Pyroclastic flows from volcanoes, dust storms in the desert, and submarine turbidity currents are all gravity currents of particles in suspension occurring in nature. Powder snow avalanches are one such flow where the density difference between the suspended particles and the interstitial air is high and the particles carry a significant proportion of the flow's momentum. This means that the Boussinesq approximation, where density differences are considered negligible in inertia terms, is not valid. Aspects of such flows, such as their internal structure and the transition from a dense flow to a suspension current are not well understood. Repeatable laboratory experiments are necessary for a better understanding of the underlying physics. Up to now, an experiment has not been designed with the correct similarity criteria for modeling powder snow avalanches. We have addressed this by designing and analyzing the results from a new experimental model of powder snow avalanches that had three distinctive features. Fine, dry snow was used to form suspension currents in air. The high density ratio of powder snow avalanches was preserved so that the currents are non-Boussinesq. Our experiments started with a dense current that then self-ignited, undergoing the transition to a suspension current. The shape and position of the current was tracked with two video cameras, and the air flow measured with a high-frequency response pressure transducer mounted in the base. We show how the air pressure data closely match theory and reveal whether a current is primarily dense or suspended.