Methods for identifying the causes of false alarms.
False alarms are the most unpleasant drawback that a fire alarm system can have. Unfortunately, nowhere in the advertising materials will you find any parameters that allow you to estimate the probability of false alarms. Even worse is that any, no matter how wonderful, equipment can be a victim of poor installation, exposure to time or interference. Therefore, installers and especially operators must know the possible causes of false alarms and be able to look for them.
The most common cause of false alarms is poor contact in the alarm loop. It is not for nothing that electronics is jokingly called the science of contacts: their absence where they are needed, and their presence where they should not be. Twists, cheap steel terminal blocks, single-core wires that break — and here you have it, in a year or two, the contact begins to disappear. A very unpleasant malfunction, depending on the temperature or humidity, it may not appear for months, but will come to the surface, for example, at minus 30 outside, so that it is «more pleasant» to look for it. Or it will appear at night, and during the day the repairman comes — everything is fine, everything works. Such a malfunction is very difficult to detect and eliminate.
Often the cause is electromagnetic interference. Moreover, interference can affect both the control panel and (more often) the sensors (detectors) themselves. This problem is typical for fire smoke detectors installed on a suspended ceiling. In this case, the cable of the loop often simply lies on the ceiling frame, mixed with lighting cables. And the gas-discharge lamps themselves with high-frequency (throttleless) ballasts are often a source of terrible interference, and they are located very close to the fire detectors.
The third most common reason is installation errors. In this case, I do not mean poor wiring, but rather poor mechanical installation of devices.
For example, the reed switch is installed crookedly, the magnet has become slightly demagnetized over time, the wooden door has dried out and become warped, and now the reed switch honestly gives the signal «door is open». Press harder — normal, pull the locked door slightly — alarm. Most reed switches have a reliable response distance of only 1-2 cm. Such a malfunction can be easily detected if you glue a magnet to the reed switch (do not forget that you have effectively disabled the reed switch — it has stopped detecting the opening of the door). If false alarms stop during the test, then this is the problem, more carefully mount the reed switch and the counterpart (magnet) on the door or replace the reed switch with a longer-range one.
By the way, the opposite malfunction is also not uncommon: the reed switch stops signaling that the door is open. This happens on steel doors if the frame itself is sufficiently magnetized.
In addition to reed switches, poor installation can also affect, for example, infrared motion sensors. The sensor hangs on one screw and sways from slamming doors in neighboring rooms. And there is a heating battery in its field of vision. If the sensor were firmly fixed, the battery would not interfere with it. But now — here are your false alarms.
In general, infrared sensors are easy to install incorrectly — opposite a window or a radiator. Theoretically, it will still work, but a window flapping in the wind or a fluttering curtain objectively ensures a rapid change in the temperature distribution in the sensor's field of view. This cannot even be called a false alarm — the sensor honestly records the movement of something warm against a cold background. Similarly, an acoustic glass break sensor can objectively react to a very strong sharp sound (practically any can be set off by clapping your hands directly in front of it). Do not unconditionally believe what they say and write about complex spectral analysis. Yes, computer programs can very accurately distinguish sounds. But for serial sensors to be able to distinguish the sound of glass from other similar sounds so well, they must also have a Pentium of several gigahertz. True, they would then consume as much power as a computer and cost the same. That's why I don't even consider the glass break sensor in the dining room, where knives are constantly dropped on the tiles, to be false alarms. If this is a problem for you, turn down the sensitivity. Or put the sensor behind the curtains near the window — then it will hear the sound of breaking glass well and will not hear the sounds of the New Year's corporate party from the room.
Now let's look at how to search for and fix the malfunction. The main principle: the source of false alarms must first be localized. This is not easy, false alarms, as already mentioned, can occur quite rarely (but often enough to make the customer nervous). You arrive at the site, tighten all the screws in the connections, check the integrity of the wires, even ring the loop with a tester (ohmmeter) and make sure that everything seems to be normal, and a week later you are again told that there were two false alarms. Well, it's time to tackle the problem systematically.
The first question: do false alarms always occur in one loop or in different ones? If the control panel has a good event log and you can view it, great. If not, you will have to negotiate with the guards on duty so that they record when and which light was on during an alarm. How to negotiate is not a question for me. If you do not know how, read about the art of getting along with people or other similar opuses. As a result, you will find out where alarms occur and when. Sometimes it is possible to correlate the alarm time with the switching on of, for example, industrial equipment — this means that the problem is in electromagnetic interference and, according to the manufacturer's recommendations, it is necessary to shield, ground or, conversely, power it from separate power sources. Discuss countermeasures with the system developer, they will not be happy, but they will advise something. Or you can simply replace the faulty detectors with other types (for example, smoke with heat) — this can also help.
