5/4/21 — New Beginnings
We had our first meeting with the course coordinator, Dr Ho. Prior to this, we threw some ideas onto a Google Document around the theme of the Environment. Some of our initial ideas included having a foldable takeaway box to reduce occupied space in landfills and a machine to recycle disposable masks. (We also toyed with the idea of an automated egg cooker to reduce queues for the egg station in hotels) Eventually, we decided on creating a system to recycle disposable masks considering the colossal amount of masks that are used ever since the COVID-19 pandemic began. Being able to recycle the plastics (polypropylene, polyethylene and vinyl) and fibres in masks would reduce the amount of new materials required to provide effective face masks, dampening the environmental effect of the pandemic.
Xiaoxuan and Sherryl met with Dr Ho online, presenting to him our ideas, research and rationale. Dr Ho was excited that we decided to incorporate COVID-19 issues into our project as he saw it as apt and timely. However, he had a different idea on how to reduce the impacts of disposable masks. Instead of making old products into new, he suggested reducing the consumption of new products instead. It was mitigative rather than adaptive. One reason why disposable masks should be replaced often, especially in high-risk medical settings such as hospitals, is due to the reduced efficiency when the mask has a higher concentration of water vapour. While cloth masks are reusable, they do not have a filtration efficiency that is comparable to surgical or N95 masks that removes up to 95% of aerosol particles with diameter around 0.3 ⲙm . Therefore, if we could create a mask that can be used repeatedly while maintaining a high filtration efficacy, we could reduce the reliance on disposable face masks.
Combining N95 masks’ electrostatic properties to remove aerosols and electrostatic dust precipitators used to remove pollutants from expelled factory gases, Dr Ho suggested we linked up with his Final Year Project student who was working on a Battery Operated Mask. She suggested having electrostatic filters in a mask that runs on a battery. This ensures that the electrostatic material can be continuously charged to retain its efficiency even after prolonged use.
We were intrigued by this idea and consulted our other group members Debbie and Shalina. Before further exploring and discussing our ambitious ideas, we had to clear a hurdle ahead – final exams.
(Reference for mask and aerosol particle: https://www.sciencedirect.com/science/article/pii/S2452199X20301481 )
Week 1 – (10-16 May)
11/5/21 — Welcome to M&T!
Exams are over! We could finally focus on doing the project. A briefing was conducted by Dr Ho to formally introduce us to M&T, safety issues and other administrative matters. We were told what to expect and what we should aim to achieve by the end of the module. It was an opportunity to get to know the other instructors in the course and each of their specialities. We were also introduced to the various roles available and we allocated the roles between us. The eventual role assignments are as follows: Debbie will be the Group Leader, Xiaoxuan is in charge of managing our finance, Shalina is the Safety Officer and I (Sherryl) took on the role of the communications manager. Thankfully, there were no qualms and everyone accepted their roles willingly. Lastly, we decided on our group name (for now) to be “Battery Operated Mask”.
Week 2 – (17-23 May)
17/5/21 — Virtual 3D printing lesson
Mr Han Yang conducted a 3D printing lesson over Microsoft Teams in the morning. In the 3 hour session, he explained how to use AutoDesk 360 to draw 3D models. We paid close attention as he went through the entire modelling process in great detail – from sketching a 2D surface and then extruding it to become a 3D object. He also emphasised the safety precautions we should be aware of when operating the 3D printer.
18/5/21 — Virtual block diagrams session
This session was conducted by Mr Tony Gan, the hearsay ‘boss’ of the M&T Lab. He went through the fundamentals of block diagrams, including its purpose and applications. We learnt how to create block diagrams on Lucidchart that simplify the representation of processes and allow others to visualise them clearly. Tony also went through some common electrical components that we might use in our project. As someone without prior electrical or computing knowledge, it was my first time hearing terms like ‘microcontroller’ and devices like ‘Arduino’ and ‘Raspberry Pi’. Thankfully, two of our members major in Computer Science and had some clue as to what was going on. Little did we know that this was just the beginning of our arduous journey with electronics and coding.
19/5/21 — See you online!
Using the knowledge gained from the lessons, we decided to have a meeting to discuss the variables that should go into the block diagram and other concepts regarding the electrical components (covered yesterday) that we may need to research on.
