Progress of project for week 4
Goal Of Week 4: Prove our theory on a small scale. Ideate and Design Everything needed for prototype 1
Description: No idea if our theory of air weighing the setup down could even be practically translated in real life on a small scale.
We did our very first experiment this week by using two generic party balloons(one stuffed into the other). The outer balloon was sealed off with helium in it while the inner balloon was connected to a bicycle pump via straws.
Initially, the balloon was floating with the straws tethered. The aim was to bring it back down with the addition of air. However, it failed to bring it down. Later, we realized that the balloon containing the helium should be non-expanding, that way, the upthrust will not increase.
After the experiment, we spent the next 3 days designing our very first prototype. Drawings shown below.We deliberated on a few things.
- Type and quantity of party balloons required to lift the setup
- Some kind of small air pump
- a gondola to hold all the electrical component
We dabbled in a few other designs too…
At the end of the week, we settled on the 3rd design (3rd picture on the right). It was simple and easy to make. The only difference was that instead of a syringe-like container to hold the air, it would be a balloon. Next week we will deliberate on how air will be delivered in, the circuitry required and the amount of helium balloons needed to lift.
Progress of project for week 5
Goal Of Week 5: Still wanted to prove our theory, but now with a half completed prototype 1.
Description: Gaining traction from week 4, we decided to go ahead and put together an early version of prototype 1…
We spent the first few days of the week planning on the four problems: Balloon, Pump, gondola and circuitry and this was our solution.
1. Balloon
On the left, the flesh-colored balloon would act as the inner balloon with air being pumped in, whilst on the right, the foil balloon would act as the lifting balloon with helium sealed in.
Using a simple weight comparison method, we were able to work out the upthrust generated by one foil balloon (27g). After we bought all our components, we weighed them and determined the number of balloons needed with an extra safety factor for wires (the setup was to be tethered).
2. Pump
The pump was the most worrying part. It had to be strong, yet light and does not draw too much power. We were able to find a simple 6V DC rotary vane pump that weighed 70g and was cheap ($15). The mass flow rate was 2L/min (no load).
This pump was light, reasonably strong and consumed very little power (drawing 420mAh at 6v)
3. gondola
The gondola was kept simple for now. we did not want to waste time and money designing a gondola when we have yet to finalize a completed design for prototype 1. We opted to use a plastic bag just for this week.
4. Circuitry
This was slightly different. This circuit was designed with forethought. We could have hooked the pump up to a power source and that would be it. But we wanted to stimulate an experiment closer to the actual prototype, it would also be a good opportunity to see whether air’s weight was significant enough to effect change on a (relatively) heavy setup.
with that in mind, we bought the lightest Arduino (nano). A MOSFET to protect the Arduino and some wires and a miniature breadboard. The Arduino was bought with the expectation that we would need to remotely controlled the setup and the motor speed as we progressed on.
Power was (for now) the least of our concerns. The setup was tethered to a DC power supply. It was also not RC capable. We put all these together, and the second experiment was underway.
Notice that the flesh-colored balloons that were supposed to contain air was hanging out. Also there were two balloons because for this specific iteration of the experiment (we did a few) we decided to use two pumps for double the effect. The key observations between the double pump and singular pump iteration were exactly the same.
Those key observations were… :
1.The setup did not go down even when the balloon grew almost 4x the size compared to the picture above.
2.The pump(s) was surprisingly strong albeit really slow (took about 15 mins to reach the size in that picture, 40 mins to grow 4x the size)
3. The setup was extremely sensitive to weigh changes. while preparing the experiment, a tiny smidge of blu-tack was enough to bring the whole thing down
4. The setup was extremely sensitive to external forces. A person walking past the chair would generate that minuscule amount of breeze. That breeze would push the setup flying away in another direction.
Observation 1 was the most worrying. We spent time thinking about it until we realized that the air balloon had to be either inside a helium environment OR placed in a rigid vessel. Otherwise, the increase in size would increase its upthrust, almost perfectly negating its increasing weight.
