It was evening, there was nothing to do... we decided
I decided to tell you a little about my experience in creating a multifunctional device for protecting ATMs. The article is more of a narrative rather than a technical nature, but I think it will be interesting for someone to read and look at the photographs. The article deliberately does not contain all possible names of actors/devices/firms.
Description of the problem
In our country we have gotten into the habit of blowing up ATMs with gas. So much so that after this the room resembles a military training ground.
Here are some photos from the explosion in the city of Vyazma, Smolensk region in March 2016.
The method, I must admit, is interesting: the door of the safe is knocked out from the inside by a blast wave, but the cassettes with the money remain intact - grab it and run. The funniest incident was in Ufa in June 2016. A thief came to a bank branch with balloons filled with gas. And laughter and sin...
Video from surveillance cameras:
What is currently available
At the moment, most ATMs are equipped with vibration security detectors that are triggered by mechanical influences: shock, tilt, vibration. Try to hit the ATM with your foot 3-5 times, the show masks will immediately appear.I specifically went to the 24th international exhibition of security equipment Securika in St. Petersburg in November 2015 to find out what was on the market in this area.
Offtopic:
One of the devices inside looks like this:
A fairly simple device with a digital accelerometer, a piezo vibration sensor with two amplification stages and a PIC16 microcontroller.
Its older model (picked from the ATM wall):
What is necessary
We need to measure the presence of gas in the ATM safe. Well, don’t forget about tilt/vibration; a bank won’t install several different security devices in one ATM. There was also the need to react to internal smoke, because a gas cutter smokes quite well, but we must protect ourselves from everything.There is also an option for protection against explosions, and not just an alarm: place a cylinder with a phlegmatizer (a substance that counteracts the explosion of a gas-air mixture) in the ATM and send a control signal from the gas sensor to the valve. If the device senses gas, it opens the cylinder valve, the phlegmatizer is released into the ATM volume and ignition of the mixture becomes impossible. This option is very expensive, but, if desired, we can implement it by installing additional equipment.
Suggested Solution
Build a multifunctional device that will have sensors:- flammable gas (based on the thermochemical measurement principle);
- tilt (digital accelerometer);
- smoke (optical camera);
- vibrations (piezo plate on a metal pig);
- separation from the protected surface (limit switch);
- opening the device body (limit switch).
... and all kinds of switches for issuing a control signal to the control panel (most often to open 12 V).
Gas sensor
I admit, at first we tried to assemble the device using the Chinese MQ-2 semiconductor sensor, so beloved by Arduino people for its simplicity. I'm ashamed, but the decision was made above. There was little hope of combining gas and smoke in one sensor, since it was stated to be sensitive to smoke. But we quickly abandoned it: we were unable to achieve stability in the readings; it floated beyond the required errors. We switched to the familiar thermochemistry.Information from Wiki
The operation of the thermochemical sensor is based on the fact that when a gas-air mixture passes on the surface of the catalyst, combustion occurs and the generated heat increases the temperature of the ball. The resulting increase in the resistance of the platinum coil is registered by a bridge circuit, the second arm of which does not have a shell - a catalyst. At low concentrations, the change in resistance is directly dependent on the gas concentration in the environment. The typical voltage on the sensor is several volts, the current is 0.1-0.3 amperes.
Thermochemistry in terms of circuit design is also not particularly demanding: a controlled voltage stabilizer on the op-amp with the ability to provide a smooth start in 40 seconds and a differential amplifier to amplify the signal from the sensor bridge.
Accelerometer
In the comments to my yesterday's article on GT about the production of printed circuit boards, they discussed the possibility of mounting small cases on wires... I just mentioned this particular case there: we had to turn the accelerometer on its back and solder it with a thin MGTF wire. Then hot glue was dripped on top and successfully submitted for testing. The accelerometer was taken in an LGA-14 case (2x2x1mm), you can estimate 2mm using a ruler, and what it was like.
When installed on the wall of an ATM safe, the device remembers its initial position and subsequently calculates the deflection angle from it. If the specified threshold is exceeded, a control signal is issued to the relay. The accelerometer in this case is used only to determine the angle of inclination. Of course, we tried to remove the vibration from it, but this did not bring much success.
Smoke detector
An optical camera of this design was used as a smoke sensor. The principle of operation is the same as in many smoke detectors: an infrared LED shines into the center, the light on smoke/dust particles is scattered (I carefully drew the dust in Paint) and a certain amount falls on the receiving photodiode. The intensity of smoke/dust is proportional to the signal from the photodiode.True, for a long time we could not find an equivalent optical density. After all, it is impossible to use a conventional optical measure, the optical density of which is normalized along the direct path of the optical signal (we had these available), since the LED and photodiode in the camera are located at an angle of 120 degrees to each other. Has anyone encountered such fire detectors? How are they configured?
For calibration, I had to use a real smoke meter and set fire to rags in the smoke chamber. Thus, we selected a set of equivalent optical measures and adjusted the sensor response threshold. True, I stunk quite a bit back then, and it was a shame to wait in line at the cafeteria later...
So that the article wouldn’t turn out to be too heavy, I divided it into two parts.
