WO2012049548A1

WO2012049548A1 - A notification device with audio emission and strobe light

Info

Publication number: WO2012049548A1
Application number: PCT/IB2011/002368
Authority: WO
Inventors: Bernhard Piller; Mario Casamassima
Original assignee: Bernhard Piller; Mario Casamassima
Priority date: 2010-10-11
Filing date: 2011-10-07
Publication date: 2012-04-19

Abstract

A digital addressable notification system for paging and public safety is provided. The notification devices (ND) comprise a loud-speaker (6), one or several strobe lights with different colours and angular orientations, an amplifier (3), a digital volatile storage device, a digital non-volatile storage device, a microprocessor (1), a microphone (7) and a network interface. The audio information for paging is received as a digital signal and is decoded by the microprocessor (1) and converted to an analogue signal, which is amplified and sent to the loud-speaker (6). Stored messages can be played to the speaker (6). The system uses various techniques to improve the power efficiency and has special features suited for guided evacuation and a two way intercom function is provided. The digital data may also be sent over the same pair of wires as the power.

Description

A notification device with audio emission and strobe light

Technical Field

The invention pertains to a system for notification of messages, paging, or evacuation in case of fire or other dangers to life of health. The invention pertains furthermore to a notification device for use in such system and a method for operating such system.

Background

Systems of this type are typically installed in buildings or similar closed structures such as school houses, hospitals, factories, ships etc. They are used for evacuation purposes in case of emergencies as well as for notifying or paging people present in the structures. Known voice evacuation systems use amplifiers at one or several locations of the building and many speakers attached to these amplifiers. Typically, a 25 Volt RMS signal (about 40 V peak to peak) or a 70 Volt RMS signal with as much as 100 Watt of power is supplied to all the attached speakers, which are coupled to the wires through

transformers with adjustable tabs. The available power for a speaker is usually in the range of ¼ Watt to 10 Watts and the actual power is selected by shorting one of the tabs of the transformer. Often a strobe is also mounted on the speaker housing. Common problems of known systems that are commercially viable in the area of evacuation with a large number of speakers are the inability to select individual speakers for paging, the inability to play different messages to speakers on the same line, the inability to change remotely and dynamically the volume of individual speakers, the susceptibility to interference with other signals on cables running next to the cables of the evacuation system, the need for thick cables to transport the high currents, and the inability to use the speaker and the strobe for guided evacuation.

US 7,508,303 discloses an alarm system with a number of notification devices having speakers and visual alarms. A system controller within a notification device controls the strobes and speakers of the notification devices. The notification device may have a micro controller, which controls a visible or audible indicator. The system uses one audio amplifier for all the speakers, and the audio is broadcasted as an analog-power-signal. Digital signal communication is limited to commands and requests for status information and excludes the audio signals. The notification devices connect the speaker-line to one of the available tabs of the transformer. The power transmitted on the control bus is only used for the notification device and optionally for attached strobes, but not for the speaker.

US 2005/0231349A1 proposes a voice over (VoIP) solution based on H.323 protocol to broadcast voice over Internet protocol (IP) based on Ethernet. IP-based speakers are well-known, however they are not suited for wide-spread use in the domain of mass notification because of their inherent high cost and are therefore not used in fire alarm and evacuation systems.

Ethernet based networks need shielded cables (Cat 5) with 8 wires, where each cable has to be connected to a switch. Figure 6a shows an example of a typical evacuation system of the prior art with IP-based architecture having a head end or central control station, two switches with 5-8 ports or repeaters, and a number of notification devices ND. The notification devices are connected to the switches by means of shielded cables in a star-like architecture. Two-wire cables lead from the switches and notification devices to power supplies.

IP-networks requiring shielded cables, switches or repeaters are too expensive when applied to evacuation systems. Moreover, the electrical characteristics of Ethernet-based networks do not allow multi-dropping and t-tapping, which is a common practice in evacuation systems engineering. Finally, the need of switches requires a star

architecture, which is disadvantageous due to additional cabling.

