saved from url=(0063)

All-band HF antenna system for difficult EMC environments.
Using the SGC 'Smartuner'* in a balanced antenna configuration.
* 'Smartuner' is a registered trademark of SGC Inc.

In the old days, when I first had my amateur radio licence, setting up an HF amateur station could be as simple as tying a piece of wire to a half-brick an tossing it into a tree. The secret lay in the p network, which inhabited the zone between the output stage and the antenna socket of every transmitter, and which appeared to match practically anything against a moderately good earth. The problem of course, was RF in the house, lots of it; but then again, there was nothing much for it to upset. There was perhaps the odd television which gave trouble, but the solution was invariably straightforward; and domestic radios by then had ferrite rod antennas, which seemed to be immune to strong electric fields.

Then came Hi-Fi systems, each with a cute folded-dipole antenna with a loudspeaker at the end of each arm, and a habit of collecting RF and injecting it into the negative-feedback loops of two large audio amplifiers. We learned to tame these beasts with ferrite beads and bits of co-ax; but it was a loosing battle in the face of what became known as the 'consumer electronics boom'. Then came the total breakdown of society, and burglar alarms; with their webs of unshielded cable leading to boxes designed to scream blue-murder in the event of any electrical anomaly, and the game was over. Transmitters mutated accordingly; the magic p-tank disappeared, to be replaced by the mute and forbidding 50W output terminal and a set of dire warnings about the consequences of applying a reactive load. Operating practices mutated also, and gave way to the cult of the co-ax fed resonant antenna, and the various abominable hybrids designed to operate on more than one band. The logic is simple; get the antenna away from the building, and get the currents equal and opposite on both sides of the feed line so that the fields cancel out. Your near-field EMC worries will then be over (well almost), as will your ability to radiate on any arbitrary frequency.

A centre-fed dipole antenna in free space has a characteristic impedance of about 73W at its fundamental resonant frequency. Bring it a bit closer to the ground and the impedance drops to somewhere around 50W. If the capacitance to ground of the two arms is not equal, you can wind a coil in the co-axial feed line (called a choke balun) and force the feed currents into symmetry. The near-field is also predominantly electric, so that it is not very good at inducing currents in ring-mains and ferrite rods. The dipole is clearly a front-runner as an antenna for difficult EMC environments but, if you follow the advice given in the text books, the measures required to get it to work on a wide range of frequencies appear troublesome to say the least. It was in considering how to re-establish HF operations at a new QTH, and having experienced EMC mayhem on trying to load up an end-fed wire, that I almost settled on the unsatisfactory compromise of a trapped-dipole and weeks of messing around to make it work on a subset of the available bands. I was however, then suddenly gripped by the heterodox but unoriginal thought; 'why does the dipole need to be resonant anyway?'. A cursory reading of the literature will reveal that there are no efficiency implications in using a random wire, provided that the reactance at the operating frequency is canceled out. Surely what is needed then, is a remote-controlled antenna tuner, on a stick in the middle of the antenna, with a current-balun between it and the transmitter? (In fact, you don't even need to put the marching unit right up in the antenna, you can also feed it with an open-wire line). It was in discussing such matters with other amateurs that I heard about the SGC 'Smartuner'.

The SGC 'Smartuner' is not merely a remote controlled p-network; it is an automatic p/L-coupler, with a built-in control algorithm which is nearly as good at tuning-up as a human, but much faster. There are various models in the 'Smartuner' family (link to SGC website given at bottom of page), but a popular model is the 1.6 - 30MHz 'SGC-230', which is nominally rated for 200W PEP.. A view of the SGC-230, with its cover removed, is given below:

