Drill Powered Bike

 As a mechanical engineer at CU Boulder, I have the great opportunity to take a class known as Component Design. This class is structured around a group project where teams build a drill-powered bike with the goal of competing in a race against one another. There are several rules that constrain the designs, such as no battery swapping during the race, a budget of $200, a weight limit of 50 lbs, and no modifications to the drill. Besides this drill, teams have to source their own materials and parts while staying under budget. Additionally, there are three race events to choose from: endurance (most laps in under 30 mins w/one battery), hill climb (fastest time to climb a designated incline with a certain amount of added weight), and maneuverability (fastest to make it through an obstacle course). 

Our team decided to do the endurance race and we each got designated roles. I chose to be the CAD engineer but was heavily involved in the manufacturing process because of it. I designed our bike with inspiration from a Razor electric scooter because I had ridden it before with a ride-sharing service and felt like the proportions were very ergonomic. 

Initial sketch

                             Exploded view of final assembly

I used an image of the Razor scooter I found online and scaled it up accordingly so that I could import it into SolidWorks. Then, I used the wheelbase and handlebar angles to create a rough model for the frame. Due to its price and easy weldability, our group decided to 0.065" wall square steel tubing for the frame members. The wheels, seat, and handlebar/steering assembly were sourced from a used bike parts bin that was provided to our class. With these components, I was able to integrate the steel tubing into CAD and design a frame with enough clearance for the drill, wheels, and steering tube. With the CAD being close to done, I generated a cut list for the frame so that our team could start the manufacturing process. The class requires drawings for almost every manufactured component so I learned how to properly tolerance items and what dimensions needed to be called out vs. suppressed. 

Frame with drill plate mounted

Frame drawing

Besides the welded steel frame, the other main component of our bike was the drive system. It consisted of a water-jet cut drill plate with various holes and cutouts to mount our drill (with hose clamps) and pillow blocks (holds axle and sprocket) along with frame mounting points. We used this plate to make our frame more rigid as it was bolted to various members. 

Drill plate drawing with revisions

The drill plate was by far the most complex part of our bike due to the spacing and mounting considerations in order to get the sprocket on the drill axle to line up with our bike's wheel sprocket. In the end, we were happy with how it came out, only needing minor modifications to work properly.

Drill plate assembly

For the most part, the build was pretty simple. The only thing that cost our team a lot of time was our coaster-brake wheel that the drill drove via a chain. We decided on using a coaster-brake hub because our wheel size was 18" and most kids' bikes did not have a normal gear cassette. Coaster brake hubs allow for the brakes to be applied to the rear wheel when pedaling backward. The internal mechanism is more complicated than you might think as it uses a spring, clutch, and threaded collar to accomplish this. The original rear wheel with the coaster brake hub in it that we used was taken apart when we found it in order to take the existing sprocket off. This was a big mistake because it seems like we assembled it incorrectly or the part was faulty to begin with and this caused a lot of headaches during the build process. In the end, our team just simply bought another coaster brake wheel from a local used bike shop and it worked as intended. It allowed our bike to coast when the power was off but then drive the wheel when the drill was turned on. 

To achieve the gear ratio that we decided was adequate (about 4.33:1), we needed to replace the sprocket on the kids' bike wheel with something a lot bigger. We found a 39T bike spur gear in the spare parts bin and decided to use it with the 9T sprocket that was mounted on the drill plate. The only issue was attaching this larger spur gear to the smaller hub of the coaster brake assembly. We decided to use the existing spline that the smaller outgoing gear had used in order to mount a larger plate. After measuring the 3-bump spline profile dimensions, a member of our team put it into CAD and had this plate water-jet cut. It included five mounting holes that the larger spur gear would bolt into and would be secured onto the coaster brake with a snap ring. With a little bit of filing and sanding, we were able to persuade this adapter piece to properly fit onto the coaster brake hub. 