If false alarms occur more or less evenly in all loops, the problem is probably with the control panel. Replace it, preferably with a different model. If that doesn't help, consider that the system is simply neglected as a whole (or that the detectors are equally low-quality everywhere), and begin to fight each loop in turn (if there are many loops in the system, then it is better to fight several at once). During such a fight, parts of the system are switched off for a while and the safety of the facility is reduced, so do not forget to coordinate this with the person responsible for security. It may even be necessary to temporarily deploy a backup system, for example, a radio channel, it is easier to quickly install and then dismantle.
So, troubleshooting a separate cable. The only scientific method is the method of dividing in half. You break the cable in the middle, move the end-of-line resistor there (or better yet, install a new end-of-line resistor) and wait for some time. If false alarms used to occur about once a week, you need to wait about a month. No false alarms — the problem is in the cut-off piece of the cable. We connect it back and cut this piece in the middle, so that now the cable remains connected.
If there were false alarms at the first stage, then the problem is on the connected part (the cut off piece may also have problems, but we will try to catch at least one by the tail first). We divide the nearest piece in half again (the loop remains connected) and wait again.
And so on until we find a specific sensor that is giving false alarms. Note: if, for example, you have electromagnetic interference and false alarms are given uniformly by all sensors, then as you cut off pieces of the loop, alarms will occur less and less often. If this is the case, increase the exposure time. The whole epic, if false alarms are not very frequent, and the loops have many sensors on each, can stretch out for months.
The second method is to replace the equipment. It is especially appropriate if there are many false alarms on different loops. Choose one of the loops and replace all the sensors on it with the most reliable and expensive ones you can afford. For one loop, this is usually not that expensive. Although it is very labor-intensive and often ugly in terms of the cheapest ones — reed switch security sensors. If it helps, then in the case of a security loop with different types of sensors, you can gradually put back different types of sensors and thus find out which sensors are the problem. It is more difficult with fire alarms — there, the entire loop usually consists of the same sensors, and if replacing them with good ones helps, it means that they were all bad before. It is not that they were all hopelessly bad. Perhaps in other situations they can work, but specifically in yours, at this facility, they are unsuitable.
In the case of fire detectors, there is another reason: cheap products can have a very wide range of parameters. Half of them, for example, are quite resistant to interference, while some are triggered, as they say, by a sidelong glance. If it is economically justified, you can gradually, several at a time, put back the old detectors. Perhaps you will be able to select those that do not give false alarms.
Addressable systems are a special case. Of course, addressable detectors are usually more expensive and of higher quality than conventional ones. But there are no ideal products. In many cases, they can also give false alarms. But it makes troubleshooting much easier. Firstly, you don’t have to bother with dividing the loop in half, you initially know which detectors are giving false alarms. This will save you several months. Secondly, all the addressable systems I know have good event logging tools, so you can get information accurate to the minute or even second when false alarms occurred. Finally, addressable detectors often provide the ability to perform detailed diagnostics or adjust their parameters. You can change some parameters, or at least just make the sensitivity rougher. I won’t give any specific recommendations, it all depends on the types of devices.
In general, troubleshooting an addressable system is much more pleasant than in a non-addressable one. Instead of running around the site with a stepladder and tools, most operations can be performed from the system control panel. However, the same labor-intensive and time-consuming method of dividing in half may also be required in an addressable system. This is usually necessary if the problem is an irregular loss of connection with individual detectors. If the problem is a poor contact (a break in the loop), then the location of the damage to the loop can be determined by analyzing which detectors the connection is lost with and with which it is always stable. If the cause is a short circuit in the communication line, then you will have to divide in half. However, even in this case, the situation is easier than in a non-addressable one. When dividing in half, it is not necessary to completely disconnect the rest of the loop; it is enough to insert one or more short-circuit isolators. When the short circuit makes itself known, it will disconnect the damaged section, and you will know where to look for the problem.
In conclusion, we will describe recommendations for combating electromagnetic interference. This activity is not so much a science as an art. Some consider it shamanism. Indeed, in complex systems consisting of hundreds of products connected by kilometers of cable and located among many other electrical installations, it is simply impossible to accurately calculate the influence of one device on another. The same actions in one case may help, in another only worsen the situation. But there are general principles that should be understood so as not to try all possible combinations by trial and error.