Instead of using continuously variable tuning elements, the SGC antenna coupler uses relay-switched capacitors and coils, which are selectable in binary increments. The input (TX side) capacitors and relays are on the left in the picture above, the coils lie across the top, and the output capacitors are on the right. Note that the coils are arranged to minimise the magnetic coupling between them, so that they may be selected without errors due to mutual inductance. Note also, that the output capacitors are switched by pairs of relay contacts in series (hence 10 relays for 5 values), this being done to reduce the risk of flashover when attempting to drive high impedance loads. The antenna terminal is the high-voltage insulator on the right of the box, the earth terminal is the braid at the bottom left. The input 50W co-ax. line from the transmitter goes to the small orange barrier strip close to the earth terminal (the supplied connecting cable has been removed in this photograph). The available tuning element values are summarised in the table below.
Tuning element Values Step Size Max combined
TX side 0, 100, 200, 400, 800, 1600, 3200 pF 100pF 6300pF
Inductor 0, 0.25, 0.5, 1, 2, 4, 8, 16, 32 mH 0.25mH 63.75mH
Antenna Side 0, 25, 50, 100, 200, 400 pF 25pF 775pF

From the step-sizes used, the matching unit might first appear to offer a rather coarse-grained solution to the problem of tuning up; until you realise that the device offers some 2^19 (ie., 524288) possible element combinations.

The coupler is operated by a 6805 microcontroller, with an embedded proprietary tuning algorithm. The processor clock crystal is on the NTSC colour subcarrier frequency of 3.579545MHz; slap in the middle of the 80m amateur band; but no reception interference is caused because the clock is turned off in the absence an of RF input from the transmitter. Tuning is accomplished by sampling the following quantities: Input Frequency, Forward and Reflected Power, Input Impedance (V/I), and Input Reactance (V - I phase difference, f). The RF detector section of the unit is shown below:

Outputs from current transformer T1 and voltage transformer T2 are rectified and combined to form an impedance bridge, which is balanced for 0V output when the impedance looking into the antenna coupler is 50W. T1 current sample, and a limited voltage sample from the line are also combined in a Mini-Circuits SBL1 double balanced mixer, to form a phase bridge which gives 0V output when the input to the coupler presents a resistive load. Positive going signals proportional to forward and reflected power are obtained from a directional coupler* using T2 and T3. Frequency is sampled by limiting the signal from the line and applying it directly to the ripple-clock input of a 74LS93 counter.

[* Stripped of the output for the phase and resistance bridges, the directional coupler is an implementation of the Sontheimer-Fredrick bridge configuration used by John Grebenkemper KI6WX in his directional wattmeter article in QST, Jan 1987. See also ARRL Antenna Book, 19th Edn 2000, ISBN 0-87259-804-7) "The Tandem Match" ch 27, pp9-19]

The tuner has a non-volatile memory in which it stores tuning solutions as a function of frequency. When forward power is detected for the first time after powering-up, it sets the tuning relays according to the data in memory for the current frequency and then checks the SWR. If the SWR is less than 2:1, the 'tuned' line on the control cable (see below) goes low and the processor halts. If the SWR is greater than 2:1, the control program branches to the 're-tune' routine; the object of the exercise being to cancel the reactance of the antenna and choose inductance and capacitance values which effect the required impedance transformation. If the re-tuning is a success, the 'tuned' line goes low, the new solution is stored in memory, and the processor halts. On the next transmission, the processor awakes briefly to see if the frequency has changed or if the SWR has risen above 2:1, and goes back to sleep if they have not. In this way, the unit learns the required tuning element values for a particular antenna at a particular frequency. If a good solution has been found previously, the tuning process takes about 20ms, which is the time it takes to determine the frequency and close the appropriate relays. If re-tuning is required, the process can take up to about 3s as the unit steps through its inductor and capacitor values; resulting in an initially high SWR on the line from the transmitter, which rapidly drops as a solution is approached. SGC claims that the unit will usually produce an SWR of better than 2:1 on all frequencies for which a tuning solution exists. In practice, in the author's installation, it usually achieves an SWR of better than 1.7:1 in the working range 1.81 to 29.7MHz, which results in reasonably stress-free operating conditions for the transmitter PA (but how much of the energy is radiated is another matter). Since some transmitters will start to reduce power at about 1.4:1 SWR, it is preferable to use a transceiver with a "built-in ATU" (known more realistically as a "line flattener"), to ensure that the transmitter always sees an ideal load and gives maximum output with low intermodulation..