Speaking of power, we used a standard bike brake cable with a hinge and rubber band to activate the drill's trigger. We originally wanted to use a metal spring but the one we had proved to be too weak to pull the hinge back after the user pressed the brake cable throttle. The rubber band worked well and allowed me, the designated driver, to have variable throttle control during the race. In hindsight, it might have been a lot easier to just use a set of brake calipers to engage the throttle rather than going with the hinge design. 

During the final assembly, a few washers were added to make sure the sprocket on the drill axle lined up with the spur gear on the rear wheel so that the chain didn't flex too much. This bike was working and now all we had to do was test it. After some testing, we realized that the drill-axle combo was moving around a bit due to vibration and thus causing the sprocket and spur gear to become misaligned. This often resulted in the chain coming off or the drill seizing up. To fix this, we salved some shaft collars from my roommates' drill-powered bike from the semester before. By placing these and some nylon washers between the sprocket and pillow blocks, the problem was fixed and the chain remained roughly in alignment. 

Along with the frame, I also designed the dropouts for the rear axle. These were water-jet cut out of 1/8" steel and had a lot of room to adjust the rear axle location in order to get the chain tensioned properly. After adding the proper hardware, the dropouts held tension pretty well. 

Our final testing after fixing all of the chain alignment issues included dialing in the correct drill torque setting. We needed our bike to be able to climb the inclines on the course without overdrawing current from the drill but also be fast enough to speed up on straightaways. It took a bit of trial and error but we pretty much went with the max torque setting in order to get up the hills and decided that I should try and coast as much as I can on the straightaways to save battery. 

On race day, I was a bit nervous that something small would go wrong and sabotage the event. However, our bike performed very well and had to stop due to a low battery (which is what we wanted to happen). Plus, we put a speaker on the handlebars so our bike was a crowd favorite. The one thing I think we should have done was integrate a drill setting change lever. Essentially like a gear shifter, changing the drill torque settings while racing would have been a lot more efficient. While our bike made it up the inclines with no issues, it was slow when it reached flat ground. We played around with more drill settings after the race and managed to make our bike go pretty fast on flat ground. 

Me riding the beast during race day:

Although a drill isn't the ideal motor to make an electric bike, this project taught me a lot about designing for manufacturing, sourcing parts, manufacturing, testing, creating engineering drawings, and teamwork. Component design has been my favorite class so far as an undergrad at CU Boulder and I definitely think it made me a better engineer. 

See our project report here: https://docs.google.com/document/d/e/2PACX-1vT3_HnMEKLe94v-CIfJtcIcVYe_P1eczUPctAy4cKxC89i7UhKW2cjwsy3dQyRuNQKgFslAL-t0cv3t/pub

A Lot of Boring Work

Digging tunnels might not seem like the most cutting-edge industry, but the need for rapid and precise boring has become a recent hotbed for innovation. Most of the momentum in the space can be credited to the Boring Company, one of Elon's many "side projects." After joining the CU Hyperloop club at my school, the University of Colorado Boulder, I learned about what goes into creating a tunnel boring machine (TBM). As part of the Boring Company's Not-A-Boring competition, universities and hobbyist groups from around the world design and build a TBM with the goal of digging a tunnel the fastest. For the last competition, the tunnel had to be 30 meters long and half a meter in diameter. 

There are many aspects that go into building the machine that builds the tunnel. Our team was split into subteams such as Excavation, Soil Removal, Tunnel Support, Propulsion, and many others. I joined the team as a freshman and quickly was thrust into the design process. Our club had roughly a year to design, build, and test our machine before the competition. I took interest in the Soil Removal and Tunnel Support areas, becoming the lead for both subteams after one semester. 

When dirt is excavated by the cutting head, it has to be processed and sent to the tunnel entrance. This encompasses the Soil Removal subteam. For this competition cycle, our team decided to take the slurry approach. Water is introduced to a soil chamber directly behind the cutting head and then pumped by a sewage pump to the tunnel entrance. There was a lot of design work that I did in terms of figuring out how to integrate soil mixing and processing mechanisms. One example of this was the agitation rods attached to the back of the cutting head. Although a lot of time and effort went into the soil removal system, most of it was changed for this year's new machine due to clogging issues with the pump and large water requirements for the slurry. 

