Although acousticians use many different descriptors for the quantities they measure, the basic quantity behind all of them is sound pressure. It is not practical to maintain a standard source which generates a Pascal of sound pressure, instead all acoustical measurements are traceable through accurately calibrated microphones.

Development up to 1970

The condenser microphone was first described by Wente [20] who also built the first pistonphone in order to calibrate the microphone. The pistonphone consisted of a 45 cm³ chamber placed over the microphone diaphragm, into which a piston moved. The piston rod was connected to a fly-wheel which was driven by a motor. When the motor operated, the change in the volume of the chamber could be calculated from the diameters of the piston and fly-wheel and hence the sound pressure could be calculated from the adiabatic gas law.

The pressure generated by modern commercial pistonphones cannot be calculated accurately, such devices are calibrated by microphones that have already been calibrated by some other means. There are modern calculable pistonphones [21] but they only operate at frequencies below a few hundred Hertz.

In the first half of this century various methods were used around the world to calibrate microphones and hence maintain a measurement standard for acoustics. In France the preferred method was the thermophone [22,23]. The basic principle of this device was that as an alternating current passed through a conductor it would generate heat. If the conductor was made thin enough it would have a very small heat capacity, so its temperature would closely follow the alternating current. The air surrounding the conductor would be heated by it and the periodic expansion of the air would be a sound wave. The strength of the sound source could be calculated.

Meanwhile at the National Physical Laboratory (NPL) in the UK the preferred method for generating a known sound pressure involved measurements with a Rayleigh disc [24,25]. The Rayleigh disc is a light circular disc suspended by a delicate fibre so that it can rotate about a vertical axis against the restoring torque of the fibre. If such a device is suspended in a current of air it tends to turn itself so that its plane is normal to the direction of the air stream. This tendency persists if the direction of air flow is reversed so that a steady deflection will be obtained in an alternating air current and the device may be used to measure the particle velocity in a sound wave. The displacement of the disc can be measured by reflecting a beam of light from it onto a screen and the pressure at the microphone can be calculated from the particle velocity.

At the National Bureau of Standards (NBS) in the USA an absolute method of microphone calibration was being developed which did not require a known sound source [26]. This was called the reciprocity method. It relies on a special property of the condenser microphone - that its sensitivity as a transmitter is the same as its sensitivity as a receiver. When two microphones are coupled acoustically, one driven to emit a sound and the other receiving, the product of their sensitivities can be found from the current in the transmitter, the voltage from the receiver, and the acoustical impedance of the coupler. The sensitivity of an individual microphone can be found by taking three microphones and measuring the sensitivity products of all pair-wise combinations.

Uncertainties on the results of microphone calibrations by either the thermophone or the Rayleigh disc were quite large, the effect of vortex motion around the disc alone contributed 0.2 dB [27], so when the International Electrotechnical Commission wanted to standardise a method for calibrating microphones in the 1960s the reciprocity method was the only way whereby uncertainties of less than 0.1 dB could be expected over most of the audible frequency range. IEC 327 [28] was published in 1971 and the world agreed on the principle to be used for realizing the primary standard for acoustics.

Development Since 1971

Over the last thirty years the method has been continuously refined:

A comparison [29] between European laboratories in the late 70s identified an omission from IEC 327 by highlighting the need to pay particular attention to the measurement of air volume contained in the front cavity of the microphone [30]. This cavity contains a screw thread and consequently the volume had to be measured by an acoustical method.

The effort required to pilot this comparison convinced NPL that an automated system was required for making the calibrations if the calibration method was to be refined further and it developed the first such system [31].

In the 1980s, the EC funded further European work to extend the reciprocity method from the type LS1P microphone (commonly called a 'one-inch' microphone and based on an American designTechnical Guide - Sound Measurements[32] from the 1940s), to the new, smaller, LS2P microphone [33,34]. During this development many of the models which underpinned the method were challenged and refined. Models for the calculation of impedance of narrow tubes [35], the effects of heat conduction [36], and the effect of microphone volume [37] were all improved. The 1980s also saw advances in the philosophical understanding of reciprocity [38] and verification of the reciprocity theory as applied to condenser microphones by an independent method [39].

To test the implementation of the then current IEC Standard, the IEC organised a comparison of the national standards of sound pressure held by 17 laboratories[40]. At frequencies up to 1 kHz, the results from all but four laboratories fell within a span of 0.06 dB. At higher frequencies, results from all but three laboratories fell within a span of 0.15 dB. These findings were broadly consistent with the overall calibration uncertainties claimed by the laboratories. Some discrepancies could be traced to the neglect, by some laboratories, of significant systematic influences e.g. the finite impedance of the microphone diaphragms, and the inadequate allowance for these influences in the uncertainties claimed.

The work of the 80s resulted in the IEC revising their Standard [41] and the key laboratories turning their main attention to the free-field reciprocity method [42]. However the publication of the revised Standard was the impetus for many other laboratories to upgrade their systems in the 90s. and a group of European laboratories held a comparison of LS2P microphone calibrations [43] through Euromet, essentially extending the EC work of the 80s to a wider group of laboratories.

As more laboratories adopted the new Standard, and sought advice from the laboratories that had participated in its development, it became apparent that comparisons of microphone calibrations were an inefficient way of determining the causes of differences between laboratories. NPL developed[44] a 'coupled microphone simulator' as a stable device that could verify that the electrical measurements associated with a reciprocity calibration were being performed correctly while a Euromet project developed reference data files for checking the associated calculation programs.

At the end of the 90s the development cycle is about to start again. Regional key comparisons are under way within Euramet, SIM (Sistema Interamericano de Metrologia) and APMP (Asia/Pacific Metrology Programme) and the new CIPM Consultative Committee on Acoustics, Ultrasound and Vibration is also organising one. The IEC has just decided to revise the Standard again, although the changes should be minor this time.

Recent development work at the major laboratories has concentrated on refining the methods for transfer calibrations of microphones [45,46,47,48,49] and some of this work will form new parts of the IEC Standard on measurement microphones.

The Future

The reciprocity method has been under development as a method for realising the primary standard for sound pressure for over 50 years and has risen to each new challenge. However there are some fundamental limitations in sight. There is no great drive within acoustics for greater accuracy beyond current levels but there is a drive for ever smaller transducers because they disturb the sound field less. As they become smaller they become increasingly difficult to calibrate [50] to existing uncertainty levels. If any absolute method can be found either to generate or measure a sound field then we should be giving it serious consideration for the future.

References >>