It takes about 3W of forward power to activate the automatic tuning system. Clearly there will be some insertion loss due to the two current transformers in the line, but is not an issue in normal circumstances. The presence of various limiting diodes however has implications with regard to receiver performance: in that non-linear devices in the antenna system may give rise to cross-modulation and harmonic generation from strong nearby transmitters. No data are available regarding the tuner's performance in this respect, but it has been noticed by the author that the second harmonic of a nearby amateur station ("400W" on 14MHz, at a distance of 2Km) is considerably enhanced in comparison to the signal from a simple wire receiving antenna.

The interface cable to the SGC 230 consists of a single sheath containing the RG58 feed-line and four wires: +13.8V, 0V, 'Tuned' and 'Reset / Lock' (in the manual, the 'Tuned' line is called 'Tune', but this is a misnomer since its purpose is to indicate that the tuning process has been completed). In a minimal configuration, it is only necessary to feed RF into the line, which is supplied fitted with a PL259 plug, and connect the 13.8V supply. The basic model has no provision for tuning-up in the absence of a transmitter, and so the unit is not ideal for short-wave listening.

The only real disadvantage of the SGC-230 from the author's point of view, is that it is not designed for use with balanced antennas. It is also usually a bad idea to use a balun transformer on the feeder to a non-resonant antenna. The tuner however, does not need a ground reference in order to work; and so it is perfectly sensible to suppress common-mode feeder currents by winding the feed-line and the power / control cable together into a choke balun. The author's first attempt at such an installation was successful, and is shown below:

The tuner is mounted at the top of a 2.5m fibreglass pole, which is clamped by means of a straight-coupler to a standard 2" diameter aluminium mast. One side of the dipole goes to the antenna terminal, and the other side goes to the earth terminal. The arrangement, although physically asymmetric, is electrically symmetric due to the action of the choke balun. The cable supplied with the tuner is wound around the fibreglass pole (13 turns) and brought to a junction box, which is made from an old underwater light meter housing fitted with three IP68 cable glands. Inside the junction box, the RG58 feedline is connected to an RG213 cable from the transmitter (the original PL259 plug is connected to an SO239 free socket) and the power and control wires are connected to a 4-core shielded cable rated at 7A (the current consumption of the SGC230 is less than 1A, but a decent cable is needed to avoid voltage loss on long runs). By leaving the supplied cable intact (feeding it through the junction box gland from the inside and then reconnecting it to the tuner) the tuner can be restored to its factory supplied condition if necessary, but care is needed in re-sealing the weatherproof enclosure properly. The junction box splits into two parts, with the cables from the house attached to the cover-plate and the cable from the tuner attached the box, so that the tuner can be disconnected easily for service. To make the choke balun, the two cables from the junction box were taped together at intervals and then wound together into a flat 10-turn coil of about 40cm diameter, which is held together with tie-wraps. Down from the mast, the two cables enter the house through a soffit vent, and are coiled into yet another choke (this time 5-turns) in the attic-space, before passing into the radio-room. Total length of the RG213 feed-line, including the chokes, was 39.8m. The fact that the installation worked first-time (as expected) indicates that the exact details of the choke balun arrangement are not critical. Pi Tank on a stick.

The arms of the author's first dipole were each 17m long, with a 1m gap between the insulators at the feed-point, to give a total length of slightly more than 35m, with a measured resonance at 3.9MHz.

On commissioning the antenna, tests were made to see if the SGC-230 could tune-up the antenna on all traditional and WARC bands, as indeed it could, and to see what EMC problems would result. Surprisingly, given the intractable difficulties encountered previously, the only immediate problem was that of one computer; which flagged a keyboard buffer over-run error and promptly crashed when a signal was emitted on 40m. The computer in question had an unusually long coiled keyboard cable, which looked as though it might be resonant on 7MHz. A ferrite ring at the keyboard socket (right) fixed the problem. Otherwise, the choke-balun arrangement was so effective at keeping RF out of the mains, that a Maplin mains intercom system (120KHz FM), which emitted loud un-demodulated SSB noises in previous experiments, is now usable while the transmitter is running. The ghastly alarm-system from hell however, still has a tendancy to go off when signals are transmitted on frequencies of 7MHz and above; but the problem can be eliminated by putting the alarm into 'engineer' mode (in which case it ignores tamper signals).

As far as propagation characteristics are concerned, the antenna system described is suitable for NVIS (near-vertical-incidence sky-wave) communications at low frequencies, moving gradually to lower-angles as the frequency is increased; the frequency at which the low-angle radiation becomes useful being a function of the height, particularly at the feed point. In the author's original configuration, the feed point was only some 12m above ground, and only 7.5m above the flat roof of an annex building. Using 100W PEP, G3YNH (near Exeter, in Devon) recieved signal reports of 59+ from all over the British Isles and Western Europe on 40m and 80m. On 160m, where the antenna is electrically short and the system much less efficient, 59 reports appear to be confined to a radius of a 350 - 500 km, depending on atmospheric conditions, but perfectly good 58-59 QSOs have been had with stations in northern Scotland (about 1100km). The transmitter site is approximately 550 ft ASL, with a good take-off to the West, and nightly transatlantic 57-59 contacts could be made without difficulty on 20m. Long-path VK QSOs under suitable conditions elicited signal reports of 56-57. Later changes to the arrangement of support masts raised the average height of the antenna to 15m. This modification increased transatlantic signal reports on 20m by about 12dB, and elicited 59+ reports from canada on 80m (under suitable conditions). The antenna has therefore proved itself to be a very acceptable all-rounder, but is, of course, no substitute for an HF beam.

Note that on higher frequencies, the antenna is effectively a centre-fed long-wire. This means that the radiation pattern breaks up into multiple lobes, with gain in some directions and loss in others. In the authors's installation. with the wire running on a bearing of 110, signal reports give evidence of a null in the direction of central USA on 17and 20m. Such nulls are a fact of life with long-wire antennas and, if possible, should be taken into consideration when choosing the orientation of the wire.

The only significant drawback with the antenna system as described is its poor performance on receive on some bands until a carrier has been sent to the SGC-230 to make it tune-up. This is entirely the author's fault, for using 40m of feed-line contrary to the recommendations in the manual; but is also a necessary consequence of using a choke balun. The attenuation is not serious on frequencies close to odd-order resonances (ie., on 80m and 20m), but is significant on frequencies close to even-order resonances (ie., on 40m and 10m). The difference in received signal level between tuned and untuned states on 40m is about 30dB. One short press of the Morse key will bring the antenna to life, but this can be anti-social; and there is no proper provision for listening outside of the amateur bands. In practice however, due to the high noise levels on the HF bands, the lack of receive sensitivity in the untuned condition is not usually a problem.

In addition to the receiver desensitisation problem when the antenna is untuned, there may also be encountered frequencies on which the tuner has difficulty in finding a unique tuning solution. If this occurs, the tuner will keep on trying to retune during the course of an SSB transmission, and thereby disrupt station operation. The problem occasionally crops up on 160m in the installation described above. SGC is aware of it, and has provided a solution in the form of the 'reset / lock' control line (revision T onwards), which can be tied to the +13.8V supply to prevent retuning. Access to this functionality can be had by purchasing the optional 'Smartlock' unit, or equally well by building the simple control box shown here:

(Legend: NO = Normally Open, NC = Normally Closed)
The SGC-230 must be supplied with power even when receiving on an untuned antenna, because it powers-down with all of the inductors in series with the antenna. At power-on, all tuning elements are removed until the first transmission, and so the bypass button allows this condition to be achieved without turning off the PSU (which, to avoid earth loops, should preferably be the same as the one used by the transceiver). Tuning, or retuning, is prevented by moving the switch to the 'Lock' position; while pressing the 'Retune' button in the auto position forces the coupler drop all tuning elements and retune on the next transmission. It might appear, from this, that the 'retune' and 'bypass' buttons perform the same function; but the 'retune' button forces the tuner to branch to the 'retune' routine regardless of the data in memory, whereas the 'bypass' button allows the tuning to be taken from memory when possible. It is useful to be able to drop all tuning elements when changing bands, and using the bypass button enables retuning to take place in the shortest possible time after having done so. Note that the 'Tuned' LED has a series resistor of 150W, while the other LEDs have 620W (for approx. 20mA @ 13.8V). This is because the 'Tuned' line already has 470W in series with it inside the coupler unit. In the author's installation, an additional high-brightness 'Tuned' LED with a 680W series resistor was placed inside the clear junction box on the mast; the idea being that the LED in the control box, with its much lower series resistance, would steal current from it and prevent it from lighting once the control box was built. In practice, both LEDs light perfectly well.

At the time of last update (June 2002), the SGC-230 retails in the UK for 369. While this may seem like a substantial investment; it is worth noting that it would cost about as much to buy the components required (don't forget the weatherproof housing), and it is difficult to imagine a more straightforward solution to the problem of how to radiate on all HF bands without making your garden look like RAF St. Eval.

Post Script (Sep 2002):
The installation described above was taken out of service in May 2002 and the equipment passed to Andy Cowley M3ABC (also M1EBV), who will report his experiences in due course. Reasons for dismantling the intallation were not due to dissatisfaction with its performance, but due to a desire to increase transmitter power to the UK legal limit (400W PEP). When the unit was opened after 18 months on the mast, the interior was found to be completely dry and pristine; although it should be noted that during the period of use the 12V supply was always left on, so that the 9W or so of static power disippation would help to prevent internal condensation.

Were I to build this installation again, I would use a W2DU ferrite bead balun (see references below), installing small ferrite rings on the composite cable supplied with the tuner. A W1JR-type toroidal cored balun is probably a better choice when using thin coax, but the cable in this case is too thick to be wound around a small toroid, and a large core, having a longer magnetic path, may not provide sufficient choking inductance on 160m. The change to a W2DU balun would reduce the length and losses of the feed line, reduce the weight and windage on the mast, and improve the aesthetics of the installation.

Calculated losses for the 40m of RG213 feedline were 1.03dB at 14 MHz for 1:1 VSWR. For 2:1 VSWR at 14MHz, feedline losses increase to about 1.5dB. Don't use long lengths of RG58, especially the 'cheapernet' varieties with tinned conductors.

In many installations, the tuner and dipole may work perfectly well with no balun whatsoever. In this case, there may be some radiation from the feed line, but this is largely irrelevant in the absence of EMC implications. The balun however, guarantees good behaviour and makes the success of a first-time installation more likely.

D.W.Knight G3YNH, Original Nov. 2000, last updated May. 2003.

References and Further Reading.

The ARRL Antenna Book, 19th Edn 2000, ISBN 0-87259-804-7.
1) Input impedance of centre-fed dipoles, ch 2, p2-6
2) KI6WX Directional Wattmeter: "The Tandem Match" ch 27, pp9-19.
3) Ferrite bead balun: "The W2DU Balun" ch 26, p21-23.

"Building and Using Baluns and Ununs", Jerry Sevick, W2FMI. CQ Communications 1994. ISBN 0-943016-09-6.
Chapter 1 on the 1:1 Balun explodes many myths and is therefore essential reading. Book is available from Amidon Associates:.

Useful Links:
SGC Web Site.

Home | Radio Index page |