US 2007/0041590 A1 proposes a method with directional speakers to show the egress path to occupants to ensure an efficient evacuation of a building.

Summary of the invention

A system is disclosed for notification, paging, or emergency evacuation of people from a structure comprising a plurality of notification devices arranged in selected positions within the structure, where each notification device includes a loud-speaker able to release an audio signal or message into a designated area. According to the invention, each notification device in the system comprises an amplifier, and a controller, which are dedicated to a notification device's loud-speaker and where the controller, amplifier, loudspeaker are arranged to form a notification device as a combined entity at a given selected position within the structure. The controller in each notification device is configured to provide a built-in intelligence or logic. It consists of a microprocessor having a volatile (RAM) and a computing speed of at least 5 mips (million instructions per second) during an audio operation. Furthermore, the system is configured for exclusive digital data communication with base-band encoding (i.e. no modulation) and that allows the use of unshielded cables and comprises a wiring architecture that has multi-drop capability.

The multi-drop capability allows the connection of several devices in parallel, where a point-to-point connection used in Ethernet systems allows only two devices to be connected. A digital data communication that allows the use of unshielded cables includes for example data communication according RS-485 standard, which is the preferred 4-wire interface (data and power on two wires each).

A cost effective means to combine power and data on a single pair of wires, on the other hand, consists of AC-coupling the data onto to a DC-power line decoupled by means of inductors, which is the common practice in data splitters used for ADSL- or DSL-modems.

The microprocessor is configured for a computing speed of 5 mips or more, which is significantly higher than the computing speed used on common network devices. It allows the processing of digital audio including optional data decompression during audio playback of stored messages or live audio broadcast.

The system, as such, is configured for an exclusive digital transmission of audio signals throughout the system, that is an all-digital transmission of audio signals between a central node of the system, hereon referred to as a head end, and each notification device with its amplifier and loud-speaker. A system according to the invention can have one or more head ends connected on a network and each having a power supply and one or more line interfaces. In a particular embodiment, a notification device can comprise a microphone and an optional display.

In addition, the digital communication is suitable for an unshielded two-wire cabling with separate power or is suitable for a combined signal that carries power and base-band encoded data over one pair of unshielded wires.

Conventional and commercially available systems, in comparison, may use digital audio, however they rely on an analogue transmission (baseband, frequency modulation, amplitude modulation etc) between a main amplifier and the speakers, where the amplifier providing amplification for all speakers is arranged spatially distant and separate from the speakers.

Unlike evacuation systems of the prior art using digital communication to IP-speakers on Ethernet basis, as for example in US 2005/0231349A1 , the system according to the present invention allows the use of unshielded cables, which are advantageous in terms of cost and installation effort.

Furthermore, the type of architecture employed by the present is particularly suitable for evacuation and mass notification systems because it allows bus communications lines with multi-drop capabilities. Since the system uses the wiring architecture of existing installations, a retrofit of existing installations is very easy because the wiring of the building can be preserved and only the head end and the notifications devices have to be replaced. The use of an all-digital transmission of the audio signals to the speakers provides the advantage that the system may be operated with an increased immunity to

electromagnetic interferences as they may be encountered for example from lighting systems or when a cable is run next to a detector line of a fire alarm system.

Furthermore, the digital transmission allows not only for audio signals but also commands to be sent to the notification devices. It provides the ability to individually adjust the volume of the audio messages and to individually supervise the functional state of each loud-speaker.

Each notification device as a separate entity includes at least these components - controller, amplifier, and a loud-speaker. The electronic board for controller and amplifier may, in an exemplary embodiment, be mounted on the back of the load-speaker or in a separate housing next to the speaker.

Each notification device may comprise one or more additional optional elements, which allow for further system functions as described in exemplary embodiments hereon.

