Summary: This version had a lot of manufacturing problems that prevented it from being functional. The design could have been simpler to achieve its functionality.
System Goals: This system was only intended to operate during daylight hours and log when when bees would cross its infrared sensors.
Problem 1: Reflow of Solar Cells
It was very difficult to assemble the solar cells because I assembled the bottom side first (where the MPPT circuitry was).
As you can see, the heat damaged the solar cells which prevented power from being generated.
Problem 2: Incorrect Packaging of IR Sensors
The design called for the which is surface mount component but I assembled the QRE1113 which is the through hole version. This required bending of the pins and made soldering very difficult. I could not source the QRE113GR (SMD version) so I would recommend putting further work into finding another paired or testing a matching pair of IR LEDs and photodiodes (). Problem 3: Untested Power Stage
The is a popular MPPT IC but its operation was never verified in the design (because of the faulty solar cells). This might not be a problem because in hindsight a standard LDO or buck converter may have been a better option if the intention is to only operate the device during daylight hours. This would save $8 from the BoM cost due to the high cost of the and reduce the price down to a few pennies.
Problem 4: Particle Argon
The particle Argon is a great device for starting up a few hundred instances of an IoT product but their pricing was quite aggressive for a consumer product. It definitely made more sense for industrial and agricultural customers who benefit greatly from the data collected, but we wanted to move away from this pricing model for our application.
Improvement Summary: This design could accomplish its task better if:
reflow-ability of the solar cells was considered in the manufacturing process (use of an infared oven or an assembly service like JLCPCB or PCBWay). a reliable source of infared sensors were tested with objects passing through the mechanical opening. a simpler power conversion stage was used. May I suggest . The wide input range allows you to play around with various solar cell configurations. The only thing to watch out for is the maximum input voltage of your power converter and the VOC of the solar cells. (Also don’t forget to add a schottky diode (or other low forward voltage diode) to the output of the solar cells.) the use of a wireless modem that supports your project requirements was selected. A particle Argon would drastically increase the BoM cost. Particle sells lower cost modules for production, but there are other modems on the horizon, such as the ESP32-C6 which supports Zigbee/Thread (not much firmware support as of Fall 2023).
Summary: The second iteration of the project abandons the original architecture completely because the project requirements drastically changed. In this project, we are relying solely on the ESP32-CAM to take pictures of the bees at a deterministic rate.
The concept for this design is a house enclosure that contains some test tubes for bees to nest in. The image above shows a native bee building its nest in there. The designer of this project mentioned that it’s not a good idea to check on the bees too often because it exposes the eggs to sunlight. We wanted to automate the checking process by embedding a wireless camera which stores images of the tubes and detects when something is inside any of the tubes.
Design Inspiration 2:
started stocking Voltaic’s Solar Cell lineup. These solar cells are quite robust and have undergone extensive stress testing by Voltaic; the tests can be found on Voltaic’s website. I can attest to their survivability from my cats playing with it a lot. To keep the BoM cost low, The was tested to see if it could successfully charge a battery. To get some idea of the testing, an 18650 3000mAH battery was charged from 3.8V to 4.0V over the course of 2 hours when the solar panel was under direct sunlight.
Design Inspiration 3:
The $3is actively being programmed (cheaper on aliexpress). The library is being used to connect the device to the consumer’s WiFi network. The ESP32-CAM will take photos of the bee nest once every hour. The architecture involves storing images on the SD card and users being able to access stored photos via local FTP. An application will handle the synchronization of these photos and run detection algorithms on the phone.
Design Problem 1:
While charging the project in the sun, the battery overheated and released a lot of fumes, stinking up my apartment. After a few minutes of cooling, its temperature was measured to be 40C. I presume it hit temperatures of above 60C. I also saw cats playing with the device so it is unclear if the cats caused a temporary short circuit or if the battery simply overheated in the sun. In future designs, thermal shielding of the battery with the chassis is recommended and air holes are recommended for outdoor wind cooling. Also these batteries were sourced on Aliexpress and a more reliable manufacturer that garuntees a wide temperature operating range should be selected.