A low-cost electronic olfactometer was developed, based on virtual sensors
array (an array of signals from a single metal oxide sensor) and on our pro
prietary semi-automated sampling system. This concept enables us to drastic
ally improve the sensitivity and selectivity of the entire olfactometer to
make it suited to food products analysis. Each step of the prototype was ch
ecked separately in earlier work, and finally the system performance was ev
aluated first on pure chemicals, and then at-line for grape juice during gr
ape-harvesting in Champagne. It is shown that over a 3-week vintage period,
the drift of the unique-sensor was quite low (absolute baseline value: avg
., 2152; std., 150; response to test solution: avg., 5.08, std., 0.39), and
the transient response to the temperature steps provided information for t
he discrimination and mapping of these products. The results correspond wel
l with the separative quantification of the aroma by high speed GC analysis
made in parallel on the same samples. (C) 2001 Elsevier Science B.V. All r
ights reserved.