After learning about software defined radios, I wanted to build one. I have always enjoyed modular designed systems. They are very good for prototyping and well built electronics. I can easily replace a "module" of a circuit to compare and test the radios performance. Modular design also ensures as minimal interaction between circuits as possible. RF sealed enclosures for each circuit help avoid problems with stray signals and "birdies" as well as other problems with signals being where they shouldn't.
This is the side wall that holds the connectors for the +12V and LO Out. The walls measure 1" tall to provide enough clearance for the local oscillator circuit. The SMA connector is one of many I pulled of some old microwave radio modules (Digital Microwave Corporation) that I acquired at one of my former jobs. It provides a small yet convenient connection that isolates the RF and allows me to feed it into another module and have minimal interference with any other portion or the radio.
Front and back view of the assembled connector PC board wall. Notice the lead I attached to the back of the SMA connector, this allowed me to attach the output of the Local Oscillator and any future modifications easier. I drilled the SMA connector to mount on the inside, however this made the hole a little too big to be covered by the connector. I made note to drill the hole smaller next time and mount the SMA on the outside. At 28 MHz the small holes will probably not have a large effect or even noticeable due to the skin effect of RF!
The completed Local Oscillator without the cover. Notice that I used spare RG-174 coax shield as a gasket to seal the top cover to RF. After soldering the braid to the edge of the board. I realized that the best way was to ONLY tack the braid on the corners. I tried to solder the braid along the outside edge (with it resting on the top edge as a seal) and it ended up wicking the solder into the whole braid making it stiff and not very effective. Next time I will only apply solder to the four corners. The circuit is built "dead bug" style which is very common for prototyping in amateur radio. It helps make quick experiments with different layouts and parts easier.
After extensive testing, the 2n3904 transistors were NOT suitable for the oscillator. After much debate and help finding cheap but suitable replacements I have settled on MPS5159 transistors which have an FT=2000MHz! I should get a lot more gain from each transistor with the replacements. I have yet to build the new circuit however I will when I get the time. The old circuit was removed from the RF box and what remains is a nice copper clad PC board enclosure.
I had a long conversation on the www.qrz.com forums with W6YK who helped design a new oscillator with suitable RF amplifiers. He also suggested the MPS5159 transistors. His schematic and computer simulations are shown below. I hope to build a oscillator based on this and make adjustments where needed to fit the project goals!
I constructed the divide by four counter Johnson Counter "dead bug" style as well. It will take the 28.238MHz local oscillator signal and convert it into 14.119MHz (freq/2) and 7.0595 MHz (freq/4). This will allow me to create the I and Q signals needed for quadrature demodulation.