There were two parts to the project. First, we had to build the electrodes and the required electrical components to attract the aerosol particles. The variables we wanted to change were the number of electrodes and the distance between the electrodes to vary the amount of electric force exerted on the aerosol particles. To ensure that the device is still effective when a person coughs or sneezes, we wanted to vary the air speed as well, so that even with high velocity air, the electrodes can still attract all the aerosol particles.
The second part required us to test the effectiveness of the device. We had to find a way to measure and visualise the removal of aerosol particles to prove that the electrodes are doing their job. This was something we had great difficulties with. We brainstormed ways that could allow us to obtain evidence of the aerosols decreasing in amount after passing through the electrode. Some of our ideas include spraying coloured water across the electrode, measuring humidity, using Cobalt (II) Chloride paper to detect aerosols after the electrodes and using a laser sheet to visualise airflow.
The discussion provided clarity about the direction of our project, making sure that all members were on the same page. We also began working on the Powerpoint slides for the progress update presentation next week.
21/5/21 — The Four Maskerteers
This meeting marked the birth of The Four Maskerteers! The main goal of this meeting was to create the initial design and ensure that it could work theoretically. We spent a great amount of time researching the specific factors that may influence whether the aerosol particle gets attracted to the electrode. We considered things like the average and maximum airflow rate from breathing and coughing, the average size of an aerosol particle and the corresponding electric field strength and electrode distance required to remove the particle. After reading multiple papers written recently due to the COVID-19 pandemic, complemented by some materials from Ms Shi Yu, we decided that theoretically, it was possible for electrodes to create a strong enough electric field that would attract aerosols within a short distance. We came up with a rough design of alternating positive and negative electrodes powered by a DC source. Instead of using metal plates for all of the electrodes, we decided to use wire mesh between two metal plates to create an inhomogeneous electric field that is required for stronger attraction of aerosols. Air would then pass in between the oppositely charged electrodes and aerosols within the air should be removed by the electrodes.
Week 3 (21-30 May)
27/5/21 — Risk assessment briefing
We gathered online to attend a briefing by Dr Alicia Ng about the filling up of the risk assessment form online. She went through the different components of risk assessment, such as the hazard, sub hazard, possible accident, existing risk control and additional risk control. We were also taught how to evaluate the danger of a risk based on the severity, likelihood, and resulting Risk Prioritisation Number (RPN).
28/5/21 — M&T presentation
This was a virtual meeting where we could share our ideas and plans with the instructors and the lab personnel. It was also a chance to learn from other groups and get inspired by their project. We shared about the design specifications of the electrostatic aerosol removal device, the components required and some factors (such as air speed and distance between electrodes/number of electrodes) that we would be varying to test the optimum conditions for the aerosol device to work. This was also an opportunity for us to seek advice on how we can test the effectiveness of the device to prove that it is effective in removing aerosols. Dr Ho suggested using a laser sheet to visualise the movement of the aerosols. We’ll explore that in the future!
Week 4 – (31 May -6 June)
1/6/21 — Delving Deeper
Heading the feedback received during our presentation, we took a closer look into the specifics of the electrostatic precipitator. We finalised details such as the number of plates we wanted, the design of the device. We prepared for our meeting with the Year 4 student whom we had arranged a meeting 2 days later to learn how she did her simulations. We decided to ask her how to use the system in COMSOL, how she predicted the voltage and electrode dimensions, and her thoughts on using a laser sheet to visualise particle movement.
3/6/21 — Teach me Senpai
Dr Ho referred us to Ms Shi Yu, a Year 4 student who completed her Final Year Project under his supervision. Her project investigated the theoretical possibility of using electrodes to attract aerosol particles in a face mask, with the goal of maintaining high filtration efficiency while ensuring breathability. She used simulations such as COMSOL to determine if particles would be completely attracted to the electrodes within the thickness of the mask.
During the meeting, she ran through how the simulation worked. We were introduced to the creation of physical boundaries for the inlet and outlet airflow portions, the use of symmetry plane, the types of mesh to use, and the need to indicate the different forces in the simulation. We also learned that we have to set the simulation to a stationary study if it is not time-dependent. She pointed out that it would be difficult to learn how to use COMSOL from scratch within the short runway of our project. Instead, she offered to let us use her simulations as a starting point, and edit the variables from there.
