Kellyville is maintained and run by Stanford Research Institute. The 32 m dish located there is normally used for incoherent scatter radar studies of the ionosphere. It has an estimated gain of 49 dBi, equal to a beamwidth of approximately 0.5 degs. All transmitter equipment is made for low duty cycle pulse transmission and not suited for amateur radio service. On the evening of our arrival we went to Kellyville to look at the installation and talked to the site manager, John, to get an overview of how we should hook up to the system. We wanted to use the system with as few changes as possible to quickly get on the air. Due to limited pre-information on the system and a note from Gudmund, SM2BYA, regarding the polarisation, we were very concerned about how it would work out. The transmitter for the ionosphere scientific studies is a 3 MW, 1291 MHz pulse klystron amplifier connected to a waveguide that runs some 100 m to the feedpoint/polariser. We gained access to the waveguide at the klystron output and inserted a coax-waveguide transition for our TX-signal. From a distance the 32 m dish looks nothing special, but standing on the 18 m high pedestal gives a whole new perspective of its actual size – it is BIG. The RX-port of the polariser ends in an insulated cylindrical 'chamber' between the 2 elevation gears. (The chamber is big enough for 4 people!). The RX-port is a waveguide-coax transition, so all we had to do was mount our pre-amp and optionally use a relay to provide additional TX isolation. We measured the isolation between TX and RX-port on the polariser to approximately 30 dB, so we needed the isolation relay. For RX we used the existing 1290 MHz-28 MHz down converter at the feed point. The LO frequency is controlled from a signal generator inside the building. The LO frequency is divided by 3. In normal operation, the LO is on the high side, so the signal is inverted. We programmed the LO to place 1296 at 28 MHz with the LO on the low side. We used an FT-1000MP. Its dual receiver capability allowed us to monitor for others while simultaneously working calling stations. The 28 MHz IF signal were carried to the building by existing RG-214 cable.