In one embodiment, each notification device comprises a flash memory able to store digital audio messages and has a capacity of at least 100 KByte.

Conventional voice systems have party lines, for example as many as 100 speakers, which are connected to the same amplifier and where the wiring is supervised by means of an end-of-line resistor. This kind of supervision has the disadvantage that a faulty contact on the speaker side of the transformer, which is needed to transform a 25 V RMS signal to a 1-4 V RMS signal, cannot be detected. For this reason, such traditional voice installations must be periodically tested by a procedure, where a person has to check every speaker in a building whether a tone can be heard or not.

In an exemplary embodiment of the present invention, each notification device comprises a microphone, mounted on-board or on or next to the speaker. The microphones provide a built-in self-test feature, whereby the system is able to supervise the function of each notification device. The self-test may be operated by commanding the controller of the notification device at regular time intervals to send an audio signal to the loud-speaker and to verify by means of the microphones whether the signal was emitted by the loudspeaker and could be received by the microphone or not.

The system according to the present invention is operated by a DC voltage that carries the power to all the speakers. Conventional systems, on the other hand, embed the power in the signal itself. It is well known that a sine wave of a given amplitude only carries half the power of an equivalent DC-signal of the same amplitude. The situation in the case of voice transmission is even worse since the amplitude has a burst-like shape with short time interval of high power and long intervals of very low power (< 10% of peak amplitude). It may very well happen in the case of a traditional installation dimensioned for 100 Watt and capable of transmitting 100W in the case of a sine wave, that in the case of voice transmission, only 10 Watts are utilized on average.

The system according to the invention, in contrast, is able to utilize its full power also when transmitting voice or audio signals. In order to deliver a clear and strong voice signal without any clipping, the present system is operated, in addition to the DC powering, using local energy buffers that can supply a peak-current during short intervals of time (for example, 1-2 seconds).

In an exemplary system of notification devices according to the invention, each

notification device comprisesith an energy buffer with a capacity of more than two times the average power times 0.2 seconds.

As an example, the energy stored in a capacitor is U^*U^* C/2, so a 10mF capacitor charged at 40 Volts would store 1600 V*V * 0.01 C / 2 = 8 Ws (where V = volt, C = coulomb, W = watt and s = second), which would allow the notification device to deliver to the loud speaker a peak power during one second, which is 8 W higher than the power supplied by the line. If only a subset of the notification devices on a line is used in a specific audio notification, the energy stored in the buffer of the inactive notification devices can be used by the active devices, thereby increasing even further the peak power. In a particular embodiment of the invention each notification device of the system is configured to allow the current to flow from the device to the power line if the line voltage is lower than the voltage of the energy buffer of the device.

Traditional notification devices comprise an input diode in order to protect the device from line reversal or polarity reversal. The notification devices of the system according to the invention on the other hand, are configured without such input diode. Instead, it comprises for example a diode arranged between the common ground and the positive terminal of the power line causing a short if polarity is reversed.

This configuration allows the device to effectively feed back energy to the power line, which can be used to supply additional power to other notification devices, which need more power than the line interface of the head end can provide.

This enables the system to command a very high audio volume level of audio signals played by devices, which are remote from the head end and where a drop of the line voltage is encountered due to the resistance of the power line. The system can be operated in that the audio signals are transmitted to the individual speakers of the notification devices using a time spreading technique. The method introduces incremental time delays of the audio signals played on the different speakers such that each speaker plays an audio signal or message with a time delay relative to another speaker.

In the case of stored voice messages the time delay can be accomplished by delaying the start of the message of each device. In the case of live voice messages, which are broadcasted from the head end by an operator synchronously for all devices, the time spreading is accomplished by means of a digital delay line. The incoming audio stream of the voice message is stored in a temporary storage (RAM) and used for digital analog conversion after an individual delay. The size of the digital delay line is the number of samples per second times the maximum delay in seconds. In case of a 8 kHz sampling rate with a sample size of 1 byte and two second delay, the size of the digital delay line would be 16 KByte.