We decided to try out the “helium-environment method” first. We spent some time sourcing for a suitable vessel to contain both the flesh-colored balloon and helium. We settled on the SAF standard trash bag. It was large enough, it was thick and it was free and readily available. The setup was quickly put together and the experiment was performed.
Note the difference in this picture, now, the air balloons do not hang out but is kept inside the army trash bag filled with helium.
Some key observations were made:
1. The setup went down once the air balloon was filled up enough.
2. Sealing helium inside a trash bag with a huge opening was extremely challenging.
We ran the experiment a couple more time, before being satisfied with the result, the setup finally went down.
Progress of project for week 6
Goal Of Week 6: To bring the setup from the ground back up again. To make it un-tethered
Description: Riding on the momentum from last week’s success, we decided to work towards completely prototype 1 and testing it outside the lab. Meanwhile, some of us started to doubt last week’s results…
There were some concerns that the reason why the setup went down was because of helium leaking out of the army trash bag. As the flesh-colored balloon increased in size, pressure inside the army trash bag increased drastically; this itself can push helium out. Exacerbated by the fact that the sealing method we used was rudimentary (cable ties and careless taping).
We decided to run the experiment again, but this time adding straws inside the flesh colored balloon. The straws were folded in and plugged with blu-tack. The idea was that if air was truly responsible for bringing the setup down, then the evacuation of said air would bring it back up again.
This was the aftermath of the experiment after air was pumped in and evacuated. Compared to the picture above, this is extremely deflated. It was clear, that the experiment was neither definitive nor accurate. It was obvious that the setup did not float back up even when air was completely evacuated. With this in mind, we tested the rigid vessel idea, in hopes that we can do away with the sealing and leakage problem.
Choosing a rigid vessel presented a whole new set of problems. This was a non-stretchable plastic bag. The pump was over-ran at 7.4v instead of the normal 6v for more power. the pump only pressurize the already maximally inflated bag a little before experience a complete back-flow. We surmised the following:
1. Finding an airtight rigid (or non-stretchable) vessel that was strong enough and light enough was a difficult contradiction of constraints.
2. A device that was light, strong enough to pressurize, without consuming too much power (to keep the battery small) and (to top it off) cheap. Yet another set of difficult contradictions.
We decided, that it was more worth our time and effort and money sticking to the helium atmosphere method. The only challenge was to find a way to seal in the helium.
In the last few days of the week, we decided to focus our efforts on the electrical aspect of the setup. None of us had a good understanding of electrical circuits. We had to spend time researching and shopping before coming up with the simplest solution. The solution was a circuit made out of:
- A 80g 2s (7.4v) lipo at 1100 mAh. Smaller capacities were considered, but seeing the time it took the pump to fill up one balloon, 1100 was the minimal, seeing as lipos cannot be fully discharged.
- A electromagnetic relay switch controlled by RF. A controller came with it. This switch could only close or open the circuit. Range was 100m+
- A step-up. The battery could only supply 7.4v but the electromagnetic relay switch required 12v.
The entire setup would come in weighing at 190g. A tolerance of 30g was given in lieu of the wires and solder and tapes. Because we already knew the nett upthrust generated by one foil balloon (27g of upthrust), there was no need to include in the weight of the total volume of helium. We also previously measured the average size the balloons blew up to last week and measured the weight of air to be about 7-10g; just enough to bring it down. This was in line with the observation from week 4 and 5, that the setup was extremely sensitive to weight changes.
Progress of project for week 7
Goal Of Week 7: Testing of Prototype 1
Description: Prototype 1 was complete. Now what is left with was the experiment.
We put everything we have gathered last week together to build prototype 1. We tested it outside the MNT lab inside SPMS.
As proven by the two pictures above, prototype 1 could sink. The RC worked, the battery worked, the pump was strong (but slow). However, much to our dismay, we found out that the outer balloon (the army trash bag) was deflated when prototype 1 landed. We took apart prototype 1 and spent a few hours carefully taping the mouth of the outer balloon shut and using silicone paste (those found in bathrooms) to further shut it tight.
