A 14 MHz VCO using a series LC resonator.
(Nov 8 2009)
The main article is here

This oscillator is a design for studies of oscillator sideband noise. The design is an LC filter that is matched to 50 ohms followed by an amplifier with a direct 50 ohm output. The direct output is connected to a buffer amplifier which is connected to a second 50 ohm output which id filtered to become overtone-free.

In the above picture the input is connected to the direct 50 ohm output. In this configuration the unit provides +11.7 dBm in a 50 ohm load.

One of the purposes of this unit is to serve as a reference to compare with an identical oscillator that has a common design error in that the capacitance diode is supplied with the tuning voltage through a 1 megohm resistor rather than through an inductor. This "design error" oscillator has high phase noise because the thermal noise of the resistor and the noisy voltage drop due to the diode leakage current produces FM modulation.

This is the view from below with the lid off. Birds nest style so the oscillator is a little more microphonic than a "proper" construction.

Without the feedback coax the unit is simply a selective amplifier. The input transformer matches the 2.7 ohm resistor to 50 ohms. The 220 pF capacitor balances the reactive component.

The resonator is the coil from the 2.7 ohm resistor which forms a series resonance with the capacitance of the four capacitance diodes. (Actually they are 24V zener diodes.) The inductance of the coil is in the order of 15 uH while the diodes form a capacitance of about 10 pF. The reactance of both is about 1 k ohm at 14 MHz.

if both the capacitor (diodes) and inductor were loss-less the impedance at resonance would be about 30 ohms (2.7 ohms plus the input impedance of the grounded base transistor. A J310) for a loaded Q of about 300. The measured Q is about 47 as deduced from the 3 dB bandwidth when (weak) wideband pulses are sent into the input and measured at the direct output as can be seen here:

The total span is 2 MHz and the 3 dB points are separated by about 300 kHz at 14.37 MHz. The unloaded to loaded Q ratio is thus about 6 which means that the performance should be dictated by the losses in the capacitor and coil.

The noise generated by the amplifier is maximum at the resonance frequency where the transistor looks into an impedance in the order of 250 ohms. At the 3 dB points, 150 kHz away the impedance would be 1.414 higher and that provides feedback that reduces the noise produced by the transistor itself. By connecting a dummy load to the input and measuring the noise at the direct output one can see the noise floor which will contribute equal amounts of AM and FM noise to the sideband noise when the cable is connected and the oscillator is running.

The AM and FM sideband noise 0.5 MHz away from the carrier is about 10 dB lower than it would have been if there had been a resistor to ground at the source of input transistor. Normally one would not design LC oscillators with a series resonator like the one described here, so resistor or no resistor is purely academic. For crystal oscillators it can be however important.

The phase noise at 10 kHz frequency separation is -143 dBc/Hz as can be seen in the main article of this investigation.

The +12V supply is carefully decoupled like this:

To SM 5 BSZ Main Page