I believe my most significant contribution to the team came with my idea of implementing a continuous and flexible tunnel support system. In conventional TBMs, heavy concrete pipe sections are slowly placed into the bored dirt to support the weight of the soil. This process not only increases dig time but also is very expensive. Since our club did not have a large budget, I knew the solution had to be cost-effective. After a lot of discussion and brainstorming, I had the idea of using a "dog-tunnel" mechanism. As the TBM progresses, the tunnel support structure unfolds behind it. This allows for continuous and rapid excavation. For the to work, the support structure is held in place at the tunnel entrance by stakes. 

To keep costs low, this support structure was constructed out of heavy-duty poly-tarp and 1/2" steel rings. We made these material choices after conducting a quarter-scale centrifuge test. The test allows for a scale model to be subjected to the appropriate pressures by submerging it in a soil box and then spinning that box in a centrifuge. All the materials are kept consistent in this model, but all dimensions are scaled down by a factor of four. By doing some rough calculations, we estimated the support structure would need to handle 25 kPa of vertical pressure and 12.5 kPa of lateral pressure. To our surprise, the centrifuge test revealed an overall factor of safety of 3 (the model didn't reach failure but the centrifuge couldn't spin any faster with the given load). 

At this point, we also realized that this data was for a fully tensioned support section. Therefore, the tarp and ring segments needed to be stretched before making contact with the soil. This realization lead to the development of a release mechanism that incorporated several linear actuators along with a stepper motor-driven carriage. Essentially, the actuators would hold two rings in place while the carriage stretched them apart by traveling on a leadscrew. 

Unfortunately, financial issues caused our club not to reach an adequate budget. We didn't have funding to fully complete our TBM. But, we decided that it would still be good to conduct some tests at the competition. However, we were proud that our design proposals and technical documents submitted to the Boring Company put us into the "Digging Dozen" - 12 teams out of about 500 submissions that were invited to Las Vegas for the competition. Most teams also ran into funding issues and only one team ended up digging.

The main test we wanted to run was with the tunnel support structure. Essentially replicating digging conditions and putting a portion of the structure into the ground at an appropriate depth to stress-test it. Over the summer of 2021, the manufacturing of the support structure began. It was a back-breaking task. For our 10 meter long test segment, 1/2" steel rods needed to be bent into rings with a 0.5m internal diameter and then sewed into pockets on the tarp sections. With our funding problems, everything was done in-house. However, we were able to finish the segment in time for the competition. 

While in Vegas, I lead the efforts for conducting this stress test. First, a long sloping channel was dug by site staff to place the segment into. I designed a staking mechanism that used rebar, a plywood sheet, and several steel pieces to ensure that the support structure could be fully-tensioned while under soil load. The next issue after situating the segment was data collection. After some team brainstorming, we decided to build a sensor carriage that would be inserted into the support section and slowly pulled out with a pulley system. This system would collect one data set before soil loading and then another after loading. LIDAR and ultrasonic sensors on this carriage would record the distance to the top of the tunnel so that a deformation value could be computed. 

Although very time-consuming, the data from these tests almost perfectly aligned with our earlier centrifuge test. My support structure held up very well against the immense soil pressures and proved to be a good alternative to the conventional rigid segments. However, for this year's competition, there are some vital improvements to be made. For one, the release mechanism needs to be simplified and tested. Additionally, the problem of settlement, soil shifting after subsurface excavation, must be explored. All in all, I think that this experience was extremely valuable. As a team lead, I was honestly very surprised to see my novel, cost-effective design solution work on the first try. There is still a significant amount of work to do before the system is implemented in an actual TBM, but I think it's a pretty good start. See pictures taken throughout the competition cycle below. Click here for a cool centrifuge video. 

