Safe and Sound-EN 60849

EN 60849, it turns out, is nothing new. Except that in some regions its teeth are beginning to leave a mark, in a way that may be taking some by surprise. Paul Malpas explores the emergence and application of this historic document and its impact in the battle to ensure effective and reliable life safety systems.

Sound systems for emergency purposes are a part of our public life: any sizeable facility in use by the public is likely to be required (at least in the UK) to use a voice, by way of initiating any evacuation or other emergency action: the anonymous ring of a bell of sounder just doesn’t cut down those critical early response times the way that the authority of a human voice does. In football and other sports stadiums, where there may be between15,000 and 70,000 enthusiastic and noisy supporters, the power of the explicit instruction is mighty. Similarly, airport and railway users intent on the urgency of their transport link need more than an easy-to-ignore ringer before they abandon their travel plans and take a route straight out of the area by means as instructed.

Anyone who has worked in the specification, manufacture or delivery of systems for emergency purpose, must know that the demonstration of due diligence in the provision of safety-critical elements is a tricky business, balancing the risks of failure and ineffectiveness with the commercial risks of trying to cover every conceivable base. Unfortunately, well –meaning providers who try to build in protection against everything they can imagine are just likely to price themselves beyond the next (less imaginative) provider. Hence the role of standardisation, hopefully so that users, purchasers, specifiers, installers, suppliers and manufacturers can all (more or less) agree on how much is enough.

IEC what you’re saying:

EN 60849 “Sound systems for emergency purposes” has been around since 1998, and traces its lineage back to the early 1990’s as IEC 849. It has always included essential system performance requirements that the system shall be effective in delivering intelligible speech and that the system shall be reliable when called upon to do its safety-critical job.

Over the years, various practical requirements have crept in, intended, it is assumed, to assist in the delivery of the above essentials, in lieu of a water-tight and practicable definition of sufficient levels of effectiveness and reliability respectively. These may be necessary at the time (and evidently were successfully argued as such) in order to bridge the gap between essential requirements and specific obligations.

Specifics:
So let’s take stock of the requirements made in 60849, and how they might affect the design, specification and delivery of emergency sound systems.

Speech Intelligibility

A sound system for emergencies is of little use, reliable or otherwise, if the speech produced through it is not intelligible, any ore than a letter written in smudged black ink on a dark blue background. In 1998, IEC60849 became the first international standard to apply a requirement on achievable Speech Intelligibility. In the UK this had been in place since 1992 with BS7443 but from 1998 the Speech Intelligibility arm was given more strength in the sort of design negotiations that lead up to the formation of the contract specifications.

The achievement of speech intelligibility is a system issue in the broad sense, where the system includes all the hardware, including the choice of loudspeakers, where they are placed in the acoustic space and the acoustic response of the space they work into. Figure 1 illustrates the acoustic conditions that might be expected in typical public spaces, both with and without attention to architectural finishes. For illustration, acoustic response is shown by a simple Reverberation Time (RT) evaluation – this is a measure of acoustic ‘liveness’ with higher values representing a more live space where sounds take longer to decay. Slowly decaying syllables make it harder for us to understand subsequent syllables in the same utterance, hence high RT values, such as from a lack of soft architectural finishes or contents, are bad news for speech intelligibility.

Overlapped with this is the range of acoustic conditions typically necessary to make a distributed loudspeaker system comply with speech intelligibility requirements. Speech intelligibility in this sense is an electroacoustic issue, not just about electronic equipment but as much, if not more, about architectural acoustics.

As a system standard, with overall performance requirements, 60849 forces a bridge between the system design and the architectural acoustic conditions, and when things work well that bridge can be made at a point in the design and procurement where both banks of work can be influenced in concert with each other. If this process works, system installers and commissioning engineers will find acoustic conditions already compatible with the requirements made through 60849. This is key to reducing the contractual arguments at completion arising from disappointing sound quality, specifically those arguments over speech intelligibility where the system itself may not be to blame.

Methods of validating Speech Intelligibility

Annex A of 60849 is an informative appendix where a number of methods for the “measurement of speech intelligibility” are described. This includes Speech Transmission Index (STI) and its commonly cited derivative, RASTI, as well as a number of methods using specially selected word lists to test panels of listeners. Technically, these methods provide means of measuring a number of indexes derived for evaluating speech intelligibility through empirically derived psycho-acoustic models. IEC60268-16 is the full technical reference for applying these methods, but 60849 does reflect some of the important limitations of each method. For example, the RASTI method is usually only applicable to un-amplified speech, and can only be applied to amplified speech in substantially linear systems.

