Every component in the design has been carefully considered with a view to obtaining maximum functionality at minimal cost. The most basic fire detection system has a build cost of around US$8.50 while the most sophisticated build (with features such as AWS and Azure compliant security, as well as NFC for maximal ease of configuration) has a build cost of around US$20.00.
We’ll discuss the various subsystems that could be implemented in the following section. Obviously not all combinations result in something useful, while other sections (such as the microcontroller) are compulsory. A vast range of options are possible within these parameters.
This is the central controller for the system. All subsystems connect to the microcontroller in order to have access to all the available facilities. The software running on the microcontroller determines what operations are performed by the system. This approach gives maximum flexibility, allowing many different sets of functionalities to be implemented by loading different software onto the device.
Software is written on a normal computer in a package called MPLABX. The package is supplied cost-free by the manufacturers of the microcontroller. A “device programmer” is used to connect the Sigthsense device to the host computer for downloading software. Once downloaded, the software will be resident on the Sigthsense indefinitely, staying resident for decades even with the battery removed. If new software is desired, it may be programmed onto the Sigthsense using the device programmer. This may be done tens of thousands of times without harm. Device programmers can be bought from the manufacturer of the microcontroller for as little as US$15.00 for the most basic version, and up to several hundreds of dollars for the most professional tools.
The microcontroller used in this system is either a PIC32MM0064GPL028 or a PIC32MM0032GPL028 depending on the complexity of the software needed for the particular feature set This is a 32-bit computer packed into a microcontroller along with 8kB of RAM and either 32kB or 64kB of flash memory for storing software. This device was chosen for its combination of features, price and low power consumption (an important consideration if the non-rechargeable battery option is chosen).
The battery system powers the rest of the system. Batteries can be either non-rechargeable (called primary cells) or rechargeable (called secondary cells). This design platform offers a choice of either – the selection of a battery will depend on the rest of the selected options.
If the build only features fire or moisture detection, possibly only connecting to the WiFi when an alarm is sent, the primary cell option is better from an installation convenience point of view.
If lighting or Bluetooth or possibly WiFi (depending on what it’s used for) are added to the system, the energy demands are such that a secondary cell is a better option from a cost-of-ownership point of view. If a secondary cell is used, it will need a power source from which to charge. Two possibilities are provided.
- The first is a small solar panel. The panel can be glued to an appropriate surface to prevent theft. This will provide years of maintenance-free operation with no running costs for situations where mains power is unreliable, unavailable or unaffordable.
- In situations where mains power is available the system also features a micro-USB plug which allows a cellphone charger to be used as a power source. Because the system has a battery the mains power does not have to be present all the time. This allows the system to be used where power is intermittent, such as in back-up lighting applications.
A thermistor is a simple temperature detector. It acts like a resistor which changes its resistance depending on its temperature. This resistance is converted into a voltage and read in by the microcontroller. Once the temperature is known to the microcontroller the software on the microcontroller can process this data in many ways. A cardinal way of doing this is to examine how fast the temperature is changing. Under normal conditions temperature tends to be a slowly changing parameter, but in the presence of fire it changes rapidly. This is a well-known fire detection technique and standards for it, as well as allowable rates of change, are laid out in the EN 54 standard.
The thermistor also adds other functional possibilities to the platform. It can be used to take temperature readings which can be read out through one of the wireless interfaces as a thermometer. The temperature readings could be stored into the memory of the microcontroller which would allow a log of temperatures to be captured. This could be read out remotely to allow detailed analysis of an environment’s temperature trends. Thresholds could be set causing alarms to go off, either on the device or remotely, should the thresholds be exceeded.
The physical size of the thermistor and its placement in the device are important. A smaller thermistor allows faster changes in temperature to be captured (important for fire detection); a larger thermistor would smooth out short-term fluctuations in temperature. In this design a small thermistor was chosen. It is possible to use software running on the microcontroller to smooth out short temperature fluctuations if needed, so the smaller thermistor is a more versatile option. The thermistor is placed right next to a vent in the casing. This allows maximal heat transfer by means of convection, resulting in the fastest response time.
The moisture/contact sensor circuit is a dual purpose input to the system. It can be used either to detect moisture (flooding as an example), or it can detect a mechanical contact, a switch, possibly for an alarm application.
In the case of moisture sensing a wire would run down to a small spike with two electrodes which gets put where a potential flood would occur and when water touches the electrodes it causes their resistance to drop, triggering an alert. The spike might be pushed into the ground to detect rain or the ingress of water under a wall.
An example of a contact input would be a typical burglar alarm system which often features a magnetically operated switch mounted on a door or window frame. The door or window would have a corresponding magnet – when the door or window is opened the magnetic switch is activated and an alarm is sounded.
Passive infrared sensor
The system provides a basic burglar alarm function by means of a passive infrared motion sensor. This sensor responds to the movement of warm objects and is effective for small unoccupied spaces, or larger spaces if strategically positioned.
The sensor detects the motion of humans close to it and therefore the alarm must be disarmed remotely. The Bluetooth interface is useful for this purpose.
An important part of an alarm system is a siren of sufficient volume to alert people who are close by. This system features a siren capable of 95dBa at 1m. It is pitched at 2700Hz to be well within the audible range of even elderly people. The volume is sufficient to wake a soundly sleeping person several metres away.