After the call with the senior, we then brainstormed once more about how we would secure the plates to the device and arrange the necessary electrical connections since adjacent plates would have opposite charges.
We also decided to 3D print the electrode (as we thought it would be possible to print metal. We only found out later that it was not possible in the lab), the plate holders, the frame and the wire mesh that would be in between each plate. The diagram below shows the overall design, with the plate holder magnified. In preparation for the upcoming 3D printing workshop, we delegated the work so that everyone could try 3D modelling a part of the design.
Week 5 — (7 -13 June)
8/6/21 — First whiff of the M&T labs
We had a physical lesson with Mr Han Yang where he went through how to use a 3D model slicer software, Cura, to prepare the 3D model for printing. The slicer would be able to process the 3D model design so that it can be read by the 3D printer. We also learned supports that is needed for overhang elements, the density of infill, the size limitations (20cm x 20cm x 20cm) of the print area. Considering these, we were advised to divide our parts into smaller pieces and then combine them after they are printed. After going through the theory aspect of 3D printing, Mr Han Yang orientated us around the M&T lab. We were showed where to find the equipment and materials we needed and various safety procedures we should keep in mind. We also consulted Mr Han Yang on our project, and how we should break our 3D models apart and recombine them to reduce printing time. For example, our initial thought was to print a box with two empty sides to contain the parts. We were then advised to print U-shaped parts instead to reduce print time. Additionally, we were advised to buy the metal plates and mesh instead of printing them out as we required precise measurements of 1mm, which the 3D printer is unable to do. We were also introduced to heat set insert to replace nuts and bolts for a more seamless design when attaching pieces together.
Afterwards, Xiaoxuan and Sherryl attended a soldering lesson where they learned the proper steps to solder components to the electrical board. We would first have to heat the metal component of the board before feeding the solder. To practice, they had a chance to solder the components required to form a digital clock! It was great fun to see the fruits of our labour.
Just before heading home, we introduced our project to Tony and the other instructors in the course. They voiced concerns about the high electric field that we require as it would be dangerous for ourselves and the user. It was also difficult to test and prove that aerosol particles were removed due to their small particle size. Even with the laser sheet, we would not be able to distinguish which image is from the dust particles, aerosols or other impurities in the air. Therefore, we decided to shift our focus to the removal of dust particles which are larger, and to find an alternative dust removal method. We threw some ideas around, such as using physical barriers. HEPA filters remove small dust particles most effectively through a mesh of fine fibres. However, besides trapping just the fine particles, HEPA filters often get clogged by larger dust particles which decrease its life expectancy. We thus thought that we should shift our focus to removing larger dust particles before they reach the HEPA filter. This would ensure that the HEPA filter would be able to last longer. Drawing inspiration from nature, Tony suggested looking into how the rain is able to clean the air and how air is cleaner after passing through water bodies. We brought these ideas to Debbie and Shalina for further discussion.
10/6/21 — One Step Forward, Two Steps Back
The feedback we’ve gotten threw us off and we had to have a drastic change of plans. We held an *emergency meeting* online to discuss how to move forward with our design. We found the idea of using WATER to purify the air intriguing and unique. It was also relatively safe to use.
Upon more research, we found the concepts of dry and wet deposition to clearly describe the natural phenomena we observed. However, we also realised that it was not a method commonly used in conventional water-based air filters. Those on the market usually involved bubbling the dirty air into the water. The dust then gets trapped by the water and sinks, while the clean air diffuses out of the water. We decided to explore that kind of air filtration first.
This marked the birth of the ~water benders~
Week 6 — (14 – 20 June)
15/6/21 — The Evolution
On our own, we each did some research on the design specifications of water-based air filters in the market. When we met, we brainstormed on how to bubble the air into the water, just like other air filters. Based on our research, the conventional air filters bubbled the air from the surroundings through the bottom of a container of water. However, we could not find a way to ensure that there was no backflow of water into the pumps and to ensure that all the air that will pass through the HEPA filter would have been cleaned by the water first.
