The Distributed Measurement Systems: a New Challenge for the Metrologists by Alessandro Ferrero and Roberto Ottoboni Politecnico di Milano – Dipartimento di Elettrotecnica Milano - Italy 1 Politecnico di Milano Overview Introduction Evolution of the Measurement Instruments DSP-based architectures Virtual Instruments (VI) Distributed Measurement Systems (DMS) Metrology issues INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 2 Politecnico di Milano Signal processing & Measurement Signal processing is the basis of the measurement activity Any evolution in signal processing techniques has a direct impact on measurement systems INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 3 Politecnico di Milano Digital Signal Processing Analog signals can be converted into a sequence of digital samples The sampling theorem provides the conditions for preserving the information associated with the original analog signal The modern instruments are based on DSP techniques They have benefited of the recent, impressive evolution of the DSP techniques and devices INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 4 Politecnico di Milano The first revolution: DSP-based instruments Field Digital processing Meas. Input signals T&C ADC Mem Comp From: A measurement, an instrument To: A measurement, an algorithm INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 5 Politecnico di Milano The second revolution: Virtual Instruments Field Meas. Interf. Input signals T&C ADC Mem Comp An interface is added to: Provide an instrument-like, user-friendly Front Panel Provide a graphic programming interface INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 6 Politecnico di Milano The third revolution: Distributed Measurement Instruments Field Interf. Input signals T&C ADC An interface to an interconnection network is added Mem Comp Net Int. World INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 7 Politecnico di Milano The new paradigm The desired measurement result is provided by a network of cooperating instruments that share their resources Field ADC Measurement unit 1 Supervising unit ADC Field Internet/Intranet Field ADC Measurement unit j Measurement unit N INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 8 Politecnico di Milano Better performance … …more problems! The different units share resources and data collected under different conditions acquired by different hardware systems The evaluation of the measurement uncertainty becomes a problem. INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 9 Politecnico di Milano The most mischievous problem The lack of time synchronization between the sampling devices of the different units. When measured data are shared across a public-domain network (Internet), the transmission time becomes largely unpredictable. The risk is that the different units start working on different timelines. INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 10 Politecnico di Milano The lack of synchronization Event at time t1 Event at time t2 Unit 1 Unit 2 t t The same event is seen at different time instants by the different units Different delays are introduced Their estimation may be cumbersome The consequent uncertainty estimation useless INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 11 Politecnico di Milano The solution Clock synchronization A unique timeline is defined for all units A timestamp can be associated to each transmitted data Measurement uncertainty is not related to the transmission delay, but to the residual synchronization error. INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 12 Politecnico di Milano Clock synchronization How accurate? As always, it depends on the target uncertainty. The case of the identification of the origin of transient disturbances in the electric system. INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 13 Politecnico di Milano t2 Measuring unit 2 Measuring unit 1 t1 t0 t If t2 - t1 > 0, then the disturbance’s origin is upstream the measuring unit 1 The synchronization uncertainty must be significantly lower than t2 - t1 INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 14 Politecnico di Milano Clock synchronization How accurate? As always, it depends on the target uncertainty. The case of the identification of the origin of transient disturbances in the electric system. A GPS synchronization is required 50 ns with respect to UTC The case of the identification of harmonic disturbances INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 15 Politecnico di Milano Measuring units Load1 Feeder Load 2 Load j Load N Each measurement unit measures suitable power quality indices averaged over a given interval The synchronization uncertainty must be significantly lower than the duration of the averaging interval INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 16 Politecnico di Milano Clock synchronization How accurate? As always, it depends on the target uncertainty. The case of the identification of the origin of transient disturbances in the electric system. A GPS synchronization is required 50 ns with respect to UTC The case of the identification of harmonic disturbances A NTP synchronization is sufficient (10 ms) INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 17 Politecnico di Milano Conclusions DMS benefit of the evolution of the interconnection of the computing systems. They are the natural evolution of the remote measurement systems They are a promising tool for the solution of very complex measurement tasks The present new problems in metrology If disregarded they may lead to largely incorrect results INGRID 2008 – Lacco Ameno – Ischia – ITALY – April 9-11, 2008 18