A thermal flow transduction method combining the advantages of calorimetric and hot-film transduction principles is developed and analyzed by Finite Element Method (FEM) simulations and confirmed experimentally. The analyses include electrothermal feedback effects of current driven NTC thermistors. Four thin-film germanium thermistors acting simultaneously as heat sources and as temperature sensors are embedded in a micromachined silicon-nitride membrane. These devices form a self-heated Wheatstone bridge that is unbalanced by convective cooling. The voltage across the bridge and the total dissipated power are exploited as output quantities. The used thin-film thermistors feature an extremely high temperature sensitivity. Combined with properly designed resistance values, a power demand in sub-1mW range enables efficient gas-flow transduction, as confirmed by measurements. Two sensor configurations with different arrangements of the membrane thermistors were examined experimentally. Moreover, we investigated the influence of different layouts on the rise time, the sensitivity, and the usable flow range by means of two-dimensional finite element simulations. The simulation results are in reasonable agreement with corresponding measurement data confirming the basic assumptions and modeling approach.

This paper reports on a design and simulation study aiming at high-accuracy 2D micromachined thermal flow transducers. The scope is restricted to micromachined devices featuring a square-shaped membrane incorporating central symmetric thin-film devices. A microthermistor array probed spatial excess temperature variations while the main heat supply was alternatively established by optional heating resistors or by pronounced self-heating of the thermistor devices. Proper device designs enable leading edge transducer performance without sophisticated signal conditioning schemes. We found that a high azimuthal uniformity of flow magnitude transduction is tantamount to a precise azimuthal accuracy. The most advanced result gave a maximum azimuthal aberration of 0.17 and 1.7 degrees for 1 m/s and 10 m/s, respectively, while the corresponding magnitude uniformity amounted to 0.07% and 0.5%. Such excellent specifications exceed the need of ordinary meteorological applications by far. However, they are essential for, e.g., precise non-contact measurements of 2D relative movements of two quasi-planar surfaces via the related Couette flow in intermediate air gaps. The simulations predicted significantly better device characteristics than achieved by us in first experiments. However, this gap could be attributed to imperfect control of the flow velocity field by the measurement setup.