We then went back to the idea of the rain and the sea. Could we somehow emulate natural processes in our air filter? We decided to have a part where water sprays down on the air for the ‘rain’ component and to blow air through sheets of water for the ‘sea’ component. To ensure the versatility of the air filter, we came up with a modular system, where each part of the air filter can be removed if not needed. For example, if you would just like to just use the HEPA filter and the fans that blow air towards the HEPA filter, you could remove the water component and vice versa.
17/6/21 — Consultations
We brought our new ideas to Tony for further feedback. He raised the concerns of the HEPA filter being wet by the water, and suggested using a cooling coil to condense and remove the water before the air reaches the HEPA filter. We also discussed the parts we had to purchase and the overall feasibility of the idea. He showed us the different kinds of fans that we could consider getting and the benefits of each. He also showed us different kinds of arduino models and possible user interfaces that we could use. From this discussion, we gathered that we would require cooling coils (typically used computers), an accompanying water chiller, centrifugal fans, and aluminum profiles to build the overall supporting structure.
In the afternoon, Dr Ho came down for the informal progress update. We shared our ideas and the feedback from Tony. Dr Ho raised concerns about the efficiency of the air filter as it should be able to clean the air in the entire room within a reasonable amount of time. We concluded that 75 m3/hour was a reasonable air flow rate into the air filter. Using this value and the specifications of the fan, we deduced that we would require 4 of the fans to provide a sufficiently high inflow of air. Dr Ho also mentioned that we should try to increase the surface area of contact between the air and the water to ensure that the air particles would be exposed to the water as much as possible. We thus decided to reduce the spacing between the sheets of water to as small as 1 mm to ensure there would be a high chance that air that passes through would be in contact with the water. We also thought of blowing air at an angle to the sheets of water to create turbulence which would hopefully increase contact with the water. Dr Ho suggested using simulations to test if our ideas would work, which we placed on our agenda for next week.
Week 7 (21-27 June)
23/6/21 — SIMULATIONS
Using the computer in the M&T Lab, we tried to use SOLIDWORKS to do airflow simulations to visualise the way air would behave in the air filter. We first learned how to draw the 3D image of the air filter in the software. This took awhile as the part that we were trying to simulate involved multiple thin plates that were closed together. We eventually decided to draw it out on the Fusion 360 3D modelling software and then export the file into SOLIDWORKS. Running the simulation was tricky as we had to select the correct input and output surface and set the correct conditions. We ran into multiple errors which we sought help from the mentors to resolved.But this was not the end of our worries. Even if we managed to run the simulation, it required roughly 80 days to complete. It was clearly not feasible due to time constraints of the project. We tried to change the simulation conditions to reduce the time taken. For example, we converted the volume flow rate to mass flow rate. This shortened the simulation time to about a few days. We left the simulation to run overnight.
While waiting for the simulations to be completed, we also started drafting out the precise measurements of the air filter. This included the dimensions of each module and the length and quantity of each aluminum profile we would need to purchase.
24/6/21 — And More Simulations
We returned the next day to check on the simulation, only to find that there was an error in the simulation which caused it to stop halfway. We tried to fix the problem, only to come back a few hours later to find that the simulation had stopped once again. Disappointed, we decided to try to use rapid prototyping instead to have an idea of how the water would flow.
For rapid prototyping, we used unwanted corrugated boards found in the M&T lab to create the plates that water would flow over. Initially, we created a ‘Z’ shape design to allow water to flow down like a water feature. Towards the end of each level, there were slits to allow water to flow down to the next level. We tested this design by pouring water over the corrugated box. However, we found that the flow was not uniform as water would tend to converge towards the centre of the plate instead of being spread out. This could be due to the jagged edges of the corrugated box due to the difficulty of cutting smooth lines on the corrugated board material. Another reason for this could be that the water was released from a single channeled source (the hose).
To overcome this, we decided that the plates should be as close together as possible so that surface tension of the water would cause it to be evenly spread out. To ensure that water is dispensed as a sheet of water right from the beginning instead of from a small opening (from a hose), we thought of allowing the water to accumulate in a container and flow down the plate naturally when the container overflows. This would ensure a uniform flow of water being dispensed onto the plate.
