Principles of dielectric property measurement by microwave free-space transmission measurements are presented, and the important sources of errors in such measurements are discussed. A system, including a vector network analyser, horn/lens antennas, holder for grain and oilseed samples and a radiation absorbing enclosure that was used for such measurements is described, and the techniques and procedures followed to obtain reliable permittivity data for wheat, shelled corn (maize) and soybeans are outlined. Data illustrating linear relationships between microwave attenuation and phase shift per unit sample thickness, each divided by the bulk density of the granular materials, and frequency and moisture content are presented graphically. The linear dependence of calculated permittivity components, dielectric constant and loss factor, on bulk density is also shown, and permittivity components for wheat, corn and soybeans are listed for reference at frequencies from 5 to 17 GHz at different densities and moisture levels at about 23 °C. Permittivity values are also listed for the same three commodities, adjusted to a medium density value through use of the Landau and Lifshitz, Looyenga dielectric mixture equation, for the total range of moisture contents at 10 GHz and at the same temperature.

Microwave techniques have been considered for a long time for moisture sensing in many food processing and agriculture-related industries. They are suitable for on-line realtime monitoring and control. However, with particulate materials, bulk density fluctuations cause significant errors in moisture content determination. To overcome this shortcoming, densityindependent calibration functions are needed. In this paper, a new approach is presented in which both bulk density and moisture content are determined directly from measured microwave dielectric properties. A simple relationship between bulk density and the dielectric properties is identified, and a new density-independent function for moisture content prediction, exclusively dependent on the dielectric properties of the material under test (0 , 00), is proposed. The validity and applicability of this function are demonstrated with an extensive data set obtained from measurements on a granular material (wheat), over wide ranges of frequency (11-18 GHz), temperature (01 C-42 C), moisture content (10.6%-19.2%, wet basis), and bulk density (0.72-0.88 g/cm 3). Explicit calibration equations for moisture prediction at different frequencies and temperatures are provided. Although data obtained by a transmission microwave measurement technique were used, this new approach remains valid in general for other techniques, provided that 0 and 00 are determined accurately.

No abstract