This time spreading of the audio signals effects that the power peaks in a voice message will occur in each speaker at a different instant of time. This results in a time distribution of the power demand of the system as a whole. Since the ratio between peak-power and average power is decreased, the average power can now be increased without a risk of clipping the signal.

The improvement of power efficiency over conventional systems is given by the product of two factors. The first factor is the DC powering versus AC powering, which effects a doubling of the power efficiency. The second factor is the averaging effect by means of local energy buffering and time spreading, which can be estimated to a factor of 4 or more depending on the type of voice message. The total power efficiency improvement can therefore be a factor of 8 or more, which means that over a wire with the same linear resistance an eight times stronger voice (speech, not tone) signal can be played by using the proposed operation method.

Whereas in traditional systems, pre-recorded voice messages are stored at a central location or in a node of the system, to which several speakers are attached, the present system comprises in each notification device a flash memory able to store individual audio data for each notification device. A system with notification devices having such memory may be operated to emit simultanesouly a different and designated message on each notification device. For example a system having 200 speakers can simultaneously emit 200 different voice messages. Conventional systems are able to emit one message per group of typically 20 speakers or two to four messages for a whole building. This feature allows the system to play voice messages which are related to the exact geographical position of the notification devices.

In a further embodiment of the system according to the invention, the system comprises additionally in each notification device a microphone. This enables not only a functional supervision of the speaker as described earlier, but also two-way audio communication between each notification device and a central processor. It allows a fireman at the central processor to talk to all locations having a notification device and ask whether people are present in this area or not. Having located people in the building more quickly by this manner, the fireman could take rescue measures more efficiently and would waste less time by going to places, where there may be no people.

The microphone may furthermore be used to test whether two notification devices are acoustically isolated or not, that is whether the loud-speakers can be simultaneously heard by a person standing at a given location or not. If they were acoustically not isolated, then the speaker of one notification device can also be "heard" by the microphone of another notification device. The information whether speakers are acoustically isolated or not is important for the time spreading method (and some multi- speaker voice patterns). When two locations are isolated, a time delay of the message played on two speakers will not be noticed by occupants of a building. However, if two speakers are in the same room, a time delay between the messages played on the two speakers is absolutely not desirable, and the system would in this case not apply the time-spreading method for these two particular speakers.

In a further method of operating the system, messages are played in fragments, where separate fragments of a message are played by different speakers in order to guide a person along an evacuation route. For example, a message "to evacuate, go this way" is broken into a sequence of two parts of the message: a first fragment "to evacuate, go" and a second fragment "this way". Such operation only makes sense when an occupant can hear both fragments from the point, where he is standing and the speakers are not acoustically isolated. The system would use multi-speaker strategies for evacuation guidance only in the case of non-isolated speakers.

When a notification device receives the audio signal of the speaker of another notification device, it also can compute the approximate distance by measuring the time delay of the acoustic signal and multiplying this delay by the speed of sound (about 300 m/s). In a large space with many notification devices, by triangulation the two-dimensional relative position of the notification devices could be obtained. This is important in the case the installer wants to obtain egress paths based on the location of an event. This data of the egress path can be stored in the database of the evacuation system.

In an exemplary method, the egress path can be "learned" as follows: A technician walks through the building and activates the notification devices with a hand held sounder and on his way also triggers fire detectors, manual pull stations, and/or other alarming devices. The sound generated by the hand held device could be a beep of a given frequency, a dual tone frequency or some other signal that can be easily captured by the notification device through its microphone. The activation of fire detectors and of the notification devices is sent to a personal computer, which correlates all the information and obtains the following information: 1) The possible egress paths, the notification devices along this path and the distance between the notification devices in the case they are not acoustically isolated. 2) The relative geographical position of alarming devices and notification devices. All this information can be stored in the evacuation system. When the system receives an alarm message from a detector, it uses this knowledge to determine on which egress path the danger event has happened and in which direction occupants have to be guided in order to avoid the regions, where a danger has been reported. The advantage of this method is that during the perdiodic inspections (and testing) of the fire system, the system could perform this "learning" task without any additional human effort.

