|
|
 |
Scientific news from Division 1 |
Robust signal statistics for the realization of the unit of sound pressure
|
|
By using advanced signal statistics procedures, the measurement time required for the pressure reciprocity calibration of laboratory standard microphones was reduced.
|
|
The realization of the unit of sound pressure in PTB by means of the pressure reciprocity calibration of laboratory standard microphones requires the precise measurement of the ratio of the open-circuit output voltage generated by the receiver microphone to the current through the transmitter microphone, i.e. the so-called electrical transfer impedance. It is determined by comparison with a computer-controlled reference resistor which is traced back to the national reference. This leads to the task of measuring alternating voltages which are affected by acoustical and electrical interference as well as by drift and transients as precisely as possible with a minimal input of time. An important prerequisite for this is a reliable and robust detection of signal stability.
In order to achieve this aim, multi-level "control charts", as known from statistical process control, were implemented within the measurement control software (Figure 1).
|
|
 |
| Figure 1: Control chart. The left part of the figure shows a process which initially enters the critical range and subsequently causes a reset of the averaging. In the right part the process returns to the acceptance range after having entered the critical range.
|
|
Signal quality criteria, e.g. the cumulative sum of deviations from the average (for drift detection) and the relative standard error of the voltage readings to be averaged, are monitored by means of these control charts. Starting with a minimum number of averages, the required number of averages is derived from the behaviour of the criteria within the control chart ranges:
- Inside the acceptance range the process is considered to be stable and the averaging is completed.
- If the warning limit is exceeded, but the upper limit is not exceeded, the averaging is continued until the process returns to the acceptance range.
- If the upper limit is exceeded, the averaging is reset.
The application of this multi-level adaptive procedure enables a fast derivation of the required number of averages during the measurement and the reliable elimination of measured values which are affected by disturbances.
|
|
Contact person: |
|
Thomas Fedtke, Department 1.6, Section 1.61, Thomas.Fedtke@ptb.de
|
© Physikalisch-Technische Bundesanstalt
Page created: 30.3.2009, last update: 30.3.2009,
Dirk Röske
|
|