ANSYS FEA of Initial Tunnel Support Ring

1/4 Scale Centrifuge Model in Soil Box

Soil Box Mounted in Centrifuge with GoPro for Data Collection

Preliminary Release Mechanism Sketches

1st Iteration of Release Mechanism CAD

Tunnel Support Overview

Competition Tunnel Support Test Overview

Me (camo pants) explaining the Tunnel Support System to Boring Company Engineers

Tensioning and Staking the Tunnel Support Section

LIDAR and Ultrasonic Sensor Carriage System

RC Transmitter CAD Project

 As part of my SolidWorks CAD class, I was instructed to model an item of my choosing for the final project. I chose my drone RC transmitter. After doing some preliminary measurements and sketches, I started to realize how complicated the design actually was. There were a lot of complicated curves, fillets, and ridges in the ergonomic backplate of the controller. Additionally, I wanted the model to have functioning joysticks, so I had to create the appropriate mechanisms. My approach was to simplify everything as much as possible but still retain the key details. I decided not to model the internal electronics because I wanted my model to focus on the mechanical aspects of the design. I took the controller apart so that I could see how each component was manufactured. Furthermore, I needed to see how the joystick mechanics functioned. 

Since I'm an engineering student, I pretty much put this project off until the last week of the semester. During the three to four days that I worked on the model, my sleep schedule was pretty much nocturnal. However, after a lot of hard work and SolidWorks bugs, I finished everything in time. I'm very happy with the final product and it was cool to see the entire assembly come together. See some renders below: 

No-Sweat Biking

After realizing that skateboards are pretty damn dangerous (following my e-board build), I decided to try and make my own e-bike for the lowest amount of money possible. Although an electric skateboard may have a smaller footprint, I think that e-bikes are a lot more practical. For me, I can actually ride a bike around town without getting thrown off when the smallest of pebbles get lodged under the wheel. Additionally, the area behind the rider on a bike can be used for extra storage. 

There are a multitude of electric conversion kits for standalone bikes, but most overcharge for basic components that can be found off the shelf for lower prices. An e-bike is extremely simple. The whole system consists of a motor, battery, motor controller, and other input devices (throttle, brakes). My first decision was about which motor type I wanted to use. In general, the two main motor variants are front/rear hub and center drive. To minimize cost, I decided on a front hub motor. Although not the most powerful, a front hub is the easiest to integrate (no need to modify pedals or gears). After browsing around on eBay, Amazon, sketchy Chinese sites, Craigslist and Facebook Marketplace, I found someone selling a used e-bike kit for a lot lower than market value. The setup included a 36V 350W front hub motor, motor controller, and various accessories such as a twist throttle and e-brakes. In hindsight, I would've liked a slightly more powerful motor, but it still works well for cruising around town (for reference, 72V 1000W motors are essentially the high end of the motor spectrum in terms of power). 

Next, I started thinking of the battery setup. I knew that I needed a pack to suit the motor: supplying a nominal voltage of 36V and an amperage of around 9-10A (350W/36V). From experience building drones and model airplanes, my first choice was a lithium polymer battery. Although widely available, these batteries tend to get expensive when it comes to voltages as high as 36V (most drones run anywhere from 7 to 20ish volts). Also, LiPo batteries tend to wear quickly after many discharge/charge cycles. Because of these factors, I decided to use 18650 (18mm diameter, 65mm long, 0 for circular shape) lithium-ion batteries. These look like larger AA cells and can be found in most older laptops. Even Tesla uses li-ion batteries in their cars (although not 18650s specifically). Depending on the manufacturer, 18650 cells can last a long time and deliver a large capacity in a relatively small package. 