Prototype 1 was up flying again, but this time, the pump was not able to blow up the air balloon. When we took prototype 1 back down (we tethered it to dental floss as a backup), we realized that the outer balloon (the army trash bag) did NOT deflate. Our seal was good enough, but it was clear the 6v air pump was not strong enough.
At that point, we bought an upsized version of our 6v air pump. A 12v pump at 6L/min. But it was too heavy (an astounding 281) and offered no real improvement over the small pump. Something else had to be the “pump”.
We would spend the the rest of the week deliberating between designs. From a centrifugal-impeller type pump, to a multi-stage compressor type pump.
Progress of project for week 8
Goal Of Week 8: Design prototype 2 (Part 1)
Description: With the failure of prototype 1, the team scrambles to find another solution…
It was somewhere around this time when we were introduced to the idea of using an RC duct fan. Functionally, it was same as an aircraft engine. Suck air in, and blow it out at high speeds to create thrust. The hope was that the air blown out could be strong enough to inflate the inner balloon AGAINST the pressure of helium. It was our first time learning how to use a 3 phase BLDC motor and doing arduino programming; A lot of time had to be spent learning and experimenting and failing. But by the end of the week, the design was pretty much set in stone.
From the left:
- Screenshot of the Arduino code. One problem was that most BLDC motors were controlled by a physical potentiometer. But in the air, it is just not possible. We could use a phone app to control it via bluetooth/wifi, except our range requirement was way to large. We could buy a pre-made RC transceiver, but it was expensive and heavy and difficult to find one. The easiest, cheapest and most effective solution was to have to Arduino act as a faux potentiometer; Starting a RC duct fan immediately at throttle would damage it, furthermore, the ESC had a built it program to prevent it from happening. Using PWM and setting the frequency at 50Hz, We found a way to gradually increase the duty cycle in steps of 1/255 or 0.39%. The problem now is to fine tune the code because different ESCs arm at different low points and different frequencies have different analogwrite ranges (as compared to the standard 0 – 255).
2. Silicone paste proved to be poor at adhesion towards plastic and too viscous to fill in small gaps. We opted to to use steel epoxy coated with specialized plastic glue to seal off the most difficult part of the setup: the area around the nozzle.
3. A picture of the circuit for prototype 2
4. The duct fan proved to be extremely strong and extremely fast in blowing up the balloon. The problem now is dealing with it’s high amperage (25A-35A).
5. The circuit took some time to figure out, but this is the first iteration. A very simple load, power source, program and switch circuit. LIPO provides power, the relays act as the switch, the Arduino provides the signal to the ESC, the ESC controls the active phases of the BLDC motor, and the BLDC motor is the load. Note that there are two relays. The first relay is NOT RC capable, but rather, to support the high current coming out from the LIPO (we could not find a relay that was both RC capable and could take such a high current). The second relay is the one that is RC capable (using RF) and activates the first relay. This way, the high current would not pass through the fragile second relay.
Progress of project for week 9
Goal Of Week 9: Design prototype 2 (Part 2)
Description: With the failure of prototype 1, the team scrambles to find another solution…
Now that the team has completed the circuitry for prototype 2, we decided to go ahead with the more difficult problem; keeping pressurized inside the balloon. We brainstormed for 3 ideas and listed their pros and cons
IDEA ONE: Ball Valve cum Servo Trapdoor
We knew from the start that we wanted some sort of check valve. After much research, we decided on a ball valve. A servo trapdoor was implemented into the design as a means to evacuated air (since the ball valve could only keep air in)
Pros: Mechanically simple, Reliable
Cons: A reliable airtight servo trapdoor would be extremely difficult to manufacture. We would always have to make our own ball valve so its manufacture would be difficult.
IDEA TWO: Solenoid Valve
During our research into check valves, we came across the idea of an electronic solenoid valve. This would be a fantastic idea compared to the ball valve Cum servo trapdoor (Not without it’s downsides)
Pros: Ease of manufacture (we only need literally buy one off the shelf) , Superb sealing (provided the solenoid valve is pneumatically rated )
Cons: Severe nozzle choke. Most solenoids valves readily available (not custom made) were in fractions of inches. The largest we could find was half an inch, which was extremely small compared to the original airflow diamter of the BLDC rotor.
