Takeo Tokunari HTMAA Fall 2021

<topbar style="display:none;"> <item><a href="../index.html">Home</a></item> <item><a href="../about_me/about_me.html">About</a></item> </topbar>  <style> h0 { font-family:; font-size: 30px; color: #414040; margin-top: 50px; margin-bottom: 6px; word-spacing: 5px; } </style> ##The final outcome <iframe width="800" height="315" src="https://www.youtube.com/embed/GLJhthO7fW0" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> * What does it do? => See the video above * Who's done what beforehand? Some precedents exist for Apis bees. For example, * This research identified temperature and some other environmental variables as predictors for bee’s activities: https://www-nature-com.ezp-prod1.hul.harvard.edu/articles/s41598-019-56352-8 * This research tries the 3D printed beehives: https://3dprint.com/113066/3d-printed-honeycomb/ But research for melipona bee genus has been very limited. * What did you design? Beehive with 3d-printed honey pots and sensors for melipona bee genus. See the summary below. * What materials and components were used? * Where did they come from? * How much did they cost? Answers are summarized below. Regarding the cost, $8 in addition to the material I used from the inventory at Harvard Shop. | Part/Materials | For what | From where | Price in $ | | ----------------- | -------------------------------- | ------------ | ------------------------ | | Filament | 3D printed honey pots | Harvard Shop | n/a (used the inventory) | | 12V-fan | Output device | Harvard Shop | n/a (used the inventory) | | 9V battery | Powering fan | Harvard Shop | n/a (used the inventory) | | 1.5 batteries x 4 | Powering ESP32 and input devices | Harvard Shop | n/a (used the inventory) | | Loadcell + HX711 | Weight measurement | Amazon.com | $8.00 | | Thermistor | Temperature sensing | Harvard Shop | n/a (used the inventory) | | ESP32 | PCB | Harvard Shop | n/a (used the inventory) | | Plywood boards | Bee box | Harvard Shop | n/a (used the inventory) | | Total | | | $8.00 | For future, I estimated the entire cost with the full feature I would like to load: | Part/Materials | For what | From where | Price in $ | | ----------------- | --------------------------------- | ---------- | ---------- | | Filament | 3D printed honey pots | Amazon.com | $8 worth | | 3.3V-fan | Output device | Amazon.com | $2.4 | | 1.5 batteries x 4 | Powering ESP32 and input devices | Amazon.com | $1 | | Loadcell + HX711 | Weight measurement | Amazon.com | $8 | | BME280 | Temperature and humidity sensing | Amazon.com | $10 | | Gas sensor | CO2 sensing | Amazon.com | $2.3 | | Microphone | Activity sensing | Amazon.com | $3.3 | | ESP32 | PCB | Amazon.com | $10 | | Plywood boards | Bee box | Amazon.com | $11 | | Total | | | $56.00 | * What parts and systems were made? * What processes were used? * What questions were answered? * What worked? What didn't? See the detailed documentation below. * How was it evaluated? I had two meliponiculture practitioners, one in Brazil and one in Mexico, watch the video above for their comments. Both of them said the wood has to be thicker or another material such as recycled plastic or Styrofoam to be considered. Otherwise, they are excited to test my prototype. :) * What are the implications? I probably should focus on making sensors and 3d-printed hives before inventing the box. This way, they can simply insert my inventions to the box they are currently using and see the difference (if any) that my interventions bring. **The context and making process are documented below.** <img src="./peabiru.png" alt="photo_link_missing" width="800"/> **A Brazilian local NGO, Instituto Peabiru, explaining how to do beekeeping to the local communities in the Amazon** ##Background For the final project, I decided to make a bee box. It is not a bee box for the common apis bees that we normally see, but for the stingless bees (also known as melipona bees), a genus native to tropical rainforests in the Amazon, Yucatan Peninsula, Tanzania, etc. In many parts of the world, indigenous groups of people traditionally harvested the honey of stingless bees and used it as drug for wound, eye disease etc. In the past few decades, some local communities and NGOs have been trying to make this beekeeping (called “meliponiculture” as oppose to apiculture) as a means of earning income that does not require deforestation. Despite the extraordinary health benefits and the importance to the bio-cultural diversity and crop pollination, scientific research and technological intervention on beekeeping practice have been limited to date. <img src="./yucatan.png" alt="photo_link_missing" width="800"/> **Field visits I conducted in Spring 2021 in Yucatan