While we were working on our prototypes, we had a chat with Dr Ho. He saw that we were busy cutting out corrugated board pieces. He suggested using the plates of a heat sink as a prototype instead to test if a uniform flow of water can be created by releasing water over two closely placed surfaces.
Week 8 (28 June – 4 July)
1/7/21— Moving Forward
After deciding to use the ‘overflow’ method to supply the water to the plates, we began to design a “water feeder” that collects the water, allowing it to accumulate before flowing out through slits in the water feeder. We created a 3D model for this prototype and proceeded to 3D print it.
After a few hours, we collected the 3D print and tested it by inserting a pump to the top. However, we found that water was exiting the bottom slit even before sufficient water has accumulated. Therefore, we decided to turn it upside down, such that water is supplied through the bottom. This would allow water to accumulate much faster, and the water pressure from the bottom prevents water from flowing out from the slit as quickly. By doing so, we are able to ensure there is roughly the same amount of water flowing out of each slit.
We also decided to experiment with the number and position of the holes to find the optimum conditions for the water feeder to fill up as fast as possible. Our second design had 3 holes from the top and 2 holes from the sides to increase the amount of water entering the water feeder. We also included a horizontal divider between the input and the slits to allow water to accumulate before flowing through the slit, instead of flowing from the input hole to the slit directly. We hope that this would produce a more even flow. We then sent the 3D model design for printing. Since we were sure that we needed more inlets for water flow, we also bought more water pumps.
Dr Ho popped by the Physics lab and we explained that the purpose of the slits on the “water feeder” was to restrict the flow of water to the small hole, hoping that it would ‘squeeze’ the water out which would produce a sheet of water. Dr Ho related this to pinching a hose and having a thin sheet of water instead of a circular flow. Inspired by this, we decided to include a slit in the divider inside the water feeder. (You can see the slit from the holes in the water feeder). We 3D printed the third water feeder overnight.
Finally, we started to assemble the aluminum profiles that would form the main structure of the air filter.
2/7/21 — Trial and Error
Our second 3D printed water feeder has finished printing! We collected it and tested if water would flow through each slit evenly. When we connected 3 hoses from the water feeder to the tap, we observed that the output flow was concentrated at the bottom and middle slits. The top slit had only a little water flowing out through the edges. We thought this could be because the water was leaking from the side holes and we determined that the side holes were redundant. We also collected the third version of the water feeder and tested it as well. We realised that the slit within the divider did not improve the outflow conditions as water was flowing down into the slit rapidly instead of filling up the space in the water feeder. We thus decided to remove the slit in the divider. Learning from these three prototypes, we decided that the best design would be to have holes at the top only and no slit in the divider.
In the meantime, we started to assemble the first module. We attached the fans to a supporting aluminum profile using right-angle brackets. We also drilled holes to a stiff hose for the nozzles to be inserted. This would form our second module. We tested the strength of the “shower” by connecting it to a pump. The radius and size of the spray were not ideal as the water was not dispersed enough. We then decided to purchase more powerful pumps.
Week 9 (5-11 July)
7/7/21 — Preliminary Tests
It’s time to test the dust sensor! We bought two Sharp GP2Y1014AU0F sensors that we wanted to connect to the Arduino mega. We found connection guides online and after some difficulty (because there were many wires all connected differently), we managed to successfully connect it to the Arduino. The dust sensor worked well when we tested it by blowing strong wind into the sensor in M&T lab!
The dust sensor uses light scattering to determine the amount of dust. The LED inside the sensor shines light towards the hole where dust enters. Due to the random motion of the dust particles, they deflect light towards a light detector that is at an angle to the LED source. The higher the density of dust particles, the more light is scattered and hence the greater the voltage of the light detector. A linear relation is then used to calculate the dust density from the voltage value.
Besides working on the dust sensor, we also tested the fourth design of the water feeder. This model worked well! We finalised the dimensions and sent our orders to the acrylic supplier. We calculated the inflow rate we required for the water feeder when we scaled up the dimensions and concluded that we required 4 pumps to the water feeder to ensure rate of inflow. While the supplier was not able to assemble the entire box for us, they were able to prepare each side of the box with the relevant holes and cutouts. We also ordered 1.5mm thick acrylic sheets from them which we will use for the “plates” module.