According to a further embodiment of the invention, the system comprises on each notification device one or more strobe lights. The combination of audio signals and optical signals for notification allows support of the audio notification by means of the optical notification and vice versa. The controller of the notification device could for example turn on a green light source on one side of the notification device and a red light source on the other side and play the message "To evacuate, follow the green light". Since the acoustic and the optical function is controlled by the same processor, such a combined function can be realized much more easily than with two independent processors and a

communication link between the two processors. The installation of one device instead of two is also less expensive.

Several variants of arrangements of light sources on the notification device are disclosed by means of the figures.

Brief description of the drawings

The foregoing features of the invention will be apparent from the following description of the preferred embodiments of the invention, as illustrated in the accompanying drawings. Fig. 1 is a schematic diagram of the electronic assembly of the notification device. In this example, a bridge interface for the speaker is used. Other single ended configurations are also possible. The functional blocks don't represent individual chips and some of the functional blocks can be integrated on one device. This embodiment of the invention uses a power line circuit, which transmits power and data on the same pair of wires

Fig. 2 is identical to Fig. 1 except that a 4-wire interface with power and data

communication on separate pairs is used.

Fig. 3 shows a strobe arrangement with several lights of several colours. The lights don't need to be mounted on the side of the housing; they can also be mounted near the centre as long as they point in different directions. Fig. 4 shows strobes, which may each emit light of different colours reducing the number of strobes necessary to implement the idea of the guided evacuation. Fig. 5 shows a version with more than just two possible directional indications. This may be necessary at a cross point in a building.

Fig. 6a shows a system with notification devices according to the prior art. Fig. 6b shows the global architecture of the system comprising a head end, microphones for paging, a switch board to turn on and off functions of the system and a plurality of line interfaces, which handle the short and over voltage protection and which couple the data onto the power line. The word "ND" means "notification device" and can include also other control elements like input modules, output modules, sensors etc.

Fig. 7 to 10 show the system's power demand as a function of time and in particular the effect of the operating method according to the invention including the time-spreading and energy buffering. In each figure, the first five single-lined curves show the power demand of five notification devices in the system, which each play the same message with time spreading. This means, the second to fifth device each play the message with a small incremental delay (around 300ms) relative to a preceding notification device. In each figure, the lowest and bold-lined curve illustrates the average power demand of the five notification devices. Fig. 7 shows the power demand of the transmission of a given audio message signal using the system according to the invention. The five notification devices play the audio message with a time delay, where the second, third, fourth, and fifth device each play the message with a small incremental delay of approximately 300 ms relative to a preceding notification device. Almost no energy buffering is applied in this case.

Fig. 8 shows curves for the power demand of the same five notification devices playing an audio message as in figure 7. The power demand illustrated here is for a system using an energy buffering of approximately 0.01 seconds. Fig.9 shows curves for the power demand of the same five notification devices playing an audio message as in figure 7. The power demand illustrated here is for a system using an energy buffering of approximately 0.1 seconds.

Fig. 10 shows curves for the power demand of the same five notification devices playing an audio message as in figure 7. The power demand illustrated here is for a system using an energy buffering of approximately 1 second.

