An investigation has been made of the dependence of the length of the ultrashort electromagnetic waves generated by vacuum tubes on the magnitude of the heating current. It is shown that the normal waves (i.e., waves the frequency of which corresponds to the frequency of the oscillations of the electrons about the grid of the tube) differ from the dwarf waves (i.e., waves with frequency ceeding several times the frequency of electronic oscillations) in their dependence on the heating current. The length of the normal waves increases, while the length of the dwarf waves of all orders decreases as the heating current is decreased. Therefore, as the heating current is decreased, there is a better agreement between the observed and computed values of wave-lengths of the normal and of the dwarf waves. At low heating currents the product ${\ensuremath{\lambda}}^{2}{E}_{g}$ for dwarf waves of different orders is very close to the theoretical values ${\ensuremath{\lambda}}^{2}{E}_{g}=\frac{{\mathrm{const}.}_{0}}{{n}^{2}} n=2, 3, 4, \ensuremath{\cdots}$ where ${\mathrm{const}.}_{\mathrm{o}}$ corresponds to the value of the product ${\ensuremath{\lambda}}^{2}{E}_{g}$ for normal waves. The results of the experiments are in agreement with the theory of P. S. Epstein, which explains the discrepancy between the calculated and the observed length of the normal and the dwarf waves being due to the influence of the alternating potentials appearing on the electrodes of the tube during oscillations. In the limit, at infinitely low amplitudes of these potentials, the ratio of the values of the products ${\ensuremath{\lambda}}^{2}{E}_{g}$ for normal and for dwarf waves approaches the theoretical value within the limits of the experimental error. Neglecting these alternating potentials introduces the largest error in the results of the existing theories of the generation of ultra-short waves. At sufficiently intensive oscillations, this error exceeds all the others taken together, including the error introduced through the simplifying assumption of plane electrodes. The limiting value of ${\mathrm{const}.}_{0}$ corresponding to infinitely low amplitudes of the alternating potentials, approaches very closely the value of the product ${\ensuremath{\lambda}}^{2}{E}_{g}$ calculated from the formulas of H. Barkhausen and A. Scheibe. This shows that normal waves really do correspond to a complete period of oscillations of the electrons about the grid, i.e., to the time the electrons take to pass from the filament to the grid and back to the filament.
Read full abstract