Issue 96: The Basis for Recommending the 520 Hz Fire Alarm Signal

By Ian Thomas and Dorothy Bruck, Centre for Environmental Safety and Risk Engineering Victoria University, Melbourne Australia

Smoke and other fire alarms are intended to warn people of the presence of an unwanted fire so that they can take appropriate action such as extinguishing the fire (if practical, and appropriate means are available) or evacuating (themselves and others they are responsible for such as children or sick or elderly adults) from the building to avoid being injured or killed by the fire. To do this the detector component of such alarms must reliably detect a fire as early as possible while at the same time avoiding being triggered by phenomena other than a fire. To be effective, upon fire detection, alarms must provide a warning signal but remain "silent" otherwise. Usually the signal is a loud and recognisable sound which must be noticed quickly by people so that they can react as soon as possible. Fire alarm warnings may be beneficial at any time but it is obvious that people are most vulnerable, and the warning most valuable, when they are asleep and therefore alarm signals must be particularly effective when people are asleep. In this article we are concerned only with alarm sounds. Other warning signals are used for people who are hearing impaired.

It has been shown by us and others (for example Nober et.al. Fire Journal July 1981 and Kahn Fire Technology 1984) that relatively low level sounds of various types will awaken many sleeping people, but it is obvious that to awaken most people alarm sounds should be very loud in locations where people sleep (mainly bedrooms and lounge rooms) and, at a given sound level or volume, should be noticed and acted on by as many people as possible. Previous research, confirmed by our research, has shown that different sounds at a given measured sound level (dBA) will awaken sleeping people in different proportions, depending on a variety of characteristics as well as individual differences.

The groups of sleeping people tested have included children, unimpaired adults, adults impaired with alcohol, adults impaired with hypnotics (sleeping tablets), adults aged over 65 years and adults who are hard of hearing. Other than the unimpaired adults, these are people who are unusually vulnerable to injury and death from fire (particularly while they are asleep). 

Many different sounds have been used in our testing of the response of these people while sleeping. These sounds include the current smoke alarm signal used in Australia, the US and many other countries (an ~3100 Hz pure tone), pure tones of 400, 800 and 1600 Hz, 400, 520, 800 and 1600 Hz square wave sounds, white noise, whooping sounds (one varying between 400-1600 Hz and another between 400-800 Hz) and voice messages (child's mother's voice, voices of male and female actors). In all of these tests the 520 Hz square wave has been found to be as good or better at waking people than the others while the current smoke alarm sound has consistently been found to be the worst of those tested.

A useful comparison of the current smoke alarm sound (~3100 Hz tone) with the 520 Hz square wave sound when presented at the same level at the pillow (75 dBA except as noted below) compares the proportion of people in the various groups who slept through the alarm. (The proportion that slept through represents the people who, because they are unresponsive, remain vulnerable to the fire even though the alarm has sounded.) The comparison uses the ratio of the proportion that did not respond to the current smoke alarm sound and the proportion that did not respond to the 520 Hz square wave sound. Thus a ratio of 1:1 means equal proportions, and a ratio of 12:1 means that 12 times as many people did not respond to the current smoke alarm sound compared with the 520 Hz square wave. In the following, N = number of people tested and BAC = Blood Alcohol Concentration. Thus of:

• children aged 6-10 years (89 dBA, N = 22 for current smoke alarm sound, N = 27 for the 520 Hz square wave) the ratio was 12:1
  
• deep sleeping young adults (N = 14) the ratio was 6:1
  
• adults aged > 65 years (N = 42) the ratio was 4:1
  
• hard of hearing adults (N = 38) the ratio was 7:1
  
• young adults with BAC > 0.05 (N = 13 for the current alarm sound, N = 27 for the 520 Hz square wave) 38.5% slept through the current alarm sound, 0% slept through the 520 Hz square wave (ratio = infinity)
  
• sober young adults (N = 24) 31% slept through the current alarm sound, 0% slept through the 520 Hz square wave (ratio = infinity)
  
• adults aged >65 years (N = 10) 8% slept through the current alarm, 0% slept through the 520 Hz square wave (ratio = infinity)
  
• adults aged >65 years (N = 10) under the influence of their prescribed hypnotic 17% slept through the current alarm, 0% slept through the 520 Hz square wave (ratio = infinity)

This comparison makes it obvious that at the same sound level (75 dBA) the 520 Hz square wave is far superior to the current alarm sound for waking all of these vulnerable people.

An explanation for this superiority involves the complexity of a square wave sound. The 520 Hz square wave has been described as being a dissonant noise. It has a fundamental frequency of 520 Hz and subsequent peaks at the 3rd, 5th, 7th etc. harmonics. These multiple peaks are less likely to be masked by ambient noise than a single frequency. Several of the peaks are in the frequency range where human hearing is most sensitive, are also more than a critical bandwidth apart, and the different frequencies activate different parts of the basilar membrane in the inner ear thus increasing the perceived loudness.

More detailed information on the research and the results is available in the following references (* = review papers)

*Bruck, D. and Thomas, I., (2008). Towards a better smoke alarm signal – an evidence based approach, Karlsson B. (ed.), Proceedings of the 9th International Symposium of the International Association for Fire Safety Science, Karlsruhe, Germany, pp 403-414.

Bruck,  D., Ball, M., Thomas, I. and Rouillard, V., (2009). How does the pitch and pattern of a signal affect auditory arousal thresholds? Journal of Sleep Research, 18, 196-203.

Lykiardopoulos, C., Bruck, D. and Ball, M., (2014). The Effect of Hypnotics on Auditory Arousal Thresholds in Older Adults. 22^nd Congress of European Sleep Research Society, Tallinn, Estonia, 16-20^th Sept 2014 (P343).

*Thomas, I. and Bruck, D. (2009). Awakening of Sleeping People – a Decade of Research Fire Technology, 46 (3), 743-761.

Related Articles:
3rd Quarter 2013 – Battery Calculations for Fire Alarm and Signaling Systems – NEMA
Poor planning and missing, incomplete, or incorrect secondary power calculations are among the most common causes for rejection of a submittal to an engineer or to the authority having jurisdiction. This article shows how to use demands and durations to calculate the net required capacity for batteries that are used as part or all of a secondary power supply. READ MORE

3rd  Quarter 2011 – Professional Qualifications for Designers of Fire Alarm and Signaling Systems – NEMA
An overview of the steps NFPA has taken to define what makes a "qualified” signaling system designer and how those responsibilities differ from those of the technician. Changes that were made in NFPA 72 2010 to denote specific responsibilities are outlined. A discussion of what changes will appear in the 2013 edition of NFPA 72 is also included. READ MORE