The base board has a single push-button for user input. In cases where the mezzanine board is not implemented this would be the user’s sole button input. It is chosen for low cost and its function is defined in software. It is intended for a basic function such as muting an alarm or turning the light on or off.
The mezzanine board has three capacitive touch-buttons that can be used for controlling the system should the single button on the base board be insufficient. These buttons detect the proximity of a finger and respond. The buttons detect through the casing in order to simplify the mechanical design of the system. The touch-sensitive positions on the casing are marked by means of printing or decals. These double up to indicate the function of the button.
As an example of a system with three touch-buttons and one push-button, features could be “Light On/Off”, “Bluetooth Pair Request”, “Reset Alarm” and the push-button, which is less likely to be pushed accidentally, could be a “Panic-button”.
Candles are still an everyday source of light in poverty-stricken settings despite their cost and the fire risk they hold. A typical “taper” style candle produces approximately ten lumens of illumination. The Sigthsense system has a built-in LED that can provide 100 lumens. The LED is fairly power hungry compared to the rest of the system and must be run in conjunction with either the solar or mains input. If solar input is used, the lighting is essentially free once the unit has been procured, resulting in substantial savings for the user over the lifetime of the device. The LED is fairly power hungry compared to the rest of the system and must be run in conjunction with either the solar or mains input. If solar input is used, the lighting is essentially free once the unit has been procured, resulting in substantial savings for the user over the lifetime of the device.
It is critical that the light is not allowed to run the battery down too far if the system is used both for illumination and alarm functionality. This would compromise the alarm functionality. For this reason, the system provides circuitry to enable the microcontroller to assess the battery state and act accordingly by shutting down or dimming the LED. A courtesy warning may also be given to the user by means of audible output or blinking the LED.
The LED is dimmable. This option is provided because dimming the LED provides the ability to trade running time off against light output. The solar input system is designed to run the LED for around four hours per day at full brightness off the rechargeable battery (assuming there are three usable hours of full sunlight per day). This is typical of winter in many parts of the world.
The LED may also be flashed as a supplement to the siren under alarm conditions.
The mezzanine board has five LEDs to indicate system status. If the mezzanine board is not used, one LED can be assembled onto the base board to provide more basic indication. This is important because the EN54 fire detector standard specifically requires a red light to indicate a fire condition.
The five LEDs on the mezzanine board are allocated as follows:
- A white LED to indicate that the battery is charging
- A green LED to indicate that the battery is fully charged
- A yellow LED to indicate that a Bluetooth connection is active
- A red LED which will indicate a fire alarm condition
- A blue LED which is software defined and might indicate operations such as WiFi activity
3.The LEDs are positioned around one of the capacitive touch-buttons. The circuit board is arranged in this way so that the button may be found easily in the dark.
The WiFi module is used to connect the system to a wireless internet provider. WiFi has become a ubiquitous method of providing internet access to a huge range of internet-connected devices and is widely available from the wealthiest homes down to the poorest informal settlements. WiFi infrastructure is inexpensive and the equipment is widely available. WiFi was chosen in preference to a proprietary network because it offers much better prospects of future availability, unlike some proprietary solutions which become unsupportable when an individual manufacturer ceases production of a specific part. We also chose WiFi instead of longer range solutions such as LoRaWAN or Sigfox because of the much lower purchase price of the WiFi modules. This results in substantial savings where either the WiFi infrastructure is available or the density of devices is high.
WiFi can also be used for geolocation. Providers such as Google offer a service whereby the globally unique MAC addresses of nearby WiFi access points are submitted and the approximate location is returned. This is not as accurate as a satellite-based navigation system, but is sufficient for many purposes.
Bluetooth is a very popular option for connecting wireless devices to smartphones and computers and has been included to open up the field of smartphone connectivity and the range of available options this can offer. Bluetooth offers a reliable short range link through which the system can be controlled, or system data retrieved. In applications where the system is used as an intruder detector, it can be remotely armed or disarmed via the Bluetooth link, much like a wireless version of the control panels that conventional alarms incorporate. In addition, comfort features such as controlling the illumination provided by the system may be implemented.
Bluetooth connectivity requires a PIN code that may be used to prevent unauthorised people from operating the system. This makes it a viable option in security applications. In addition, the name of the device as shown on the user’s smartphone may be customised by the user for convenience when multiple devices are within range.
Systems which result in a real-world response – such as the dispatching of response personnel – are particularly vulnerable to online attack by malicious third parties. Online systems often counter this threat by means of cryptographic authentication, essentially allowing those systems to prove their authenticity. Because of the scale of this threat many operators of online systems mandate the usage of cryptographic authentication. This has been provided in our system.
There is a selection of security variants that may be implemented by assembling different components on the circuit board. Which version is chosen will depend on the requirements of the online service provider used. Microchip Technology’s ATSHA204A and ATECC608A are popular options that may be implemented.
Near Field Communication
Near Field Communication (NFC) allows an NFC enabled smartphone to communicate with the device by tapping the phone on the device. NFC tagging allows the phone to download settings to the system (e.g. those needed for WiFi or Bluetooth) without the need for users to enter them directly. This is a practical option in a less technically literate environment. NFC also allows the system to direct a smartphone directly to a web page or to launch an application.
The NFC integrated circuit contains a block of memory that can be used for storing settings such as logins and passwords. In cases where cost-saving is desired, and the smartphone interaction is not required, these settings may be stored in a section of the microcontroller’s flash memory.