Detailed description of the invention

A notification device ND embodying the present invention is illustrated in Fig. 1 and Fig. 2, where the only difference between the two figures is the line interface between controller 1 of the device and a head end HE of a system of notification devices. The heart of the device is the controller (1), which executes a program and as such provides the intelligence or logic to the notification device. The controller includes an ADC (analogue to digital converter), a DAC (digital to analogue converter), a serial interface for

communication and several lO's. The controller 1 receives data from a data splitter 2 or from data terminals and sends back data to the data splitter 2 or to the data terminal. The data splitter 2 is connected over a two-wire interface to a bus that carries power and data on the same two wires. As an alternative, data and power can be carried on separate pairs of wires as shown on Fig. 2. Besides the internal volatile memory or RAM the controller 1 can use an external RAM 4 with higher capacity, and in addition to an internal non-volatile memory or Flash memory 5, the controller 1 can use an external Flash 5 with higher capacity, typically to hold the digitised audio messages. The messages that the controller 1 receives contain general commands like for playing a message, setting the volume, or for handling the network or system management functions. The controller 1 also receives the digital audio streams of the microphone signal of the head end. The head end can be a stand-aone device or a node of a larger system. The audio data stored in the Flash memory 5 or sent over the network is decoded and converted to an audio signal by means of a decompression algorithm (if the digital audio is compressed) and a DAC. The audio signal is routed to the amplifier 3. A speaker 6 is connected to the output of the amplifier and the speaker can be supervised with the microphone 7. The electronic board of the notification device ND can be mounted on the speaker 6 or in a separate housing in the immediate vicinity of the speaker 6. One or several strobe-drivers 8 can be controlled by the controller 1.

When an event such as a fire or other emergency is detected in a building, different audi messages may need to be played at different locations depending on the location and type of the event. For example, the floor where the event happened needs to be evacuated first and other floors need just to be alerted. By storing all the messages on board, each notification device can be told to play a specific message and also the strobes may be individually turned on. Individual paging is supported and even several audio channels can be transmitted over the same pair of wires if enough band width is available or if compression is used. Another big advantage compared to conventional voice systems is the possibility to use a time spreading technique. A typical voice

(speech) signal has bursts of high intensity separated by gaps of low intensity. This behaviour causes a wide power distribution and the peak power is usually much higher than the average power. A hardware method consists in storing energy locally in a capacitor and to use this energy when not enough energy is available from the line. In addition to that, the power demand can be smoothed out by adding varying time delays to the notification devices. By starting the playback on the notification devices with varying delays the combined power demand of all devices varies to a lesser degree than the individual power demand, and the peak power is not much higher anymore than the average power. A conventional voice system, on the other hand, must be able to supply the peak power and the volume must be kept low enough to prevent saturation of the peak signals. The result is that, for typical voice signals, a conventional system can only output a fraction of the nominal power, which would be achieved with a pure sine wave. The digital system with the time spreading technique can output also typical voice signals at full power like with a sine wave (only square signal carry more power). The same time spreading technique can be used on the microphone channel by passing the audio signal through a "digital delay" line of individual length for each node. In order to minimise the impact of the delay on the user, the nodes, which are geographically close and which may be heard simultaneously from a given location should use the same delay.

Before the system can be used, all voice messages have to be stored to a Flash memory. They can be stored at the factory or can be transmitted over the network.

The onboard microphone of the notification device has three functions: 1) It can be used to supervise the own attached speaker, 2) it can be used to determine the vicinity of other notification devices and 3) it can be used to "learn" the evacuation routes. The supervision function of the own speaker is obvious: the processor outputs a tone to the speaker and measures the signal of the microphone. If no signal is received, the speaker is not working. This feature is considered as a self test and replaces a manual test by a technician. It is important to find out whether notification devices are close together or widely separated. In order to get an efficient time spreading, it is desirable to use a different delay on each speaker, but from an acoustic point of view, it is a problem if two speakers with a substantial delay can be heard from one location. For this reason, if several speakers are in the same room, they should be set to the same delay, and if they are in adjacent rooms, they should have different delays. The third use of the

microphones is to "learn" the evacuation route and to correlate the detectors with the speakers. The method works as follows: When the technician checks an installation he emits periodic tones while he walks through the building and activates the detectors. These tones are received by the notification devices and by knowing the node number of the activated detector, the vicinity between detectors and notification devices may be obtained. Also the sequence of detection of the tones on the notification devices allows the system to determine the possible escape paths.

The possibility to play stored messages allows the system to guide occupants of a building to the exit. One way consists in playing the same message with a short delay on several speakers. A person asked to evacuate would recognize that he or she has to follow the sound and go in the direction of the speaker where the message was played last. Another way consists in sending partial sentences on different speakers like for example: "Go (1) - this (2) - way (3)" guiding people in the direction from the speaker which emitted "Go" (1) to the speaker which emitted "way"(3). A combination of strobe lights of different colors or variable colors or bicolor strobe lights as well as strobe lights visible from different and distinguishable angles can also be used in combination with an audio message like for example "follow the green light". A person looking at the notification device and seeing a green light would know that this is the right evacuation direction whereas a person facing the other side would see a red light indicating that this is the wrong direction. Fig. 3) shows a schematic drawing of a notification device with two strobe lights on each side, Fig 4) a notification device with two bicolour strobe lights and Fig 5) a notification device with two strobe lights pointing to four different quadrants.

The ability to adjust the volume is important not only at the commissioning but also during normal operation because the ambient noise may vary considerably during the day. The volume should be as high as needed to ensure intelligibility, but it should not

unnecessarily annoy people. A simple paging message doesn't need to be as loud as an emergency call. The sound level can be adjusted on the basis of a schedule, by analyzing the ambient noise with the help of the onboard microphone and should also be dependent on the urgency and importance of the message.

The head end shown in Fig. 6b) of this evacuation system has to supply the power to each line and to couple the data into the power lines. Also, short circuit protection and over voltage protection for each line is needed. The attached microphone allows to page into the building; the touch screen and the switch board allow control of the system and select areas, zones, or individual notification devices.

Figure 7-10 show the effect on the power demand of the system if the method of operating the system according to the invention is applied. The curves show the power demand over time when the following text is played: "May I have your attention please (pause), may I have your attention please (pause) there has been a fire alarm" (the text goes on here, only a 6 second time windows is presented). The power demand of the individual notification devices as shown in figure 7, are for notification devices operated with time spreading but almost no local energy buffering. The first five curves show the power demand of five notification devices, which play the above message with a small incremental delay (around 300ms) relative to one another. The last curve shows the average power demand of the five notification devices. Each curve is normalized to the peak magnitude, which corresponds to full scale.

Compared to conventional systems, the ratio of peak to average power demand in figure 7 is now reduced mainly due to the time spreading. The average power still contains some peaks, but they are typically reduced by a factor of 5 because the system contains five speakers. The reduction is for example particularly noticeable at 570/100 or 5.7 seconds.

Figure 8 shows curves for the power demand for the same notification devices playing the same message with time spreading and local energy buffering of 0.01 seconds. The ratio of peak to average power demand is reduced compared to that of figure 7 due to the energy buffering.

Figure 9 shows curves for the power demand for the same notification devices playing the same message with time spreading and local energy buffering of 0.1 seconds. The ratio of peak to average power has further decreased.

Figure 10 shows curves for the power demand for the same notification devices playing the same message with time spreading and local energy buffering of 1 second. The ratio of peak to average power has further decreased, where this effect is mainly due to the energy buffering and the time spreading has comparatively small effect.

Claims

1) System having a head end (HE) and a plurality of notification devices (ND) each having a loud speaker (6) to emit audio signals

characterized in that

each notification device (ND) comprises an amplifier (3) configured to drive its loud speaker (6),

and each notification device (ND) further comprises a controller (1) configured to process command and audio data,

where the controller (1) consists of a microprocessor having a volatile (RAM) and the controller comprises a computing speed of at least 5 mips (million instructions per second) during an audio operation,

and the system comprises a digital data communication for command and audio data between each notification device (ND) and the head end (HE),

wherein the system is configured for exclusive digital data communication with base-band encoding,

and the system allows the use of unshielded cables and comprises a wiring architecture that has multi-drop capability.