The other con was that an overwhelming majority of the solenoid valves available were made of heavy metals like brass. It took us almost 2 full days of shopping before we found one online that was made of polyamide. It required a full week of shipping to arrive in singapore.
IDEA THREE: HYBRID
It took us awhile, but we decided two combine ideas together. We would use the Ball valve cum trapdoor idea, but we would essentially swap out the servo trapdoor for the solenoid valve.
Pros: Superb sealing from both the Ball valve and the solenoid. We eliminate the severe nozzle choke from the solenoid because now the solenoid only evacuates air.
Cons: Weight increase. We will be swapping out an approximately 30g servo trapdoor mechanism for a 90g solenoid valve. Manufacturing will be slightly easier because now we literally only have to buy a solenoid valve instead of make a servo trapdoor.
NEW DUCTED FAN
with the final setup in mind, the team has decided that the current ducted fan we are using is too old (8 uears old), too heavy, too large, and too weak. We decided to shop online for a new ducted fan.
some of the things we considered were its KV rating. We noticed that the higher the KV, the faster the rotation per volt applied but the lower the torque. In general, the higher the KV rating, the smaller the current drawn and the smaller the motor. Low KV rated motors were huge, drew over 50A of current and produced (on paper) incredible amounts of thrust ( High KV produced sub 600g, Low KV produced over 1.2KG ). We decided that what we needed was high airspeed, low current drawing and low weight.
We settled on a 11.1v 8000 Kv 41mm Aluminium ducted fan.
8000kv would be almost double of the current 4500kv motor
Therefore we can reduce the rotor diameter from 66mm to 41mm, hence significantly reducing weight and buulkiness.
An aluminium housing (compared to the old flimsy plastic one) provided more dynamic stability (no warping of the stator housing and hence no blade scraping)
We ordered it online. The item would take more then a week to arrive…
NEW GRAPHENE LIPO
After some research, the group decided to find a lighter alternative for the lipo battery. We noticed that for the same discharge rate, cell and capacity, the graphene lipo always lighter by about 30%. It was more expensive but we had the budget for it. We bought it overseas
Progress of project for week 10
Goal Of Week 10: Design prototype 2 (Part 3)
Description: With the failure of prototype 1, the team scrambles to find another solution…
While waiting for the solenoid valve to arrive, we decided to do a prototype of the ball valve servo trapdoor idea.
The ball valve was the easiest to prototype. Using nothing but empty plastic bottle, tape and a crushed paper sheet (to act as the ball), we fashioned a functional (but not perfectly airtight) ball valve. Some considerations were:
- Our team realised that the crushed paper ball would never make the seal airtight. We needed a perfect sphere. So we already drew plans to buy a ping pong ball and a rubber o ring.
- The second biggest problem was nozzle choke. There were two severe choke points (refer to the left picture). Due to the venturi effect, velocity would slow down drastically as in expanded out of the choke point. We did not know if the choke point would impede airflow and prevent the balloon from blowing up. Even if the balloon blew up, we were afraid that the chokepoint would cause the ducted fan to draw a lot more current, causing severe overheating and damage the electrical components.
To ascertain the operational effects the chokepoints had on the setup, we decided to put it together and run the experiment. Turns out, it worked. The ducted fan was able to blow the balloon up quickly. It got a little hot, but not too much. In addition, our team deduced that since the ducted fan need only run for a few seconds, overheating would not be a problem.
plus this was a empty plastic bottle, we could always increase the size of the choke point to decrease the choke effect.
The crude paper ball we made was also surprisingly effective. Although air was still leaking, we understood that if we were to use a proper ball with a rubber o-ring, the seal would be better.