As I soon realized, 18650s aren't the cheapest batteries. That's when I made the decision to harvest old laptop batteries. While this can be dangerous (old cells typically aren't safest), you'd be surprised at how many perfect cells can be recovered. The only real downside with using these recycled batteries is the time it takes to check each cell. I also found it hard to get my hands on old laptop batteries as most people I knew had thrown their computers out (except my grandparents, where I got the majority of cells). After deciding on a battery layout consisting of 50 cells, a 10s5p setup, I began the long process of acquiring the cells. For the pack design, 10s5p means that there are 10 cells in series (3.6V nominal voltage of 18650s x 10 = 36V and 2000mAH capacity x 5 = 10AH - most smartphones are around 2-3AH). To test each cell, I used my drone battery charger to charge and discharge the batteries. I removed any cells that started to overheat during the process and also observed the cell voltages following charging after multiple days to make sure the batteries held their voltage. I also purchased a BMS (battery management system) that would allow me to charge the cells safely and make sure none of them burnt out. Most li-on packs are assembled using nickel strips and a spot welder to make connections. Without a spot welder and a large budget, however, I decided to use a pack assembly kit. The kit was made up of plastic battery holders with screw threads at the terminal ends so that cells could be connected by bridging these threads with nickel strips (and securing them with small bolts). Following a long process of building and checking connections (nearly ruining the whole setup by shorting two cells with a screwdriver), I wrapped the pack in several layers of foam and finished the whole thing off with some gigantic heat shrink tubing. 

Ironically, the bike was the last part of this build that I got. I managed to pick up a decent hybrid for free from a friend that was moving. After testing all the electronics, I started mounting them on the bike. I noticed how the provided e-brakes not only clamped down on the wheels when triggered but also sent a signal to the motor controller to stop the motor. Although the wires running along the frame gave it away, the front hub blended in perfectly. I rode the setup around my neighborhood and got up to 15-20 mph with minimal pedaling. Following some cable management and minor modifications, I started going on longer rides for range testing. I managed to go about 20 miles with not much physical exertion and still had about 25% battery left (voltage in this battery goes from 42V fully charged to 30V completely drained). So far, I've had no major problems with this build and it works perfectly to get around campus (surprisingly, I still have some in-person classes in the age of Zoom). The only real negatives come mostly with the bike itself. The caliper brakes aren't the best (looking to switch to disc later) and the gears don't change very fast. Other than that, the battery is a bit heavy but does provide significant range. 

In the future, I may swap in a more powerful motor, improve the braking situation, add more weatherproofing, and attach a key-start system. Not to mention, the cheap charger I bought doesn't really display charge percentage/rate so it's slightly inconvenient. Also, I had an idea to add a generator to the pedals so I could charge the battery when pedaling (although there might be too much resistance/weight). Overall, this project went a lot better than I expected and I'm genuinely satisfied with the final outcome. See pictures below. 

Motor Controller mounted below the seat

Integrated Motor in Wheel Hub

Fully sealed Custom Battery 

A Smart Garage Door

Instead of acting like a normal person and just placing a spare key outside my house for the times that I get locked out, I decided to make my garage door "smart." By that, I mean integrating it with an Arduino-controlled motor relay hooked up to a server via WiFi. I had an old Arduino Uno sitting around and decided to put it to use. All I needed to link the Arduino and my garage door together was a motor relay. A relay basically acts as a switch that can open and close a circuit when instructed to by a microcontroller or some other device. I discovered a free Internet-of-Things service for these types of projects called Blynk. The service provides pre-written Arduino code and an ecosystem for controlling the whole thing with a smartphone app. After loading up some of the sample code and hooking the relay up, I followed the simple steps for setting up the Blynk app and attempted to test my over-engineered garage door opener. With my phone connected to a network, as soon as I pressed the button in the Blynk app, I could hear the clicking sound of the relay opening and closing. Next, I found an enclosure for the Arduino and relay while also running wires from my garage door motor. After doing some cable-management and putting the wires in their correct locations, I tried opening my garage door from the app. On the first try, the door opened, but as soon as I released the virtual button the door shut immediately. In the Blynk app, I changed the button to a switch, so it could remain open without constant pressure. This change worked and marked the end of this project. I also discovered later that Blynk has its own cloud server, meaning that you don't have to be connected to your home's network to open the garage door. Theoretically, I can open my garage door from across the world, assuming I have some sort of connection. Recently, I also found out (around 2am) that when power is cut to the Arduino, it closes the relay, triggering the garage door to open. That's really the only problem with this project so far. See pictures below. 