Peninsula, Mexico** ##Objective To help make meliponiculture more productive, and to do so, embed sensors for data collection and output devices such as fan. ##Design iteration I started with the random sketch below: <img src="./scribble.png" alt="photo_link_missing" width="500"/> As documented in Week 0, initial ideas included the followings based on literature review on scientific papers and commercial precedents. * Temperature and humidity sensors: in case of apiculture, temperature and humidity are two measure factors that contribute to the yield (e.g. if the inner temperature is too high, bees would hover near the entrance to send fresh air to the hive). I would like to use them as predictor. * CO2 sensor: Terenzi et al. (2020) investigated for apiculture that the correlation between CO2 concentration and the migration event of bees, in which bees leave the old hive, taking the stored honey. * Microphone: Cecchi et al. (2019) noted that bees communicate using a set of particular sound within the hive, potentially useful information to understand the health status of the colony. * 3D-modeled premade beehive structure: according to the local NGO, it would take two months for melipona bees to build the hive structure from scratch. As shown in the photo below, it is not the common hexagonal shape, making it difficult to recreate. I would like to see 3D print the same/similar structure to test if it helps bees save their time and allocate their energy on honey production. * Data storage: since I do not expect to have good reception in the middle of the Amazon, it is necessary to store the data until someone has a plan to visit a nearby city. * Battery: needed to power the sensors and memory drive. Inspired by Neil’s advice on spiral development of projects and skills, I decided to keep my project rather simple (but beyond my comfort zone) and set sub-goals to make the following components, each with the growth pillar I set in my mind (explained in brackets): A. Bee box, following the design practiced by a local NGO in Brazil over two decades. (subtraction manufacturing) B. 3D-modeled premade beehive structure. (additive manufacturing) C. Temperature sensor using a thermistor that monitors the temperature inside the hive. (input devices) D. Loadcell to measure the weight of honey. (input devices) E. Fan to cool down the hive. (output devices) F. Program using the Arduino IDE to activate fan when the temperature hits a certain threshold. (programming) G. Batteries to supply power for the chip and the fan. (others) The followings are what I had to give up based on my supply-side limitation (mostly time). I plan to finish the work during the winter break. H. WiFi connection via ESP32 to send data from the bee box to a web server. (networking + interface design) I. Data storage with a micro SD card. J. Other sensors, such as a gas sensor and a microphone. ##Making ###A. Bee box (subtraction manufacturing) My approach was first to follow the basic design of the bee box used in the local context over decades, as I believe it makes sense to make a baseline (i.e. I can always design different ones in future and compare the performance of such hives with this baseline model). <img src="./local_meli.png" alt="photo_link_missing" width="800"/> **Left: wild beehive / Right: artificial beehive adopted by local NGOs over decades (both images shared from Instituto Peabiru)** Below is the initial sketch I created in Week 0. <img src="./week0.png" alt="photo_link_missing" width="800"/> My idea was to make a construction kit like I did for Week 1 assignment (for laser cutting). My initial sketch included a pocket to insert an inner plate to make it cleaner from the outside (a technique I learned from the make-something-big assignment). <img src="./pocket.png" alt="photo_link_missing" width="800"/> Later I encountered a problem with Aspire when I was making toolpaths for shopbot from a dxf file; the dimension changed to one tenth (I triple checked the dimension on my Fusion 360 sketch was correct). Also, there were some unclosed paths created on the edges for some unknown reasons. Rhino could have done a quick fix, but since I have been using Fusion 360, that was not an immediate option for me. Following the advice of Joon, I decided to modify my design to eliminate the need of pocket and went for laser cutting. <img src="./no_pocket.png" alt="photo_link_missing" width="800"/> It went well with the first cut – the happiness did not last for long for two reasons. * Failure 1: material inconsistency. Although they looked more or less the same, the laser cutter could not cut the same plywood board with the same thickness for the second board even with the slowest possible rate (6.0 mm/sec with 100% energy; below this, the board began to get carbonized. I stopped experimenting the slower speed to avoid fire risk). At the end, I solved this problem by cutting the exact same paths twice. It cut well without too much burn, but another problem arose because of this operation. * Failure 2: kerf. Although I knew from the Week 1 group assignment that kerf is an important aspect to consider for laser cutting, I underestimated (for forgot) the degree of influence. The kerf affected my design in the scale of millimeters, and as the result, the board did not stick together. I could have used lots of glue, but decided not to do so and instead redesigned the board with the kerf I measured. <img src="./kerf.png" alt="photo_link_missing" width="800"/> Then it worked perfectly! After hammering them, the pieces got connected to each other so tight that I did not need a nail (although I put glue at the end to make sure my PCBs and devices do not fall down). <img src="./box_after_laser.png" alt="photo_link_missing" width="800"/> ###B. 3D-modeled premade beehive structure. (additive manufacturing) As I tried in the assignment for the 3D printing week, I decided to make a full scale version of the artificial beehive (not the hexagonal ones we normally see but something like a 2D-array of pots). I remember I spent at least 6 hours to design the honey pots in that week, but this time, it was within 30 minutes:) <img src="./3dp_front.png" alt="photo_link_missing" width="800"/> The printing went well although the hollow structure was partially filled by the supporting structures (I should have disabled them). Later I had to change the position of the hole due to the load cell, so I had to create a hole for the bees to go through. I used a heat gun to partially melt the base plate to create such a hole. <img src="./3dp_support.png" alt="photo_link_missing" width="800"/> <img src="./3dp_top.png" alt="photo_link_missing" width="800"/> <img src="./3dp_back.png" alt="photo_link_missing" width="800"/> ###C. Temperature sensor using a thermistor that monitors the temperature inside the hive. (input devices) The idea of using thermistor to make a thermometer was already there when I was doing the weekly assignment. However, at that time, I encountered the problem with SAMD11C chip that everyone else at Harvard shop encountered; the board did not get recognized by my MacBook even after seeing the success message on terminal for bootloading. This time, instead of using D11C, I used an ESP32 chip and made a generic board like Arduino (board with many pins for peripherals). I found this is so much easier once you get this board working, as the only remaining thing you have to do is to design the minimum circuit required for the peripheral. As you see, the thermometer circuit design was this simple, and it worked perfectly fine! <img src="./thermistor_schem.png" alt="photo_link_missing" width="500"/> <img src="./thermistor_route.png" alt="photo_link_missing" width="500"/> <img src="./thermistor.JPG" alt="photo_link_missing" width="500"/> ###D. Loadcell to measure the weight of honey. (input devices) I used loadcell(s) that came with an HX711 chip for signal amplification. Initially I was thinking about button-type load cells as it is easy to deploy four pieces of them, each in one corner of my beehive. <img src="./loadcell_heavy.JPG" alt="photo_link_missing" width="500"/> However, they did not respond to the physical stress no matter how strongly I pushed the board. Later I realized that button-type load cells were way too over spec – these were collectively suited to measure weight in the range of 250kg or more. I got Nathan’s help to get another one suitable for my expected weight range (bar-type for more or less 5kg). This load cell needs to be positioned in the center of gravity, thus blocking the hole made as the path for bees to go through. This made a bit of extra work to make another hole, but at least, it started working on my serial monitor. <img src="./loadcell_worked.JPG" alt="photo_link_missing" width="500"/> ###E. Fan to cool down the hive. (output devices) This was what I stumbled over for a long time. Initially, as I did in the output week, I was trying to use an H-bridge without realizing I do not need it for my fan (reversing the direction of rotation was totally unnecessary; I knew that, but I thought H-bridge contains a charge pump that would eliminate the need of the 9V power source). <img src="./fan_before.JPG" alt="photo_link_missing" width="500"/> I spent a lot of time figuring out how to get this board working, but later when I consulted Rob on this, I realized that I could use a much more straightforward circuit just enough to get the fan working. <img