| Component | Quantity | Price (SEK) |
|---|---|---|
| Cheerson CX-20 drone (pre-owned) | 1 | (2990) |
| Pycom expansionboard 3.1 (provided by LNU) | 1 | (189) |
| Pycom Lopy 4 (provided by LNU) | 1 | (465) |
| Pycom LoRa antenna (provided by LNU) | 1 | (130) |
| Adafruit AMG8833 IR Thermal camera | 1 | 366 |
| Buzzer | 1 | 38 |
| Junction box IP65 | 1 | 65 |
| 10kΩ resistor | 1 | 1 |
| 47kΩ resistor | 1 | 1 |
| GoPro camera mount (pre-owned) | 1 | (35) |
| Breadboard jumper wire male to male | 10 | 32 |
| Breadboard jumper wire female to female | 4 | 13 |
| Total | 516 SEK |
The drone we have used for this project is a Cheerson CX-20. It's quite a big drone but still affordable. We needed a drone with some lift capacity so that we could mount our device underneath. There is no data on how much it can carry but there are videos on Youtube where it flies with weights of 1.5kg. We aimed to make our device lightweight but still resistant to some liquid. The drone has pin connections underneath which suited our project. It has both regulated 5V and also 12V straight from the battery. We used these to power our device and also to monitor the drones battery level. It also has a GoPro camera mount that we used to attach the device.
To measure the battery level we had to scale the voltage from 12V down beneath 3.3V, which is the maximum tolerance for input on the LoPy pins. We solved this issue with a voltage divider. It's a simple circuit containing two resistors with calculated values.
VREF = 12.53 * (10000 / 47000 + 10000) = 2.198 V
PIN_voltage = PIN18 input / 1000
VBAT = PIN_voltage / (10000 / 47000 + 10000)
The sensor used to monitor the temperatures is a AMG8833. It has 64 pixels (8x8) that detects individual temperatures by IR with a frame rate of up to 10Hz. The temperatures is returned in arrays via I2C communication. According to the technical documentation it can detect temperatures in a range from 0 to 80°C, however, we discovered temperatures up to 157°C. The viewing angle is 60° and detection distance is up to 7 meters. Power consumption is 4.5mA.
Measured temperature will decrease with distance. The diagram below shows the result from a test that was performed with a set heat generation of 160°C. With the support of this data our belief is that the sensor should be able to detect a heat generation of 150°C from a distance of 4 meters by setting the alarm threshold at 40°C. Since paper self-ignites at about 185°C (https://www.dafo.se/Arkiv/Faktabank/Brandrisker-och-riskhantering/Brandteori/Varme/) this configuration should give an eligible function. It will also prevent false alarms when the drone is flying very near the surface since temperatures in Sweden rarely rises above 40°C. To support supervisors with plausible data we therefore calculate an estimated value to compensate for loss in distance. The multiply constant we used is 3.75 that we computed with data from diagram beneath.
150/40 = 3.75
With a viewing angle of 60° we calculate:
x = 4 * tan(30°) = 2,3 m
A = (2,3 * 2)² = 21,2 m²
To protect the electronics in case of bad weather or a crash we needed to fit it into an enclosement. We used a junction box with IP66 classing. The sensor and Pycom microcontroller fitted perfect inside. Since the sensor need clear vision, a hole was made in the bottom of the box. Small screws was used for attachment and then some hot glue was placed on the screws on the outside. The buzzer was hot glued on the outside. As mentioned earlier we needed to keep things lightweight in order to make the battery last longer and for the drone to be able to fly with the device attached. The drone was built to fly with a GoPro camera attached. GoPro cameras weighs between 100 and 200g with housing and SD-card depending on the model. Our device ended up with a weight of 186g.