8/7/21 — Preliminary Tests Part 2
Now that we have our dust sensor up and running, we wanted to test it using actual environmental conditions. We were also curious to find out if the “shower” was effective at removing dust particles. But first, we had to purchase corrugated boards to seal up the sides of the air filter.
We made our way down to the bookshop at North Spine to purchase corrugated boards and lugged them to the other side of the campus. We got down to cutting them to the appropriate sizes and stuck them to the sides of the aluminum profile. We then connected the pumps and fans to the DC power source which was set to 12V.
During the first round of testing, we realised that water droplets were splashing into the dust sensor which interfered with the results. The water droplets were mistaken as dust particles which resulted in extremely high results near the “shower”. We then started to build deflectors to help prevent water from entering the dust sensor and to prevent water from damaging the HEPA filter.
It was also then that we realised that the condenser/water chiller was not effective at cooling down the air. There was excessive heat being produced from the radiator which prevented the water entering the condenser from cooling sufficiently. This meant that we could not condense the moisture from the “shower” component. This could be an area that we could look to and improve on if we had more time and resources to find alternative radiators that have a higher cooling efficiency.
9/7/21 – Progress Meeting
We presented our progress so far to Dr Ho, Dr Alicia Ng and Mr Han Yang, telling them about our design.
Some feedback we’ve gotten was that the water droplets from the “shower” was too large, which reduced the contact with the air particles. We decided to find an alternative nozzle for the “shower”, hoping that it would be able to produce more mist rather than large water droplets. We also noted that a possible cause of the large water droplet size was due to the flexible hose which bent in our set up yesterday. To counter this, we would be changing the hose to one with a thicker material that would not bend so easily.
There were also concerns about the long wait time for the acrylic sheets that we ordered. Dr Ho suggested we experiment with plastic files while waiting for the acrylic.
Week 10 (12 – 18 July)
15/7/21 — Electronics
We met with Dr Alicia in the morning to share with her how we plan to address the concerns raised during the progress meeting. Since then we have purchased new nozzles and sent our orders for the acrylic plates and water feeder.
While waiting for the remaining components to arrive, we focused on setting up the electronic parts and testing the codes for the pumps, fans and sensors.
Over the weekend, Sherryl went down to Sim Lim tower to purchase the temperature sensor and jumper cables. We’ve got both DHT-22 and DHT-11 temperature and humidity sensors. DHT-22 is a more advanced version of DHT-11 with a higher accuracy.
Our Raspberry Pi 4 Model b 8GB arrived and we tried turning it on, configuring the settings. We learnt how to connect the computer and the Raspberry Pi remotely through the same IP address. We also learned about user interface and we discussed the different elements we would want the user to be able to see and control. There is a front-end and back-end of the software we will be using. The front-end is what people would see on the Raspberry Pi screen while the back-end is the code that programs the system. We decided that we would want the user to control the power of the pumps, fan speed and be able to track temperature, humidity and dust density values on the Raspberry Pi screen. We used Flask to create a web page that would display real-time values of the above parameters. We then linked the back-end code to the web page and successfully connected one dust sensor and a DHT-22 temperature sensor to the Raspberry Pi.
16/7/21 — Electronics Part 2
After learning about the setting up the user-interface yesterday, we tried to connect the Raspberry Pi Screen to the Raspberry Pi. We successfully connected the screen and were able to display the React Flask on the screen. We were also able to see the changes in values displayed on the screen.
After a productive morning, we faced many difficulties afterwards we tested the MOSFET with the PWM. MOSFET stands for metal-oxide-semiconductor field-effect transistor and is used to amplify the voltage or is used as a switch. The PWM is Pulse-Width-Modulation that generates variable-width pulses to represent the amplitude of an analog input signal. It is useful to convert digital signals from the Raspberry Pi to analog signals received by the fan and pump motors. We connected the Raspberry Pi to the PWM and Servo breakout board (PCA 9685). However, we ran into trouble with the code and connections. Initially we made a mistake with the wiring and connections. We were able to fix this by having an *emergency meeting* with Kelvin (our mentor). Afterwards, we had trouble with the code as we could not turn on the fans with a low duty cycle. This resulted in a very small working range of the fans. After consulting Tony and the other TAs, we learned that we needed to connect capacitors to smoothen out the input voltage to the fans. This results in a more constant supply of voltage and decreases the sudden spikes and falls in the voltage values. We were then able to expand the range of voltage values that the fans could operate under.
