Multiple-quantum magic-angle-spinning (MQMAS NMR) spectroscopy has become a routine method to obtain high-resolution spectra of quadrupolar nuclei. One of the main problems in the performance of this experiment has been the poor efficiency of the radio-frequency pulses used in converting multiple-quantum coherences to the observable single-quantum signals. As the MQMAS experiment is basically an echo experiment this problem can be related to the efficiency with which continuous wave pulses can normally achieve the multiple- to single-quantum conversion for different crystallites in a spinning powdered sample. In this paper we investigate various aspects involved in this multiple-to-single quantum conversion, in the hope to facilitate the devise of new experimental schemes that can lead to significant MQMAS signal enhancements. We examine in particular a recently suggested experiment for MQMAS spectroscopy which employs amplitude-modulated radio-frequency pulses, and which can yield substantial signal and even resolution enhancements over the commonly used pulse schemes in MQMAS experiments. The mechanisms of operation of continuous-wave and of amplitude-modulated pulses as applied to the selective manipulation of spin-3/2 coherence elements are examined in detail, with the aid of the fictitious spin-1/2 formalism in combination with quadrupolar adiabaticity arguments. New insight into the nature of the MQMAS experiment is thus revealed, and the superior performance of suitable amplitude modulations toward the formation of MQMAS powder echoes is justified. Experimental results highlighting the utility of this scheme in samples possessing multiple quadrupolar sites with varying quadrupolar anisotropies and chemical shift offsets are demonstrated, as is the relative insensitivity of the new signal-enhancement technique to the actual level of rf irradiation. Further implications and uses of this new irradiation scheme are also briefly discussed.