Construction and calibration of a new type of heated leaf replica for measuring leaf boundary layer conductance g(b)(H) is described, offering two novel features: (i) the heated and unheated surfaces are situated on the same replica, and (ii) the temperature difference is measured inside the replicas instead of at the surface. Wind tunnel experiments were used for calibration against completely heated standard replicas. Definition of a normalised characteristic length L to be used in the Reynolds and Nusselt numbers of a flat surface is proposed. Good agreement between experimental data and engineering relationships were found for circular and vine-leaf shaped standard replicas. Six partially heated vine-leaf replicas were installed in a vineyard in central Portugal together with complementary meteorological sensors. Leaf boundary layer conductance was continuously measured, at different locations within a vine, for 40 days, and compared with a reference wind speed U-ref measured above the canopy (2.84 m). The measured conductances were closely correlated with U-ref but due to local shelter effects and/or directional effects on convective transfers, the correlation depended on wind direction and location of the replicas within the canopy. Although further improvements are needed in their construction and calibration, some features of these partially heated leaf replicas may be advantageous for long term analysis of micrometeorological variations in complex canopies.