With regard to Chris Carson's device: This is a Parametric Amplifier. There is no input signal. A small initial charge on the capacitor is enough to make it begin oscillating. In a Parametric Amplifier, a signal is amplified by varying the capacitance with respect to time. This is usually done with a Varactor Diode. It can also be done by mechanically rotating the plates of an air variable capacitor, as is done in Chris Carson's device. However, mechanical rotation limits the practical frequency at which the device can operate. Like any amplifier, a Parametric Amp can be thrown into self-oscillation. When that happens, no input signal is necessarry. The thing will oscillate on its own. The variable capacitor should be wired in parallel with an LC tank circuit. For amplification (or oscillation) to occur, the variable capacitor should be varied at twice the resonant frequency of the tank circuit, taking into account the mean value of the variable cap when calculating the frequency. The correct rate at which to vary the capacitance is as follows: 1. Measure the value of the variable capacitor when it is fully open, and when it is fully closed. Call these C_min and C_max. Then calculate: C_mean = ( C_max - C_min ) / 2 The total capacitance in the system, for calculating frequency, is equal to: C_total = C1 + C_mean Where C1 is the tank capacitor. Then, the resonant frequency of the system is : f_tank = 1 / 2 * pi * sqrt(LC_total) The correct frequency at which to rotate the capacitor plates is: f_cap = 2 * f_tank The variable capacitor must be rotated at TWICE the resonant frequency of the system. This will cause current oscillations to grow without bound. Also significant, is what Mandelshtam and Papaleksi called the "modulation depth". This is the ratio of the variable capacitor to the tank capacitor. The modulation depth must be sufficiently high (but not too high) to begin oscillations. * * * Here are some LTSpice simulations I created, of simple Parametric Amplifiers thrown into oscillation. There is no input signal. A small initial charge on the capacitor of 1mV, or a small initial current of 1mA, is enough to create oscillating currents which grow without bound. Single Tank amplifier: https://drive.google.com/file/d/1v4IsPs1xRc6bK7n3AcGftZwrHesHwNDJ/view?usp=sharing The variable capacitor must be varied at twice the resonant frequency of the tank. f_tank = 1 / 2*pi * sqrt(LC) f_cap = 2 * f_tank When calculating the resonant frequency of the tank, and thus the appropriate rate of capacitance variation, it is important to take into considerating the mean value of capacitance. For example: In a tank circuit composed of a 2500pF capacitor in paralell with an inductance L, and a time-varying capacitance C2: If C2 varies from 0 - 1000pF, its mean value is 500pF The resonant frequency of the tank is then 2500pF + 500pF = 3000pF. 3000pF is the correct value of capacitance to be used when calculating the resonant frequency of the tank: If you calculate the frequency using 2500pF, no oscillations will occur. Two Tank Parametric Amp: https://drive.google.com/file/d/14CQ8e61SKObJA_75nLdL-rJkovdYSAKF/view?usp=sharing A simple "straight through" Parametric Amplifier with no input signal. In this configuration, the variable cap is varied at a rate equal to the frequency of the signal tank + the idler tank. One thing I noticed with the Para Amp is that the value of the variable capacitor is critical. In this example, a variation of 0 to 360pF causes amplification. Less than that, and the oscillations die out. For example, if it is 0 to 340pF, no oscillation occurs. Another thing, depending on the two frequencies involved, there can be a condition where the idler tank is amplified 1000 times more than the signal tank. I have read that the relative Q of the two tanks is a significant parameter, but have not experimented with this. The phenomenon can also be created by varying L instead of C. That was the principle upon which Mandelshtam and Papaleksi's device operated. Parametric Oscillator using Time-Varying Inductance In Series: https://drive.google.com/file/d/1jc8Kn1saAVttMECz-EsVzgJaYQEN8xMX/view?usp=sharing In Parallel: https://drive.google.com/file/d/1Xtmo7DG1-SMpnjwiG_Xmh79-ScztQD04/view?usp=sharing The inductor can be connected either in series with the tank inductor, or in parallel with the tank. The relative values of L, C, and inductance variation rate are slightly off in the parallel simulation, but the effect occurs anyway. The values are more correct in the series simulation, and the current grows quickly from 1mA to many billions of amps within 200 ms. References: On the Parametric Excitation of Electric Oscillations L. I. Mandelstam, N. D. Papaleksi 1934 https://drive.google.com/file/d/1ZeGq3LrQNXExyTtbZ-1nhuUcyqDVh2Lv/view?usp=sharing Parametric Amplifier Design William J. Robertson, B.E.E. Ohio State University 1959 https://etd.ohiolink.edu/!etd.send_file?accession=osu1400145234&disposition=inline Modeling time-varying storage components in PSpice Dalibor Biolek, Zdenek Kolka, Viera Biolkova Dept. of EE, FMT, University of Defence Brno, Czech Republic Dept. of Microelectronics/Radioelectronics, FEEC, Brno University of Technology, Czech Republic https://pdfs.semanticscholar.org/eae5/59dc31302cf1fd51967b0f6fc83aa9467dee.pdf * * *