TRAPDOOR SERVO
(refer to the picture of the right). This one was a lot more difficult then anticipated. There were so many variables to control
1. The trapdoor (prototyped with styrofoam) could not securely adhere to the servo arm.
2. The trapdoor would not properly seal the opening of the ducted fan. We tried lining it with foam tape, but it proved to be too sticky and messy and unreliable
3. The servo had to be Precisely placed next to the ducted fan with the precise height and precise distance so that the trapdoor can slam shut the mouth of the ducted fan. This was a time consuming task.
4. The servo had to provide a large amount of torque to seal off the opening. A slight push (not visible to the naked eye) would cause a huge air leak. But the larger the torque, the larger the servo
5. Control was a problem too. Our servo could not read its own position and coupled with the nature of our purely on/off relay circuit, it was very difficult to control the servo. We decided to go for a rudimentary timing method. We would measure the time taken to fill up the balloon, then program the servo to slam shut at that mark.
5.1 We would soon discovered that coupled with that unreliable rudimentary timing method, the servo shared the same library resource as the ducted, there fore we had to use a second arduino to link the servo motor too (adding uncessary weight) and use I2C slave-master control to link the two arduinos together (the ducted fan arduino was the master, servo its slave)
The team would waste a few days on this idea before scrapping it. what initially seemed like an easy effective idea turned out to be extremely ineffective and difficult.
HYBRID
With the success of the ball valve and the failure of the servo-trapdoor. The only logical way was to either go for the pure solenoid valve method or the hybrid method. we waited for the solenoid to arrive…
The group wanted to try to solenoid method first. It was the easiest the prototype, and the one with the most attractive weight savings and simplicity (the ball valve parts if 3D printed would weigh upwards of 150g) while the solenoid weighed 80g. Plus the solenoid was literally plug and play. the only concern was the extreme choke of 1/2 inch.
The solenoid valve was rated 12v DC two way drawing 200mA with an orifice size of 1/2 inch. But from day 1, we were beset with problems.
- We could not seem to operate the valve. We ran the appropriate current and voltage through, we head a click noise, but no matter how much air we blew in, nothing was coming out. We contacted the manufacturer but they refused to give us a reply
- The solenoid was not a straight path through, it is actually a path into the hole, through the green box, then out the other hole. This was unacceptable, the chokes were too severe
- The manufacturer lied to us. They claimed on the website that it was suitable for air and water. But after calling to enquire, they admitted to using the same template for ALL solenoid valves.
With these two problems in mind, the group gave up on the solenoid, we needed to find another way…
DR MAD 8000KV 40mm ROTOR Feat. GRAPHENE LIPO
nearing the end of the week, both the new 8000kv ducted fan and the graphene lipo came in. Due to a communication error between members, a wrongly rated graphene lipo was bought (we needed 3 cells but a 4 cell battery was bought instead). We will have no choice but to fall back on the normal 3 cell lipo
The ducted fan was nothing but good news. We quickly tested it and found it out to be far superior to the older one. It only took about 30% of its max RPM to blow up the balloon compared to the 80% max rpm for the ducted fan. it was also far lighter, weighing in at about 60g compared to the 120g. It was solid (no warping of the stator shell) and it had a constant smooth outer diamter, making the design of adapters extremely easy and convenient. This ducted fan will be used in the final product.
Progress of project for week 11
Goal Of Week 11: Design prototype 2 (Part 4)
Description: With the failure of prototype 1, the team scrambles to find another solution…
Right now, our prototype is nearing completion. Everything has been tested and ready to fly. The only bottleneck is the valve. We cannot keep the air inside the balloon! Furthermore, We wanted to find a more presentable and reliable way to seal up the balloon. A lot of time would be spent researching and trying before we would settle on a final design.
The team, deciding that time was running short and there were too many delays on prototype 2, decided to double down and re-design prototype 2 into the final product. Here are certain aspects of the new design. We would spend the whole week measuring and designing. There was no room for experimenting anymore, we had to stick to what we know would work and measure everything precisely.
- Inner Balloon Design
- Lifting Balloon Design
- RC Controls
- Valve
Progress of project for week 12
Goal Of Week 12: Station Keeping
Description: Enough delays, Lets get this thing flying.