Control UX

Relay Connected to Arduino Board

Component Enclosure

Relay Wires Connected to Existing Opener

Laptop Surgery

Oftentimes when a piece of technology breaks, people throw it away or never use it again. I've done this before, especially with old smartphones. While cleaning my room, I found an old laptop that I used in middle school. At the time I used it, around 2015, it was new. It's an Asus Transformer Book T1000 with a detachable keyboard. Perfectly usable today, albeit slightly slow. The screen is cracked in the corner, but the actual LCD panel functions properly. I realized that I had stopped using this laptop because the power button broke. It couldn't be turned on. I decided to take apart the Transformer Book the find the root of the problem. After removing the back plastic cover on the tablet portion, I located the circuit board for the power button. The internal button had snapped off, leaving only a metal contact on the board. With the cover off, I used my screwdriver to short the contact, resulting in the normal boot-up sequence. The power button board simply had a few screws to secure it and a standard ribbon cable to connect it to the motherboard. I searched eBay for a replacement power button board and found one for around $10. Once it arrived, the repair procedure was basic. Everything worked like new. As technology has become more advanced, it has also become less repairable. Companies like Apple have long restricted consumers from taking apart their devices for repair. The idea of planned obsolescence (tech intentionally made to require a replacement) has only spread. The tech industry, with many parallels to the automotive industry, isn't favoring the true enthusiast anymore. For example, if my laptop was a newer MacBook, it would've been a lot harder to find a power button module on eBay. To achieve a sleeker form factor, companies have opted to solder components directly to the motherboard and adopt proprietary designs. This recent trend is not only detrimental to the wallets of consumers in the case of a repair, but also to the environment. The life cycle of technology is decreasing, resulting in more waste. It's probably too late now, but tech companies should begin to support the tinkerer. 

Mousetrap Powered Vehicle Challenge

In a recent project in my physics class, our teacher instructed us to build a mousetrap powered vehicle that can travel at least one meter off of a table. We were only allowed to use basic materials and all the vehicle's power had to come from a mousetrap. The hard part about this project was finding a design. Most high school mousetrap car projects focus on a simpler distance-based challenge, where students build vehicles that can travel for the most distance. There are plenty of these distance designs on the Internet, but not many for the table style challenge. Instead of building a traditional mousetrap car, I wanted to do something different. I felt that a regular car would either break on impact with the ground or have a hard time maintaining momentum after the drop from the table. In my design process, I attempted to come up with the simplest design I could that had the lowest amount of moving parts. In the end, I came up with the idea to attach a mousetrap to the bottom of a ball and basically just launch it off the edge of the table. The potential energy stored in the mousetrap would be used to make the ball jump off the table where it would roll on the ground. I settled on using a ball made of floral foam that I bought at Target. The foam was stiff but easy to cut with a knife. I first began carving out space for the mousetrap to sit in. The difficult part about this process was making sure that the mousetrap was fairly flush with the ball so that the roll wouldn't be affected too much. It took some sanding to get the perfect fit. Next, I used superglue to secure the mousetrap. I tested my contraption at school the next day. It worked pretty well, but on my last test run, the mousetrap fell out of the ball. I decided to add some hot glue along with applying more super glue. Additionally, I applied a lot of duct tape around the circumference of the ball to make sure the mousetrap never popped out again. I did some more testing and my revisions seemed to have been successful. On the actual day of the competition, my ball went about 2.5 meters (measured from the edge of the table to the closest side of the ball). Most people in my class were surprised to see a foam ball brought into class when they had all built traditional mousetrap cars. Click here for photos.