The first recommendation from manufacturers of all systems is to use a shielded cable. Yes, it often helps. Although in an existing system, replacing an already laid cable with a shielded one is usually almost impossible. Nevertheless, let's consider some details. The shield on the cable itself can help a lot. Even if it is not connected anywhere. Often, it is even the best solution — to leave the cable shield unconnected. In any case, the shield equalizes the effect of interference on all wires in the cable, and therefore the differential interference signals applied to the devices are reduced. In no case should the shield be grounded (or connected anywhere at all) at both ends. Because in this case, the shield becomes not a shield, but an additional conductor through which an unpredictable current flows. This is called a ground loop, more on that below. Often, the best solution is to ground or zero the shield on the side of the control panel. It is the control panel that receives the signal from the loop, and if the shield is connected to the reference point inside the control panel, then the interference on all cable cores relative to this point will be minimal. Depending on the circuit design, grounding may not be optimal, but connecting, for example, to the PPC case, to the negative power supply wire of the PPC, or even to the negative wire of the loop. By the way, the PPC case, if it is metal, should, in theory, be grounded. But in practice, if the ground (the third wire in the power supply network) is not very high-quality (itself contains a lot of interference), it may turn out that it is better not to connect anywhere than to such ground.
In addition to cable shielding, shielding of the detector subject to interference is sometimes used. A sheet of copper foil or galvanized sheet metal is placed under the detector on the side of the supposed source of interference (for example, if there is an elevator motor or a milling machine behind the wall). Aluminum foil from a chocolate bar is ineffective, because it has a fairly low conductivity. It is often useful to connect such a screen to the minus of the detector's power supply with a separate, fairly thick wire.
Often, interference enters through unplanned contact. The worst thing is when one or more wires in the system are grounded in different places. That same ground loop mentioned above. Different points on the ground have very different potentials (the ground is not a very good conductor), as a result, a so-called equalizing current will flow along a wire grounded in several places. This can include a return current from a passing tram (in theory, it should flow along the rails, but if there is poor contact there, it will flow along your cable just fine) or a symmetrical current from a three-phase rolling mill motor. There are known cases when such a current evaporated poorly grounded cables and completely disabled equipment. The result, as a rule, is not so tragic, but the impact of interference increases many times over.
Please note: multiple grounding may occur against your will. For example, a loop laid with noodles was fastened with nails. A nail touched one of the wires and the grounded plaster mesh — and that's it, here is an unexpected secondary grounding point. In theory (according to GOST), all PPK are designed to operate with a leakage resistance in the loop of up to 50 or even 20 kOhm. But the possible influence of interference with such a leakage to the ground is unpredictable. Often, when checking loops, only the resistance and insulation between the wires are checked. Do not forget to check the leakage to the ground — from the point of view of interference, this is even more important. If the resistance to the ground is less than 1 Mohm, problems are very likely.
Another way for interference to penetrate is to lay the power line of the detectors and the signal line in different cables. This happens if remote detectors are connected to a separate power source located next to them. In this case, the interference induced on the power line and on the signal line is different, and this potential difference is applied to the detector. Again, in theory (more precisely, according to GOST), detectors should easily tolerate interference from the loop. But the possible interference is much more diverse than the test interference used during testing. Maybe everything will be fine, and maybe not.
By the way, a potential source of problems is a popular fire alarm loop. Such a loop can be a huge loop antenna, very susceptible to both magnetic and electric fields in a wide range. If the control panel does not provide sufficient insulation between the two ends of the loop (and many control panels do not insulate them at all), then if there is a suspicion of electromagnetic interference, you can try to break the loop. It may help.
Another source of interference is the power supply. Try disconnecting it. Completely, both wires. Let the system work on the battery for a while. If it helps, the false alarms stop — install an isolation transformer, stabilizer, online UPS — all these are possible ways to isolate yourself from interference coming from the power supply.
And finally, as a last resort, I can advise trying to split one large system into several smaller ones. Instead of one 48-loop device, install three 16-loop ones connected to different power supplies. Or divide one integrated system into several autonomous ones. Perhaps the problem is that the size of the system of directly connected devices has exceeded the permissible ones in this place. Again, if it helps, then subsequently, observing safety precautions, for example, with galvanic isolation of communication lines, you can reconnect the system into one. The main thing is to determine the source of the problem, then you can find a suitable solution.