Robert Wallner, Institute for Measurement Technology
Christoph Beisteiner, Institute for Measurement Technology
Bernhard G. Zagar, Institute for Measurement Technology

In our proposed paper we present a low-cost real time solution for measuring physical quantities in a 3D volume based on the open-source projects LinuxCNC and COMEDI. LinuxCNC uses a Linux kernel with real time extensions (RTAI) and is normally used for CNC controlling milling or lathing machines. We had repurposed LinuxCNC as a part of our realized measurement system. The major advantage compared to state-of-the-art 3D scanning measuring systems, is that LinuxCNC permits to traverse all axes simultaneously. Therefore LinuxCNC allows scanning arbitrarily geometries, such as conical helical lines by using linear and circular movements. We have implemented a software solution utilizing COMEDI to transfer the acquired measurement data from any sensor into the hardware abstraction layer (HAL) of LinuxCNC. Furthermore, a real time data logger, based on a circular buffer, to record sensor positions and measurement signals concurrently from HAL, was implemented. For recording and post processing, the collected data from the data logger, is transmitted via TCP/IP to a client application like Matlab or Octave.

For demonstrating the functionality of the system a Senis 3D Hall sensor, a gantry robot with a scanning volume of 135 x 135 x 135 mm³ and a data acquisition (DAQ) card NI6221 are concurrently operated. The hall sensor in combination with the DAQ-card acquires the magnitude of the Cartesian magnetic field components down to 0.5 A/m. As a test object for the measurement setup a strip board containing a deficient trace is used. The resulting magnetic field due to a current of 3A applied to the track was scanned with a speed of 400 mm/min in approx. 10 min and with a resolution of 300 x 200 pixels in an area of 1 x 1 cm², 600 µm above the strip board. The resulting magnetic distribution field indicates the position of the defect within the trace easily.

As a conclusion we realized a measurement system that is capable of sampling a vectorial quantity in an arbitrary 3D volume with a high scanning velocity of 1380 mm/min and a cycle time of 1 ms we achieved a spatial resolution of 44 pixel/mm. Furthermore the used DAQ card supports the sampling of up to 8 analog channels to acquire signals parallel from multiple sensors thereby relinquishing spatial resolution or scanning speed.