0 x2 0 0 0.375 ; one side of driven element 1 x2 y2 0 x2 0 0 0.375 ; other side 1 x3 -y3 0 x3 y3 0 0.375 ; first director 1 x4 -y4 0 x4 y4 0 0.375 ; second director 1 x2 0 0 x2 0 a 0.15 ; dipole to balun 1 x2 0 a x2 0 b 0.15 ; balun to bend 1 x2 0 b c 0 b 0.15 ; horizontal run to termination 1 c 0 b d 0 b 0.15 ; beyond termination 1 source Wire 2, end2 39 pF ; shunt capacitor for 75-ohm match 2 loads Wire 7, end1 .75 uH 3.52 pF 3 ohms ; RLC balun model resonant at 98 MHz Wire 9, end1 350 ohms ; termination to create traveling wave
I modeled a shorted folded dipole fed against a vertical conductor to determine the common-mode impedance and resulting mismatch loss of a typical Yagi driven element and coax shield. For shield lengths between 15 and 30 feet, 75Ω mismatch loss was at most 6.3 dB. This demonstrates in another way that a nonresonant feedline length provides little protection against unwanted common-mode signals.
Most 300Ω consumer baluns use two bifilar windings on a binocular ferrite core interconnected as shown. When driven at the 75Ω terminals, these baluns produce equal voltages of opposite phase with respect to the coax shield at the 300Ω terminals. Conversely, voltages common to the 300Ω terminals cancel at the 75Ω terminals. The purpose of the winding short may be to disable the bifilar winding while retaining its stray reactance for symmetry. With input and output shorted, common-mode impedance typically is less than 20Ω.
Some baluns omit the winding short. These devices produce unequal voltages at the 300Ω terminals but tend to equalize current in the parallel lines. A single turn at the center of the core and two wires connected to the coax shield instead of four identify a current balun. Common-mode impedance typically is 250Ω.
Some voltage baluns use a toroidal core with a single bifilar winding. What look like ceramic capacitors are integrated RC networks. Not all toroidal baluns have these networks, and they can be found in binocular baluns as well. The capacitance raises common-mode impedance below the balun passband, perhaps to attenuate strong AM signals or low-frequency lightning impulses. The resistance drains any static antenna charge.
To compare the common-mode rejection of various ferrite baluns, I cut a 31″ piece of 300Ω twin-lead. This is a quarter wavelength at the low end of the FM broadcast band. Fed against a surface, a conductor of this length makes an effective antenna. I terminated the twin-lead with a 300Ω resistor at one end and spade lugs at the other. I connected it to several baluns that I plugged directly into a Wavetek SAM RF-level meter, which I tuned to a local FM signal. Keeping the orientation of the twin-lead constant, I swapped baluns and compared signal levels. Except for a tiny residual, any signal was due to unwanted common-mode balun response.
After connecting the twin-lead directly to the SAM as a reference, I measured the following signal levels in dB for several baluns:
Binocular voltage baluns <-40 <-40 -35 -29 -27 -26 Binocular current baluns -20 -19 Toroidal voltage baluns -18 -18
The toroidal baluns had the worst common-mode rejection. The current baluns were almost as bad even though their common-mode impedance was much higher. I had created one of the current baluns by unsoldering the winding short in a voltage balun. When I pressed it back in place with an insulated tool while watching the meter, the unwanted signal dropped about 10 dB.
These are the two <-40 baluns, presumably binocular voltage baluns. Mismatch loss due to the longer leads should have increased common-mode rejection only 0.5 dB. As well as the best rejection, these baluns had the lowest transmission loss, 0.5 dB. The others were push-on baluns with screw terminals with about 0.75 dB loss.
Even with perfect common-mode balun rejection, nonzero transmission line spacing causes a residual differential signal. Simulation of the terminated twin-lead oriented vertically above an infinite ground plane gave a peak differential signal 44 dB below the peak common-mode signal.
You can increase the common-mode impedance of a 300Ω voltage balun by following it with a 75Ω current balun. The cascaded common-mode impedance is the sum of that for each balun.
A halfwave coaxial line that inverts signal phase makes a low-loss 300Ω voltage balun. Here Peter Körner used small-diameter coax to save space. Two ferrite chokes enhance common-mode attenuation. All shields connect. The halfwave-line center conductors go to the two antenna terminals, while the feedline center conductor goes to one of them. Keep all leads as short as possible, as Peter did here.
An electrical half-wavelength at 98 MHz is 60.1″ multiplied by the velocity factor of the coax. Use the manufacturer's value for your cable.
This circuit models the loss of a halfwave coaxial balun. It uses a lossless transmission line with a characteristic impedance of 75Ω and a delay of 5.1 ns. The T-pad models the 0.083 dB loss of a 50″ length of RG-6 (2 dB/100′ loss and 83% velocity factor assumed). The circuit does not account for any ferrite core loss.
This circuit models the common-mode voltage rejection.
This Triax FM 5 balun uses serpentine traces on a printed circuit board to form a halfwave transmission line.
The PCB bolts directly to a folded dipole. A plastic enclosure weatherproofs the balun and connections.
An L-network followed by a current balun can match and isolate a 300Ω load. When made with a ferrite core, this balun has the low loss of a halfwave coaxial balun but is more compact. It is larger when made with coiled coax but avoids any ferrite loss. Span the antenna terminals with #18 bare copper wire coiled at the center into 5 turns, ˝″ diameter and 5⁄16″ long. Using the shortest possible leads, solder the capacitor and coax center conductor to the wire ⅛″ from the coil ends. Connecting the capacitor to the coax shield yields the shortest leads, but 22 pF in one lead and 33 pF in the other can substitute for 13 pF. 24 pF and 27 pF also work. Use this Windows program to design a different 282 nH coil at 98 MHz.
Loss for the L-network balun. Blue is for coiled coax and red is for one or more ferrite cores. Core loss is not included. It increases with outside shield current since the core impedance is largely resistive at 98 MHz. Shield current does not affect coiled coax loss since its impedance is reactive.
Over the years, many kinds of baluns have been proposed and built that don't actually work. If you see a ground symbol in a balun schematic, beware. It may just indicate circuit common. But if it is an external connection, the designer has almost certainly confused RF and DC ground. Far from the earth's surface, no non-radiating RF current sink seems to exist. This includes the antenna boom, which will simply radiate any current a balun directs to it. You'll find several examples of improper baluns here.