2) System according to claim 1

characterized in that

each notification device (ND) comprises a microphone (7). 3) System according to claim 1 or 2

characterized in that

each notification device (ND) comprises a flash memory (5) with a capacity of at least 100 KByte. 4) System according to one of the preceding claims 1-3

characterized in that

each notification device (ND) comprises one or more strobe lights of different colors or variable colors or one or more bicolor strobe lights. 5) System according to claim 4

characterized in that the one or more strobe lights are configured to emit light in different and distinguishable directions.

6) System according to one of the preceding claims 1-5

characterized in that

the notification device (ND) is configured to allow the current to flow from the device (ND) to the power line if the line voltage is lower than the voltage of the energy buffer of the notification device (ND). 7) System according to one of the preceding claims 1-6

characterized in that

each notification device (ND) comprises an energy buffer with a capacity of more than 2 times the average power times 0.2 second. 8) Method to operate a system according to one of the preceding claims 1-7 characterized in that

voice messages are played on speakers (6) of several notification devices (ND) of the system, where each voice message is played with a time delay relative to the message played on another speaker.

9) Method according to claim 8

characterized in that

on each notification device a different message may be played simultaneously. 10) Method according to claim 8

characterized in that

live voice is individually delayed by means of a temporary storage in a volatile memory.

11) Method according to claim 8

characterized in that

in order to show the direction of an egress path, messages are sequentially played on the notification devices (ND) along the egress path.

12) Method according to claim 8

characterized in that in order to show the direction of an egress path, the strobe lights are sequentially activated on the notification devices (ND) along the egress path.

13) Method according to claim 8

characterized in that

the strobe lights of the several notification devices (ND) are operated to emit strobe light of a first color in a direction pointing away from a hazardous location and to emit a strobe of a second color in a direction pointing toward a hazardous location. 14) Method according to claim 8

characterized in that

an acoustic isolation between individual notification devices (ND) is determined by means of a microphone (7) detecting other notification devices (ND) in the vicinity playing acoustic signals on their speaker.

15) Method to operate a system according to one of the claims 2-7

characterized in that

the microphone (7) of a notification device (ND) is used to test the function of the speaker of the notification device (ND).

16) Method according to claim 15

characterized in that

the microphone (7) is used as an acoustic trigger by a technician. 17) Method according to claim 15

characterized in that

the microphone (7) is used for a two-way conversation.

18) Notification device (ND) for a notifying system, the notification device (ND) having a loud speaker (6)

characterized in that

notification device (ND) comprises an amplifier (3) for the loud speaker (6) and a controller (1) configured to process command and audio data,

where the controller (1) consists of a microprocessor having a volatile (RAM) and comprises computing speed of at least 5 mips (million instructions per second) during an audio operation. 19) Notification device (ND) according to claim 18

characterized in that

the notification device (ND) comprises a microphone (7).

20) Notification device (ND) according to claim 19

characterized in that

the notification device (ND) comprises

a non-volatile memory able to hold audio data of stored digital audio data and

configuration data,

an analogue to digital A/D converter that is integratable in the controller (1) and is configured to convert a microphone signal into a digital signal, and

a digital to analogue D/A converter that is integratable in the controller (1) and is configured to convert digital audio output of the controller (1) into an analogue signal suitable for the amplifier (3),

wherein the amplifier (3) is arranged and configured to receive an output signal of the D/A converter as a source or input signal or to receive direct digital data generated by the controller (1). 21) Notification device (ND) according to one of the claims 18-20

characterized in that

the notification device (ND) comprises a flash memory (5) with a capacity of at least 100 KByte. 22) Notification device (ND) according to one of the claims 18-21

characterized in that

the notification device (ND) comprises one or more strobe lights.