Prototype Time-Sharing
Multi-Band Beacon  
Ray Cracknell, G2AHU
RSGB Propagation Studies Committee
Issue 65, May 2000

The Aim

The aim of the project is to produce a cost-effective, easily duplicated, beacon transmitter to work on the 24, 28, 50 & 70 MHz amateur bands, with 40MHz and 144MHz as possible alternatives. It is intended ultimately for the beacons to be controlled in frequency and timing by Global Positioning Satellites (GPS).

The motivation for the project arose from the success of the International Beacon Project with the North California time-sharing beacon chain on 14100, 18110, 21150, 24930 and 28200 kHz and the allocation of 28MHz frequencies for time-sharing beacons in Region 1 by the International Amateur Radio Union. The overcrowding of beacons on the 28 MHz and 50 MHz beacon bands is now evident and pressures to reduce the segments allocated to beacons is very considerable. For a long while the RSGB Propagation Studies Committee has been keen to have co-sited beacons which ties into the idea of time-sharing. Further any time-sharing will require synchronisation and it seems better to use the equipment required on more than a single channel.

At this stage of the project the frequencies and timing have to be controlled by crystal oscillators and it is hoped that work on GPS control will commence this year (2000). We need short-term stability for DSP monitoring and this makes phase locking unsuitable.

The transmitter is housed in an aluminium case 32cm wide, 8cm deep and 42cm tall. A mains unit providing 15 volts for the driver stage and 12 volts regulated for the exciters, timer and keyer stages, and a variable voltage regulator for the nominal 24 volts for the PA are housed in the base. Space for a low capacity lead-acid battery is available for a 12 volt trickle-charged supply for the timer, but the supply for the PA must be housed externally. The transmitter is built on removable shelves consisting of:

a. Timer and keyer

b. Exciter stages

c. Driver stage

d. PA and filter stages on the top shelf.

The timer is built into a 130 x 75 x 45 mm box made of circuit-board with a 15 pF variable capacity for tuning and a Reset/Run switch for setting up and adjusting the timing. For obvious reasons these controls are only available by removing the back of the transmitter case. The timer was designed to provide a marker every 20 seconds to start the keyer accurately for each transmission and to provide outputs to switch the relays in sequence to energise the required exciter and change the RF output from the exciters to the driver stage.

Part of the first IC, an HEF4060, is a 4.194304 MHz crystal oscillator and the remainder together with the second IC, a TC4020, consists of binary counters to produce pulses at 2 per second.

This is followed by two CMOS counters. The first, an MC14017, divides by 10 which provides one output to the second CMOS counter and another to the DM74121, which supplies the pulses to reset the keyer every 20 seconds. The HEF40l7 distributes 20 second pulses to the first 4 outputs and output 6 is used to reset itself every 2 minutes. The 1 to 4 outputs are each followed by a 2N2222 transistor used as a switch to work their respective relays. This arrangement provides a two-minute cycle but changing to a three-minute cycle, if and when required, would be simply done by resetting from output 6 to output 9 on the final 4017.

The keyer is built on a 175 x 115 mm printed circuit board situated on the same shelf as the timer. It was kindly supplied by Lawrence Woolf, GJ3RAX, and is a matrix type designed by him for the 50 MHz beacons which were distributed to DX stations by the UK Six Metre Group. The 16 by 16 matrix is scanned by 74150 and 74159 ICs, each controlled by a 7493 IC. Reset facilities are actioned by the output pulses from the 74121 in the timer; since operation of the keyer requires earthing the reset facility, a 10 micro-Henry choke is necessary across the input terminals. Keying of the exciters is actioned by a transistor switch on the keyer board, and a front panel headphone jack is designed for setting up and testing the keyer.

The message sent consists of approximately 10 seconds plain carrier, followed by the station callsign and then ‘ar’ with a one-second delay to allow for relay lag. The message can be altered by changing the diodes on the matrix, carefully observing the polarisation. One diode is required for a dot and three for a dash. The space between dots and dashes in a symbol is one space and the space between symbols is three spaces and there are four spaces between words. A long 10-second dash is obtained by a whole line of diodes.

Each exciter is built as a separate unit to facilitate replacement with alternative frequency units. The present units were the subject of considerable experimentation aimed at frequency stability and harmonic suppression. Units are built in double-sided circuit-board boxes 135 x 50 x 40mm. Construction of all of them is very similar and consists of a 2N3819 crystal oscillator followed by another 2N3819 source follower both supplied with stabilised 9 volts. These are followed by a 2N2222 amplifier where keying takes place and it feeds balanced push-push doublers or push-pull triplers, which in turn feed a 2N3053 amplifier whose input and output are tuned to the transmitted output frequency.

The 24, 28 and 50 MHz exciters require 12465.25, 14095.5 and 25025 kHz crystals followed by doublers, but the 70 MHz exciter requires a 23350 kHz crystal followed by a tripler. A 144 MHz unit has not yet been built but a 24 MHz crystal with the 3rd harmonic extracted in the drain, rather than an overtone crystal, would be employed.

Four relays are mounted in the middle of the exciter shelf in order to leave the wiring available from the back during testing, which will be convenient but is untidy and will be altered when the final units are produced.

The driver stage uses an UMIL-5 followed by a D1001. Both are isofets and the circuits are built separately on 100 x 50 x 15 mm heat sinks. The UMIL-5 is capable of 5 watts output up to 400 MHz with 28 volts applied and the D1001 is capable of 18 watts output with 28 volts up to 200 MHz. Both are used as broadband amplifiers and give ample drive to the final with only 15 volts applied. All the isofets, including the PA, were donated by Mr Steve Thompson G8GSQ; we are very grateful to him also for his expert advice. In beacon design, under-running components is always advisable. With a shelf available for them, the driver isofets are widely separated and well screened from the PA and exciters.

In spite of this driver being satisfactory, an alternative of using a pair of isofets in push-pull was being considered. The advantage of push-pull would be to restrict the generation of even-order harmonics. However the unit was found to have insufficient power gain and larger isofets are required.

The wideband power amplifier uses a pair of D1005 isofets in push-pull. It is constructed on a 150 x 100 mm circuit board. The isofets must be a matched pair and balancing the output is of prime importance so the bias is stabilised and fed through individual potentiometers to each isofet. The transformers are all made using RG316 co-ax and two twin-hole ferrite cores for each transformer (Siemens B62152A1x1).

The amplifier is mounted directly on a 150 x100 x 4O mm heat sink with the flanges of the heat sink outwards through the back for ventilation.

The PA is followed by three low pass seven-element filters. The prototype uses one filter for both 24 and 28 MHz and separate filters for 50 and 70 MHz. Jeremy Whitfield, G3IMW built the filters and they have now been installed. The RF output circuit will be designed either for a single 50 ohm co-ax lead to the antennas, or separate outputs for each band as required. Input and output from the filters is controlled by relays activated by the timing system.

Individual recipients of beacons will be free to design their own antennas. Jeremy Whitfield, G3IMW is responsible for antennas; several experiments have been conducted by him and advice will be freely available, but the decision must remain with the group responsible for the beacon.

Trap vertical half-wave antennas will probably be the most popular choice but will need to be specially designed to incorporate 70 MHz. Omni-directional would be desirable in central locations but any located on the fringes of Region 1 may care to use directivity towards the centre. Polarisation is sometimes important but does not affect propagation via the ionosphere but antennas do affect ERP at low angles. JE7YNQ in Fukushima, Japan has shown itself the best of the beacons from Japan during March this year and uses a turnstile antenna on 50MHz and stacked dipoles on 28MHz. This tends to confirm our findings from GB3BUX and GB3NHQ that turnstile or similar low-angle antennas are better omnidirectional DX antennas than verticals, and serious consideration should be given to them where possible.

As mentioned above, the recipient will be required to provide the mains supply, which we assume will probably be 230V 50 cycle AC. Thus a 28 volt mains unit will be necessary. If batteries are used then they will be a safeguard against relatively short periods of mains failure and will cope with the wide variations in current requirements in the transmitters. Although the voltage stabiliser will control the voltage without batteries, their use will result in use of a 10 amp supply being sufficient but without them a larger unit may be required. However the cost of two 12 volt accumulators may outweigh the difference, and in that case a small rechargeable 12 volt battery across the supply to the timer is advised, so that it will continue to keep time in the event of mains failure.

Most of the experimental work has now been completed and the time has come to enlarge the beacon development group and to commence the work on controlling the accuracy of both the timing and the frequencies of transmission using GPS.

To return to the archive page click here