4 Application to Standardisation: ISO 9533
4.1 A-weighted measurements. The criteria of ISO 9533 Para. 7.4.4.2.1 and 7.4.4.2 mean that the A-weighted level of the alarm must be equal to or greater than that of the machine. However, it has been shown above that this is not a necessary condition for a broadband alarm to function effectively.
4.2 One-third octave band measurements. The criterion of ISO 9533 Para C.3.1, requiring the dominant sound of the alarm to be 13dB greater than that of the machine in the dominant third octave band of the alarm, has been shown above to be inappropriate for broadband alarms. The alarms do not have a dominant band and their levels could be higher or lower than those of the machine, depending on the machine spectrum.
As the energy of a broad band alarm spreads over a range of frequencies, any single third octave band is lower in level than the third octave band which contains the main component of a tonal alarm. Typically, the difference might be up to 10dB in the dominant third octave band for a tonal alarm when comparing tonal and broadband alarms of similar A-weighted level.
5. Conclusions This work has shown that criteria which are necessary to ensure that tonal alarms function correctly are not transferable to broadband alarms, for which different criteria require developing
Appendix.
It has been noted that the CAT 725 spectrum (Fig 9), measured at Sevenoaks, had a dominant peak in the 125Hz band, giving the sound a tonal nature. As the spectrum differed markedly from those of other machines, and is unusual for diesel powered equipment, there was a suspicion that this machine was not producing a typical spectrum. Subsequently, the opportunity arose to make measurements on a second CAT725, used at a land fill site at Southend. The machine was run at its maximum governed speed 2100 rpm. This machine was about a year old, and had a lower noise level. The spectra in Fig14 compare that for the machine used in the listening tests with that measured subsequently, both at similar distances to the rear of the machine.
It is clear that the original machine, as used for the listening tests, has a different spectrum from that of the second machine. It is also likely that, if the spectrum of the second machine had been used in the listening tests, third octave band alarm levels of around 55-dB would have been adequate. This would have given an alarm A-weighted level lower than that of the A-weighted level of this second machine.
continued...