Week 11 (19 – 25 July)
19/7/21 — wire trouble
While fixing the PWM, we learned that the PWM has a switch (with a low voltage) that turns the input voltage from the 12V power source on and off. From the previous week, we decided that we need a capacitor to fix it.
Connecting the other electronic parts together, we decided to use these connectors (shown in the photo) to connect the wires. We were having difficulties fitting them into the connector and then connecting them into the PWM board. After doing up the wiring, we still could not turn on the fan modules. We suspected a problem with the capacitor and tried replacing it with other capacitors. In the end we realised we used the connector clips incorrectly. Within the clip, the wires are connected horizontally, such that one clip should be used for one type of wire.
Afterwards, we faced problems with the DHT 22 sensor as we found that it is not compatible with the port expander. We thus could not use DHT22 as the temperature and humidity sensor.
We went back to the drawing board and drew new block diagrams to figure out how we should reorganise our connections. We also decided that we needed to decide where we would place our sensors and how many we would need.
19/7/21 again — Everyone/thing deserves a second chance
The new nozzles and accompanying pipes arrived and Debbie and Sherryl decided to try them out. Initially, they thought that the second one was better as the water vapour produced was more misty. But they decided to try the original nozzles again. They found out that when the first one was fixed properly and the kinks smoothened out, it was actually more powerful and had a better flow. (Possible life lesson! We should allow things/people to improve and give them another chance :D)
We previously purchased a 90 degrees connector to connect the hose to outside of the module. However, we realised that the connector that we bought was not the right size. We then thought of using the connector to connect the hoses inside the module so that there would be no sharp kinks in the hose, which ensures that the nozzle receives as much water as possible. We tried to tape around the hose to make it thicker and eventually managed to squeeze a larger pipe in with BRUTE FORCE.
“The solution to EVERYTHING is BRUTE FORCE.” – Debbie 2k21
21/7/21 — Rough Day
The additional dust sensor, MOSFETS and analog to digital converter (ADC) (MCP3008) we ordered finally arrived. We tried to connect the dust sensor to the ADC since the dust sensor is analog while the signal from the Raspberry Pi is digital. However, most people don’t use the dust sensor with Raspberry Pi so there is no available code for us to refer to, and we had to modify existing codes. Another issue we faced was that the sensor uses 5V voltage whereas Raspberry Pi is 3.3V, this inconsistency meant that we needed a transistor. We then spent time troubleshooting but were unable to find the source of the problem. We decided to find alternative dust sensors.
Week 12 (26 July-1 aug)
29/7/21 — At long last
We started the day checking if the sharp dust sensor was working well. Previously, we were having problems with connecting it to the Raspberry Pi. But thankfully, it worked this time around and gave logical values when we ran the code. However, the dust density values varied significantly, and this is one area we could look to improve in the future. Our environment sensor (M5stack) arrived and we checked if it was working as well! It was a much better option as it was an I2C sensor that could be connected directly to the Raspberry Pi.
At long last, the acrylic sheets ordered a few weeks back arrived! We decided to customise acrylic sheets to fit the exact size of the water plates and the water holder box. However, the company we engaged was only able to supply individual sides of the box separately and did not glue them together. Therefore, we had to use acrylic glue to assemble the acrylic sheets to form the water feeding box. Debbie and I worked on the acrylic box, but we messed up the first time when we glued the sides wrongly. Once again, BRUTE FORCE saved the day! We pulled the parts apart and applied glue once more, this time the correct way. We also discovered that it was insufficient to simply apply glue on the side where the two acrylic sheets would meet each other. In order to secure the parts, we had to apply a second layer of acrylic glue (solvent cement) to both the point of contact, and on the outer surface.
