01/26/2015 Lately, I've been cultivating and refining my knowledge of radio frequencies; the great "black art" of electronics. I recently bought a pack of subminiature pentodes, and wanted to put them to work. They're quite nice devices: few transistors can match the minuscule reverse-transfer coefficient of a pentode (typically < 0.1pF grid to plate capacitance), which allows for simple operation: high gain and no neutralization required for operation up to the HF range. While the noise figure is worse (due to full shot noise of the screen grid current), the gain remains high. They can very much be likened to a JFET or MOSFET, with extra shielding (dual gate MOSFETs come the closest in many characteristics), and merely ten times higher operating voltage. Indeed, the 5702 subminiature used in this project has transconductance comparable to the average RF JFET (5mS, or should I say, mho to keep with the historical setting!).
Click to enlarge:
Overview of the breadboard. Construction is primarily bare PCB stock, solder, and copper wire.
RF front end. Antenna connects to the BNC connector (soldered edge-launch style to the PCB), is stepped up by the coil, and amplified by the first tube. The local oscillator is mounted above, in its own shielded box.
The plate of the first tube is loaded by an inductor, which resonates at the broad RF frequency range. This is coupled directly to the grid of the mixer, also a 5702.
You can see the prototyping method used for this build: the tubes were first mounted to vertical PCB strips, which were drilled on 0.1" centers to accept the pins. Copper was removed from the PCB strips in strategic locations, giving double sided copper to mount to the base, and single sided copper pads for the electrical connections. The height above the ground plane minimizes stray capacitance, an important consideration in tube circuitry (which generally operates with impedances of kΩ).
The mixer (right compartment) is driven by the RF amplifier (rightmost tube). At the same time, the local oscillator (LO) drive comes in via coaxial cable, filtered by a parallel tank (the red coil and an attached capacitor) before driving the cathode.
Notice the smoothed edges on the cut pieces of PCB. Panels were cut nearly to size with shears, and usually sanded to final shape using wet-or-dry sandpaper. Most solderable surfaces (like the main base plate) were cleaned by sanding before assembly.
A diversity of coil constructions were tried and used in this project. All are air cored. On the right, a low-value coil is self-supporting (and mechanically tunable, hence the deformed shape). Behind it, a narrow nylon tube serves as former; it is supported on stiff wire leads clinched into holes in the former. Left, several coils were wound on thin wall teflon tubing, which is an excellent dielectric as well as mechanical support (it does not melt under soldering heat). The windings on these were cemented in place with superglue. The coils in the filter section (not shown) were constructed on paper tube, cemented with superglue, and epoxied to strips of etched PCB stock.
A highly selective IF bandpass was desired, so crystals were used. Unfortunately, having only two equal parts on hand limits the sharpness somewhat. Typical examples from other amateur designs use five or more crystals. However, the result is still excellent, dropping by 40dB over several kHz on the edges. Two tuned solenoid coils bookend the filter, also providing impedance matching between the plate (IF-2 amplifier) and grid (IF-3 amplifier).
The filter was designed by iteration of a typical design, using SPICE simulation to evaluate performance. No filter synthesis tools were found to be helpful. The major drawback: these crystals have many spurious modes, ranging from just a few kHz to 100s of kHz above the fundamental resonance mode, leading to many peaks in the IF bandpass that are undesirable. It seems this cannot be avoided, unfortunately, except perhaps by staggering crystal frequency and cut.
Final IF (IF-2), tuning coil, and detector. This coil is wound on teflon tube, cemented with superglue, and supported on etched PCB stock: wedged in place rather than bonded, since... what's the point of using glue against teflon?
The detector is a 6AL5 (left). Unfortunately, I don't have any subminiature signal diodes, so unless I want to give in and use solid state here, a miniature will have to do, as out-of-place as it is. The AF output coupling and AVC time constant are also set here (bottom left). Finally, the power wires (100V "B+" and 6.3V heater) enter here (front and center).
Top view. The miniature socket is mounted, with nuts and bolts, to holes in a pair of vertical risers. These are reinforced with right angle stiffeners, to prevent bending stress during insertion and removal of the tube.
Rear view of the amplifier and detector.