src="./fan_mosfet_schem.png" alt="photo_link_missing" width="500"/> <img src="./fan_mosfet_route.png" alt="photo_link_missing" width="500"/> <img src="./fan_after.JPG" alt="photo_link_missing" width="500"/> ##F. Program using the Arduino IDE to activate fan when the temperature hits a certain threshold. (programming) Now that both the thermometer and the fan are working well, it was time to put them together. I wrote the code below to (1) take the voltage changes due to the change of resistance in thermistor (i.e. the higher the temperature is, the more electrons manage to pass through the thermistor more easily and voltage goes higher) and (2) turn on/off the fan when it passes the threshold value. I have not calibrated the thermistor to actually display the temperature in degree Celsius, but it should not be a big problem. The weight measurement program is very simple so far. It works well to convert the voltage changes (caused by the tiny strains from tension and compression) into weight. Before the program run, I pre-loaded another program for calibration. I got this code from this website: https://www.electroniclinic.com/esp32-hx711-and-load-cell-based-digital-weighing-scale-iot-weighing-scale/ <img src="./program_everything.png" alt="photo_link_missing" width="700"/> ##G. Batteries to supply power for the chip and the fan. (others) I thought it would be the easiest part as all I had to do was to connect two cables, one from the battery to the board and the other from the board to the battery. While the answer should be still yes, it was also so easy for me after 3am to connect the wires another way around. It immediately generated smoke from my chip – after that, the board was no longer functional. Now that it was the day before the final presentation, all the stocks had run out – most notably ESP32. Ibrahim and Ben, who offered to destroy their old board to recycle a functional ESP32, saved my life - it took me several hours to put everything back. Lessons learned: * Be 100% sure about that the wires are connected in the right order. * Put an LED in a way to protect the chip; i.e. stop the current when the wires are connected another way around. ##Integrating Integration was not an easy, straightforward task. You should solve the 3D puzzle of geometry to nicely place all components and connect them with wires (actually this is when my brain got confused and I mistakenly fried my ESP32 while trying different connections). I decided to put most of the stuff underneath the top box so that they do not show up during my show-and-tell session. <img src="./nice_box.JPG" alt="photo_link_missing" width="700"/> <img src="./wiring_behind.JPG" alt="photo_link_missing" width="700"/> There was one problem when integrating everything; almost all “shields” needed to connect to the +3V3 and GND pins of the ESP32 board, and I only had one pin for each. I was ok with a breadboard before integration, but it was too big and ugly to be fit in the actual project. The solution I invented (,which I think has been invented by many other people) was to make a very small version of breadboard by myself – i.e. put a 10-pin header onto a tiny copper board. This way I did not have to worry about the lack of pins while saving space in my box by eliminating the need of the huge breadboard. [pin] Post mortem and future work * The beehive has the standard functionality I wanted to put now (except the WiFi connection to send data from the chip to a website). However, it is not really ready for the real use; for example, for the bees the bunch of wires are certainly annoying, fan may slice them or its wind could disturb their ability to fly inside the hive, and they may want to seal everything (such as the space between the load cell and the 3D printed honey pots) with their propolis. I need to troubleshoot with the local people who have been using such bee box in Brazil and/or Mexico to integrate their feedback. * Although it was not one of the most impressive projects in the class, for someone who started with no relevant background, the very fact that I was able to complete a project primarily by myself (well, of course with the huge help of TAs and other classmates such as Chris Zhu, Chris Wang, Ibrahim, Gabby, Benjamin, etc. etc.) made me a bit proud of myself. I am motivated to keep expanding my capability to try different features so that one day I can say I know how to make almost anything. Huge thanks for Neil, TAs, and the lovely Harvard Section community! <img src="./neil_and_takeo.JPG" alt="photo_link_missing" width="700"/> [original files](https://hu-my.sharepoint.com/:f:/g/personal/ttokunari_mde_harvard_edu/Eo_MW5iJhp1GvL7Oqn-jOrYB6G4nvFlrPQ0QWN5hT0okVg?e=J0TZIo "original files")