Meanwhile, Xiaoxuan assembled the circuit boards onto the corrugated board. We decided that all the circuit components should sit safely outside the air purifier so that it does not get water damage from the wet components. We learned how to utilise a stripboard to accommodate multiple connections to the same pin on the Raspberry Pi. The header pins that are also on jumper wires had to be soldered onto the stripboard so that it acts almost like a reverse breadboard that allowed more organised connections. Xiaoxuan is the official master at soldering and she took on the responsibility of soldering 6 rows of 4 pins onto the stripboard. The pins were for the power, ground, SCL and SDA connections. After planning how to arrange the different boards onto the corrugated board, we started to assemble and connect the different electrical components together.
30/7/21 — Getting it Together
We started to connect the pipes together, but we realised it was still a little too loose for the acrylic water feeder. We tried to use sealing tape to reduce leakage, but it was not totally effective and it started leaking again after a while.
To better secure the hose to the acrylic water feeder, we decided to 3D print pipe holders with customised pipe sizes. (The pipe holders are shown in pink) This was tricky as the dimensions did not seem to fit well the first time we tried it. We had to go through rounds of editing and measurements before we found an excellent fit for both the acrylic water feeder and the hose.
While Debbie and Sherryl worked on the pipes, Xiaoxuan connected the wires that will bridge the electrical components across the different modules. This was the first time we were using this connector so it required a great deal of patience.
Week 13 (2 Aug – 8 Aug)
2/8/21 — Oppsies
Today we gathered at the sofas outside the M&T lab as the Physics labs were closed in preparation for the new semester. Over the weekend, Sherryl bought stripboards (or so she thought) and wire clippers from Sim Lim Tower. Being inexperienced in electronics, little did she know that dot boards exist. What she thought were stripboards turned out to be dot boards. We then spent time attempting to turn the dot boards to strip boards. First, we tried to solder across the empty holes in the board. That didn’t work as the solder would overflow to neighbouring columns which would cause a misalignment in the connection. Then we tried to solder a connecting wire through the column. This was difficult as the thickness of the wire made it hard for the solder to attach it to the extruding pins.
Meanwhile, Xiaoxuan and Shalina attached the wires to the connectors to be stretched across the different modules.
5/8/21 — Coding master
Today we had trouble turning on the Raspberry Pi so unfortunately, we spent a lot of time trying to troubleshoot the issue. We eventually realised that we had to set a static IP address for the Raspberry Pi to access it. Xiaoxuan then edited the src and api.py code for the interface to take into account the addresses we soldered onto 2 of the pca9685. Xiaoxuan also edited the code to connect all the pump controls onto our interface.
6/8/21 — Getting back
The next day, Xiaoxuan and Shalina fixed the wiring for the Raspberry Pi, and made the change to thicker wires for the wires connecting the components to the power source. This enabled the wires to support a higher current. We then decided to use lan cable connectors instead, for a more stable connection between the Raspberry Pi and the sensors. Xiaoxuan then desoldered the old white wires and soldered on the new lan cables (i2c). Shalina then attached the acrylic sheets to the sides of the aluminium profiles but we then realised the dimensions of the acrylic sheets were off. We mistook the measurements of the screw holes to be the distance from the edge of the holes instead of being the distance from the center of the screw holes. We had to make do so only 3 of the 16 holes were able to fit the dimensions of the aluminium profile. We also realised that due to the way we attached the corrugated boards to the aluminium profiles, on the outside of the profiles instead of on the inside. This increased the distance from the acrylic box to the supports on the other end of the box. This meant that the acrylic sheets we ordered were too short to span the entire length of the box. To salvage this, we decided to elongate the supports for the acrylic sheets and 3D printed new supports.
Week 14 (9 Aug – 15 Aug)
This week we collected our prints only to realise that we had changed the design of the box again to only have space for 6 acrylic sheets instead of 9 due to restrictions from the supplier’s end. Hence we had to edit the previous 3D model again to meet the specifications of the current acrylic box. We also decided that we had to print supports for our acrylic box to attach it to the aluminum profile. After printing the new supports according to the dimensions of the pipes, we realised that the sealant blocked up the holes we modeled for the pipes. So it was back to the drawing board to edit the dimensions to take into account the sealant. We then decided to cover up the modules with corrugated board to contain the water from our modules, mainly the shower module which would spray water everywhere.