The following morning we returned with all our equipment, including 2 complete setups in case something failed or had become damaged or lost. The connection to the TX-port was only a few meters from our operation position in the transmitter/workshop building, and a short length of 1/2" LDF was used. The pre-amp and isolation relay was mounted and the relay was controlled from our sequencer via the existing control cables. After a final check to ensure that the relay and pre-amp was working, it was finally time to test the TX-line and listen for our own echoes. The tracking of the moon was done by computer controlled motors. We were thankfully not required to turn the dish manually. We performed a TX and SWR check, then aimed the dish at the moon. All of us were anxious to see if our system was working. We transmitted some dashes, switched to receive and instantly heard our own echoes with S5-7, YES it worked. It was just in time for the Eur moonset. We immediately called CQ and soon the 1st station was in the log. We were very relieved and happy to confirm that the dish polarisation was in accordance with the amateur convention! There were some minor oscillations in the auto-tracking system, which made the dish scan the moon with +/- a few tenths of a degree, enough to cause some QSB. Normally the dish is used to track fixed objects/positions in the sky. Tracking the moon meant that the computer had to do some extra calculations, which took more time, causing the system to become a little unstable. This problem was fixed by John the next day. We'd always had good strong echoes, but in a QSO with Stig, OZ4MM, he told us that our signal had changed dramatically from the time at our moonrise and some 5 hours later. We now realised that there was an offset problem, which John confirmed and he told us that they had an earlier suspicion to an offset error. We could easily enter offset corrections into the tracking computer, but it was an educated guess as to where to start. The dish would frustratingly return to the park position (Az=0°, El=90°) to begin a new event every time we changed something. After some time (days!) we had the offset routine optimised. The best way was to turn the dish to a point where the echoes would just disappear on either side of the moon for both azimuth and elevation and then simply aim for the centre point in both axes. We could hear our own echoes with 1W, but could not easily reduce the power further to find the limit. The moon-noise was quite evident, which is uncommon on 23 cm. In our preparation we had calculated to be able to work stations having 20 dBi antenna gain and 100 W, so we were especially alert during Eur moon set and NA moon rise. I don't think we actually worked any non-EME stations, but the best effort in real weak signal stuff was our "QRZ PA3..." to PA3DZL calling and 'testing' with only 10 W and a 2.5 m dish. To keep track of common moon windows, we used the VK3UM EME planner. Our common windows with JA and VK were limited because the dish could not operate/track below 10° el. The limitation was implemented to avoid any inadvertent hazardous radiation when operating at a 3 MW level. Our plan was to use 4 yagi systems on 2 m and 70 cm, which would give us reasonable antenna gain allowing us to work a fair number of stations while keeping the mechanical effort to a minimum. The antennas were placed approximately 30 m apart on opposite sides of the radio shack. The 70 cm antennas (4 x 28 el 432-9WL) were mounted on the "scaffold" and the 2m antennas on a wooden post. Both systems were connected by 1 5/8" LDF for the major part of the TX-line, which limited the loss considerably. Both systems were mounted in a H-frame with an azimuth/elevation rotator, G-5600. It had been our original plan to pre-assemble the H-frames and antennas and to test them thoroughly before everything was shipped to Greenland, however due to lack of time this was not accomplished. Unfortunately, as feared, we encountered problems when mounting the antennas. The G-5600 azimuth/elevation rotator was simply too small for an EME antenna and could not elevate the system without the antennas being fully balanced. On 70 cm we managed to mount all 4 yagis, and with the help of small counter weights consisting of small bags filled with gravel, the system was able to elevate. After several tries on 2 m we had to settle for only 2 yagis, as it was better to be QRV with only 2 yagis than QRT with a broken rotator. Our amplifiers were a 4CX1500 on 2 m and a 2 x 3CX800 on 70 cm. Both were build by Michael, OZ2ELA/OX3LG. By Tuesday evening everything was ready, with all the equipment installed and the antennas beam heading calibrated. Sun noise was quite strong and we could even hear thermal noise from the ground so we were keen to get started, but of course we had to wait for moonrise on Wednesday morning. Finally we had moon rise, but for the 1st 20 mins the surrounding mountains and inland ice were obstructing the path. Only after 5 degs el was reached were the 1st signals heard. On 70 cm we soon heard our own echoes, but on 2 m it was harder with the smaller system. The 1st QSO on 70 cm with DL9KR was run by Michael, OX3LG, with Bo, OX3LX, eagerly watching. Bo then took over while Michael recovered from being the 1st ever to work a 70 cm QSO outside of Greenland. To avoid any interference between 2 m and 70 cm, we had to synchronise the TX and RX periods. We chose to use 2 min periods normally used on 2 m. All skeds were 30 min where OX2K would transmit in the 1st period. It was quickly evident that this was not a good solution however, as instead of 2 min periods, suddenly a 4 min period was present at every full and half-hour! This produced great confusion. We used split frequency operation transmitting on 144.085/432.085, in sked listing on 090 and random 080-085. The majority of people calling us on random used 085, which caused a lot of QRM, especially when the timing sequence went wrong. It took some time before the systems were running correctly and on Wednesday we finely adjusted the beam heading to make sure we hit the moon. The poor readout on the control units for the Az/El rotators did not make things any easier. Unfortunately one of the control cables for the 2m system was damaged and we lost a couple of skeds on Wednesday before the fault was corrected. On Thursday the input circuit in the 2m PA suddenly failed. This unit was finished literally hours before it was due to be shipped and had been tested for only a limited time with a dummy load. However Michael "the wizard" fixed the input circuit and we were quickly back on the air, without having to setup the spare PA. Unfortunately some innocent HAM's suffered from this incident and they could not understand what had happened. Despite the limited system on 2m we succeeded with many of the skeds and also some random QSOs. The best signals were obtained every morning, when the moon rose over the inland ice with 5-15 degrees elevation. Ground gain must have been a factor with the gently sloping ground for the 1st few km to the east. A good tool to find stations and aim antennas was a DSP FFT program (FFTDSP 4.2) where we could "see" even very weak signals in the LF bandpass. It was the first time any of us had seriously used such a program and it will definitely not be the last!

When all the OX2K stations were up and running we activated 2 HF setups, a 50 MHz setup, a 2 m setup, a 70 cm setup and the 23 cm at Kellyville. Consequently due to lack of operators, it was difficult to keep all the stations running, which meant that some EME-operators had to do 20 hours shifts. This was far from ideal, especially for the operators, but the result was reasonably good. Sometimes the 2 m EME operator was assisted by a non-CW/EME operator to press a button on the memory-keyer or to "look" for signals using the FFTDSP program. In retrospect it is fair to say that the expedition was a huge success. The results were on 2 m were 36 QSOs, 36 initials, 15 dxcc (12 OX firsts) and 3 continents, on 70 cm: 39 QSOs, 35 initials, 15 dxcc (all firsts) and 4 continents, and on 23 cm 113 QSOs (22 on SSB), 76 initials, 20 dxcc (all firsts) and WAC. We worked HB9Q, SM2CEW and WA4NJP on all 3 bands! Given the remote location, the limited amount of equipment and the fact that we managed to work EME on 3 bands simultaneously, the result has to be rated as VERY good. For all of us this was probably the greatest experience we've ever had as radio amateurs, and most of us are already thinking where we should go next.
For more info see the OX2K homepage:

and general information on Greenland: