G4BAO

The idea came from musings on the success of the 10 GHz beacon, in emails between Bernie G4HJW, Sam G4DDK and me in August 2007. I already had a 12 GHz multiplier, and Chris GW4DGU wanted to try his ultra-stable “DISTAW” oscillator. From concept to on air took 13 months. GPS locking in May 2010 made it, as far as I know, the first 24 GHz beacon in the world to run JT4G.

• Even minutes: “GB3CAM JO02BI” in JT4G, then “GB3CAM” in Morse.
• Odd minutes: unmodulated carrier until 32 seconds past, then “GB3CAM” in Morse.
• JT4 tones: 24048.869798, .870113, .870428 and .870743 MHz. CW key-up tone is 24048.8696 MHz.
• Site: Haddenham Cambridgeshire, JO02BI. Horizontally polarised 10 dBi slotted waveguide at 10m AGL, in a clover-leaf pattern.

The GPSDO and reference controller

Site: Invicta House, Margate Kent, JO01QJ81.

Antenna: Horizontally polarised 15 dBi 90 degree sector antenna at 80m AGL, pointed at Rotterdam.

Here is a link to the full description of the WebSDR

https://g4bao.com/Kent WebSDR_1.pdf

Indoor Unit Block Diagram

The Indoor unit

The Outdoor unit

Two Ionica base-station PAs, joined with hybrid combiners made by cutting the surplus 3-way combiners down to 2-way. Three layers of Kapton tape on the hybrid pull it from 3600 down to 3400 MHz, and the duplexers retune as 3400 MHz band-pass filters.

• 2 mW in gives 19 W from each PA and 34 W combined, drawing 3.2 A at 48 V (154 W).
• The supply is the Ionica’s six Nemic-Lambda PH modules: 9.5 V modules for the PAs, run from an unregulated 35 to 50 V at about 4 A.
• Semi-rigid SMA leads must be exactly the same length either side of the amplifiers, to keep the phase relationship.
• Hybrid tests at 3400 MHz: input return loss 26 to 27 dB, insertion loss about 3.2 to 3.3 dB.

The Ionica base station has a single PCB with three complete double-conversion transmit strips, once seen at rallies for £1. I cut out the main upconverter and used it with an Ionica receive converter.

• Oscillator: 90.444444 MHz crystal ×9 to 814 MHz, then ×3 to 2442 MHz.
• Replace the first IF filter with a 958 MHz helical or SAW filter. The Ionica “pizza box” has an identical one.
• The PIN diode attenuators need 10 V to forward bias them, or remove the diodes and bridge the tracks.
• Solder vertical screens along the gold borders, or 3256 MHz (2442 + 814) leaks into the output. Check it on a spectrum analyser before going on air.

Surplus Andrew cellular PA modules, rebuilt with copper-foil matching lines worked out in Appcad and a Smith chart program, since there is no impedance data above 960 MHz. For the MRF9060 module I reduced it to a single device.

• Results: about 70 W from the MRF9060 at 5.3 A and 35 W from the MRF9045 at 3.3 A.
• Bias: start at zero gate volts, then set the quiescent drain current to 350 mA (9045) or 450 mA (9060).
• Always disconnect the drain and gate supplies before trimming the foil, and trim no more than half a millimetre at a time.