Back in the 2013 I was visiting my friend and near neighbour Briain Wilson, GM8PKL, who has a very good view from his from living-room and patio to the north over the City of Edinburgh and beyond. He told me that on very clear winter days he could see snow covered peaks which he thought were in the Cairngorms which are at a distance of around 140km away. A few weeks later I was visiting Chris Tran GM3WOJ (a well known HF DXer/contester) who lives near Tain in Ross and Cormarty and he happened to remark that from his patio he could see Ben Macdui on a clear day. I began to realise that these summits were probably LoS from both ends of the path between my home location in Edinburgh (IO85JV)and Chris’s location (IO77WS). I suggested that it would be interesting to try a 3cm microwave contact scattering signals over the peaks between the two locations. In the back of my mind was the thought that both locations could be done with more than the usual comfort of portable operation, and as we could probably run about 20 watts at each end of the path it should work. A bit of path profiling confirmed that in fact both ends were in fact LoS to these peaks and it was likely that a signal could be scattered over them on 10GHz.

The GM3WOJ 3cm beacon story html 17686c97

Figure 1. Path between GM3WOJ and Edinburgh

The GM3WOJ 3cm beacon story html 2ad1fbbd

Figure 2. Path profile from Edinburgh to the summit of Ben Macdui.

GM3WOJ beacon story html 2cc26664

Figure 3. Path profile from Ben Macdui to GM3WOJ.

Path profile

Figure 4: Direct path profile from my QTH to GM3WOJ.

At this point Chris suggested that we could do one better and he would be very happy to host a 3cm personal beacon on either his house or one of his towers provided I built it and set it up. He was happy to keep an eye on it and mind it. I jumped at this and said I was very happy to provide the necessary hardware.

Over the winter of 2013/14 I assembled the necessary bits and pieces. The exciter I used one of Andy JNT’s LMX2541 frac-N synthesiser PCBs programmed by a PIC microcontroller to do the keying as per Andy’s recipe. The JT4G keying is timed from GPS. The frequency reference is a well-aged 10 MHz OCXO. It was felt that the extra complications of GPS frequency locking were probably not worth the effort from experience of running the 23 cm beacon GB3EDN

This generated a signal on 3456 MHz which is tripled in frequency (With a bit of recycled Sat TV PCB) to drive a modified QUALCOMM Omnitrak PA (courtesy of Chris Towns, G(M)8BKE). The output is a few hundreds of mW which is fed to a small 35cm BSB dish (remember those?). This gives a beam width of around 7 degrees which from IO77WS gives good coverage of a lot of the UK southwards. The all important housing is from a surplus DMC 22 GHz link unit which had already yielded lots of useful stuff for 24GHz.

The GM3WOJ 3cm beacon story html m2ad15773

Figure 5: Block Diagram of hardware.



Figure 6: ODU hardware.


After an initial trip to Chris,GM3WOJ’s in mid May to reconnoitre and get some measurements to fix the beacon to one of his smaller towers ( at 11m AGL) I got all the hardware together and tested to my satisfaction. This tower was chosen as it had a clear view south and was not obstructed by trees. It all had to be pretty rugged and weather-proof as I did not want to be driving up and down the A9 to effect repairs. In mid June I returned and between us we installed the beacon. Remarkably all went to plan and we had it on the air within that day. Monitoring the signal locally all sounded good but it remained to be seen if it would get past the mountains to the south which incidentally are the highest in the UK second only to Ben Nevis. An interesting problem was ensuring that the dish was actually pointing in the correct direction when the tower was elevated as there was no received signal to pan it on.

The GM3WOJ 3cm beacon story html 79f077b6

Figure 7.The beacon hardware with builder.

Chris tower1

Figure 8: The beacon installed on tower and the take-off south over the Black Isle.

The next day I set off south down the A9 heading home to Edinburgh. I had a Horn antenna fixed to the car as all the best microwave geeks have and monitored it going south. The signal was good on backscatter from the Black Isle but largely disappeared in the vicinity of Inverness. Further down the A9 just before Aviemore where the Cairngorms are in good view from the road it returned and at reasonable strength which was reassuring. I drove up to the Ski slope at Cairngorm and it was not audible LoS as the area is shielded by the direction of the glen. Pointing the antenna at any of the peaks in the vicinity however gave a reassuringly good signal. After that I headed south back on the A9 home with no more signals. It worked with back-scatter but it remained to be seen if forward scatter would work. The next day I was keen to see if it was audible in Edinburgh and I pointed a small (40cm) sat TV dish with a modified LNB as a pre-amp out of my shack window to the north. After a bit of waving it about in the right general direction I was amazed to find the signal was clearly audible at 216 km with an 1800 m obstructed path. The signal was always audible after fixing the antenna to a bracket on the roof.

The GM3WOJ 3cm beacon story html m7d826019

Figure 9: The 40cm receive dish on my roof in Edinburgh (IO85JV). Note the blue sky! It blew away a couple of weeks later.

This arrangement worked well until the winter but in January gales the dish was ripped off the roof and landed in the back garden complete with LNB. Very kindly Briain Wilson GM8PKL offered to host the receiver at his QTH which is LoS to some of the grampian peaks. Figures 10 and 11 show the new dish installation. It uses a re-purposed Sky Dish (painted "gorse green" to camoflage it from curious neighbours) and a modified LNB with an external LO feed. The LO is locked to the 10 MHz output of a Morion DOCXO and comes out on a 432 MHz IF. The receiver has to have excellent frequency stability and the Morion reference DOCXO provides this. It sits within 3-4 Hz of 9936 MHz!. These old Morions are amazing It is used to provide the reference frequency for a Chinese ADF4251 demo board. The ADF4351 is programmed by and Arduino Pro-Mini to give an output on 3312 MHz this is amplified and tripled in frequency to 9936 MHz where it is amplified to a level of around 40 mW. The LO generator is housed in Briain's cellar which is adjacent to the dish and LNB but undercover and protected from the elements. The cables are fed to the dish through and air brick and kept as short as possible to minimise losses. As he is on a high spot with a good view over the City of Edinburgh the weather can get quite taxing at times! Pictures of the LO "indoor unit" and a block diagram are below.

20201222 165036

20201222 205903

LO gen

The LO signal is fed to the LNB via a length of good quality microwave coax but none-the-less the loss is is in the region of 4-6 dB., hence the high LO level inserted into the coax.

The receiver is a FT790R2. This was obtained from Ebay as "Sold for spares" and was a bit of a basket case but I was able to resurrect the Rx side of it and it has proven to be frequency stable and reliable. The audio from the receiver is streamed onto the internet and is at and click on “M3U” if it does not open automatically. The status of the audio stream can be viewed at .

The stream is generated on a Raspberry Pi computer using Darkice and Icecast is the web-server` programme. These programmes are open-source and intended for internet “radio” stations. The audio interface is a USB soundcard.


Figure 9: The “gorse green” dish at GM8PKL's QTH.

20150228 140103 FILEminimizer

Figure 10: Briain, GM8PKL panning the dish.

Viewing the Streaming Audio with Spectran.

The streaming audio can be monitored with Spectran on your PC. To do this you need to install a “virtual audio cable” program to pipe the audio directly from your media player programme output to Spectran. I use the standard Windows media player programme with VB Cable for this purpose. VB Cable is available here at no cost, . Instructions for setting it up are also provided at that URL. To setup Spectran click Setup and then Select soundcard  as shown below. Spectran should see the cable input and output as sound card ports.



Figure 11

Select Cable Output for the Spectran input sound card and Speakers High Definition Audio as shown and click OK. Start Spectran and you should see the signal on the waterfall display. The plot will lag behind the actual signal because of latency on the internet making it difficult to decode with WSJT software (but it can be done!)



Figure 12

Here a period of carrier, followed by a CW ident, more carrier and the start of the JT4G keying cycle can be seen.  As usual the signal is quite dispersed by rainscatter.




Figure 13

Here is a nice Doppler trace from an aircraft. At the time there was a BA flight leaving Inverness Airport shown below.



Figure 14.

Update: May 2017

In January 2016 the beacon was moved form Chris's small tower to a larger one higher up on his ground. The reasons for this were two-fold. Firstly silver birch trees (weeds!) had grown up into the LoS from the dish and also to reduce the weight of the ODU on the tower. Chris had some concerns about the loading on the tower so the beacon was rehoused in an IP65 ABS plastic box as opposed to its original DMC surplus ODU case which weighed about 5 kg. With the forced move the opportunity was taken to rebuild the beacon. This worked well until the weather started to warm up in May 2016 and a slow drift was noted on the frequency and on warm days the signal whould cut off for short periods. These problems were attributed to poor cooling of the electronics in the ABS box and voltage regulators going into shutdown.

As a result of these issues there was a rethink of the whole design of the beacon. It was decided to house the signal generation circuitry at ground level in the original cast aluminium DMC ODU case and feed the signal up a length of coax to a mast-head PA. At the same time it was decided to GPS lock the frequency to eliminate the frequency drift problem. This was relatively easy to do with the surplus Symmetriicom UCCM 10 MHz GPSDO units that have been available until recently on Ebay.The signal is now generated at a level of about 0.5 Watt at the base of the tower and feed up a length of about 20 m of Anderws LDF4-50 coax cable which has a loss of about 10 dB at 10 GHz. That along with a 10 dB attenuator reduces the signal to about 5 mW which is adequate drive for the masthead PA. The PAs are both old Qualcomm Omnitracs units which were obtained from a silent key sale.The output at the masthead is about 1.2 watts. The antenna is still the original BSB 30 cm dish with a beamwidth of about 9 degrees at 10 GHz.



Figure 15: The signal generation circuitry. L to R the GPSDO in diecast box, the PA and the keyer/PIC/LMX synthesiser and x3 multiplier.



Figure 16: The ODU at the tower base with output power monitor.



Figure 17: The dish on the tower. The PA housed in an ABS IP65 box cab be seen below the dish.

Update August 2018

Early in 2018 it was realised that the hardware was unreliable and the beacon had been off the air for considerable periods. I major rethink of the design was carried out and the decision was made to simplify the design to reduce the weight still further, reduce the complexity and also cut the power consumption.With the addition of GPS frequency locking it had become difficult to power the transmitter because of the voltage drop on the buried low voltage power cable. It was decided to remove the requirement for a GPS frequency locking and GPS timing of the keying. In essence we were going to go back to an old style "steam" beacon that just sent its call sign and QRA. The would allow us to house all of the electronics in an 8 inch by 5 inch by 2 inch Eddystone diecast box. Over the winter of 2017/18 the new hardware was constructed. The frequency reference is a Symmetricom OCXO, the exciter is a Chinese ADF5355 PCB modified to improve its phase noise performance and the PA is a narda X band LNA followed by a Qualcom PA to give 1 watt out. The keying is done using an Arduino Pro-Mini over the SPI bus. The software is based on OZ2M Arduino beacon keyer mofified for cw only and to improve the keying control of the ADF5355 by re-initialising it every keying cycle. This was to prevent loss of the output signal if the chip registers became corrupted.

20180210 12321820180210 123034

20180210 123029



Figure 18  Close up views of the hardware in the diecast box.


The current drawn by the beacon when the OCXO is warm is around 1.4 Amps. Internal regulators in the box provide stabilised supplies for the internal modules.

Chris did not want any SMPS regulators in the unit as he was concerned about noise being radiated into his HF antennas (it is a nice quiet HF location!) hence the used of linear regulators throughout. A result of this is that the box dissapates around 22 watts of heat (and radiates 1 watt of RF!) and care had to be taken with the thermal design. To overcome this problem the box was mounted on the tower sandwiched between two over size plates of heavy gauge aliminium for cooling. These can be see in Figure 19.




20180213 104509


20180213 104522


Figure 19. The beacon mounted on the tower showing the cooling/mounting plates.


And finally........

 20160219 094712


Figure 21. Chris beside an earlier iteration of the hardware.



The vital statistics of the beacon are:

Callsign GM3WOJ

Location IO77WS62


Frequency 10368.400 MHz

Reference OCXO (Usually within 100Hz of nominal carrier frequency)

Power 1 W

Antenna 35 cm off-set parabola (Ex BSB dish)

Beamwidth 7 degrees

Beam heading ~165 degrees

JT4G timing GPS

To date the best DX spot is from G4KUX in Co Durham, but with rain-scatter I would expect that it may get further. Up-to-date spots can be viewed at

Postscript: Needless to say we have not had that QSO yet but we still have plans! That’s the trouble with beacons!


Chris Towns G8BKE – Donation of PA

Chris Tran GM3WOJ – The location.


Updated: August 2018