Annex B also provides a method of relating the various indexes to something called the Common Intelligibility Scale, or CIS. This is a convenient concept for applying contractual satisfaction tests, but beware of over-reliance on the relationships shown as it is quite possible to fool the CIS scale by multiple conversions.

There is also a simple statistical evaluation method provided, in order to avoid potentially misleading evaluations where the absolute minimum measured value across a whole potential listener area must be found and compared for compliance. Instead, sampling of locations is permitted and the requirement is for the mean measured value minus one standard deviation to comply with the numerical value set (eg STI 0.5). This, assuming a normal statistical distribution of values, would effectively give a value for which there was 95% confidence of exceeding for any selected location over the area from which the samples were taken. An even distribution of samples is important for this to be valid, and not one that finds more ‘good’ spots than is representative.

The Standard does not specify the noise conditions in which to evaluate the chosen speech intelligibility metric (though it does specify that the signal and noise conditions are measured during the evaluation, and how this should be done). However, the difference between the system broadcast signal level at the ear and the background noise at the same location and moment (signal-to-noise ratio) is a critical factor in the resulting values. For example, a system measuring STI 0.53 in quiet conditions (such as might be found ‘out of hours’) may only measure 0.48 in more representative noise conditions. Where this might be critical to the success of the system (ie in most cases where the issue is raised) the signal-to-noise conditions should be stated, or at least the evaluation noise conditions (assuming the system commissioner is at liberty to set the appropriate signal level, though some controls on the upper and lower limits of this set-up scope are necessary). The effect of this noise condition must then be applied if the resulting data is to be valid in confirming a contractual satisfaction.

System Resilience

Essentially, the system must be reliable, which means:
- resistant to physical damage (and functional or performance loss) to equipment, cables etc from maintenance, use, wear-and-tear or climatic conditions;
- resistant (but not immune) to damage from the effects of a fire
- resistant to the loss of mains power, either in the 24 hours leading up to any incident and/or for 30 minutes during one;
- resistant to the failure of any critical component, such as an amplifier, power sources, circuit protection, emergency microphone, message playback, emergency system interface or any critical software-controlled process;
- fully self-monitoring of all potential critical failures, with prompt and clear reporting;

As described earlier, the degree of reliability has not been defined, or any objective method of evaluating it. Instead, there have been a number of prescriptive equipment-level requirements, as well as conferring the requirements of the relevant fire alarm system standards in the (common) case of where the emergency sound system is installed in combination with a detection system. None of the requirements are particularly onerous if equipment is being developed with 60849 in mind. However, the use of pre-existing equipment may often appear to detract from the more detailed requirements of 60849, forcing the designer or supplier to make a careful risk-assessment argument for the equipment. Assuming that a satisfactory level of resilience can be demonstrated, balancing factors would include the implied or demonstrable reliability advantages of specific items of equipment produced in high-volumes for allied markets demanding high reliability. Another factor would be the ability of the system to deliver intelligible speech by the use of specialist equipment.

The owner or operator are also conferred obligations in the standard, relating to operations, record keeping and maintenance, as well as the nomination of a responsible person to ensure these obligations. Of course, the purchaser also has the power to insist on adequate and appropriate specification of the system (and the acoustic conditions) at the start of the job, which is probably the most critical factor in ensuring a satisfactory outcome.

The Future

Around 2003, there was a draft of the third edition of 60849 prepared, which would have constituted a technical revision and may have moved closer to specifying the kernel of performance and reliability requirements, freeing up the details for the designers and manufacturers to respond to creatively. However, it is understood that there was a huge volume of (probably conflicting) comments, possibly because of the regional diversities involved, and of the revision was shelved. The future of this standard is therefore uncertain, but remains active. In Europe, it seems that the series of product standards in EN 54 (from CEN) are set to succeed 60849, but some are fearful that this would herald a loss of the ‘system’ approach. Whereas this would undoubtedly be more convenient for some manufacturers, whether it results in more effective and reliable systems remains to be seen. At the international level, IEC and ISO have realised that 60849 was less appropriate to be continued as an IEC, essentially because of the system aspect and the architectural acoustic factors, and an ISO Committee has taken over development, as a product standard for the "Sound System control & indicating equipment" (ISO 7240-16) and a performance standard for the "Design, installation, commissioning and service of sound systems for emergency purposes" (ISO 7240-19). Watch this space!


The author Paul Malpas works for consulting firm Arup Acoustics.


Article Categories




Most Viewed