The blaster has 12 different firing patterns available, based on both the firing modes and types. The firing modes include the individual left and right magazines, an alternating mode, and a mode where both magazines fire at once. The firing types include single fire, 4-round burst, and fully automatic. The blaster can be in any combination of modes using the user interface on the top of the blaster.

Motors were a very light press fit into the flywheel

The intended controller for this project was a custom PCB with an STM32F411CEU6. However, there were numerous problems with this circuit board. Firstly, during the creation of the custom footprint for the beam breaks, the emitter diode pins were swapped, causing them to not emit at all. Additionally, the voltage divider of the buck converter had to be adjusted as it was originally outputting 6V rather than 5V. This is where the known issues end, and the unknown issues begin. 

We built two custom MCU boards, each a duplicate of the other. However, they each had different issues. On one board, the 5V and 3.3V power lines were somehow tied together, likely internally through an IC. Likely culprits include 4 mosfets used to light LEDs and a few level shifters used to convert 5V signals to 3.3V for the MCU. The other board had functioning 3.3V and 5V lines, but the MCU became extremely hot when powered. No solder bridges were visible, so we swapped the MCU on this board, thinking that the original may have been damaged. The second MCU also became extremely hot, indicating that our short circuit was somewhere else on the board.

Having shorted 3 different MCUs, we decided to cut our losses and used a BlackPill development board instead. This caused us to have to remap all our pins to allow for clean(ish) wire routing. The beam break sensors received a new mounting system and then wires had to be hand soldered to each as shown below.

Ultimately, the hand soldering of the system using a BlackPill resulted in a functional final project, but not an exceptionally robust one. With many wire connections being made purely with solder and no other mechanical connection, connections break with extended use due to the violent vibrations of the solenoids. We also continuously had problems holding the beam break sensors in place. They kept wanting to lift up and away from the flags attached to the solenoids.

This 24V board worked exceptionally well for the application and allowed us to drive the solenoids at an extremely fast rate of fire. This is done through “Solenoid Clamping” which rapidly decays the current generated by the solenoid when returning using the spring. This allows the solenoid to return much quicker than if the current was just freewheeling through a diode. The diagram below shows how this H-bridge configuration works for solenoid fast discharge circuits. 

Here's what the front of the solenoids look like. The aluminum flags trigger the beam break sensors mounted above

Software

The blaster has 3 different firing types: Single, 4 shot burst, and Full Auto. These are complemented by 4 different firing modes: Left, Alternating, Both, and Right. This results in 12 different firing patterns. The Left and Right modes only use the left or right solenoid respectively. The alternating mode switches between the left and right solenoid every shot, and always starts on the opposite of whichever was last fired. The “both” mode always fires both solenoids at the same time, resulting in two darts launched per actuation. These modes and types are displayed on an individually addressable LED grid, which shows the currently selected mode and type through a crosshair.

The blaster is able to execute each firing pattern due to the use of the beam break sensors. These give feedback to the microcontroller which allows it to only actuate a solenoid when it has completed an entire stroke. This way the control of the solenoids is not dependent on the time constant of the solenoids, which changes based on friction or battery voltage. The friction is an especially important aspect, since the force from the spring loaded magazines changes based on how many darts are loaded.

Our software was primarily written in C++, with the use of the RC receiver class we made in a previous lab. The logic to control the firing patterns is implemented through multiple finite state machines (FSMs) that use a combination of timing, a dart counter, user inputs, and the beam breaks to trigger the transitions that fire the darts. 

Debounce was an issue we ran into several times throughout the project, which required additional logic in the software to account for this. Originally, we had implemented debounce circuits into our circuit board for all of the various mechanical switches in our project (5 total). However, we did not have said debounce circuits in our switches after switching to a BlackPill development board, so we used state machines with flags and measurements of the elapsed time in order control both the rising and falling edges of signals, primarily our mode and type selection buttons. This implementation is shown in Figure 12. We chose to prioritize input latency, so the debounce is triggered on any rising or falling edge, and the filtered signal is forced to that state for a period of time before it is allowed switch again. The trigger uses a 10ms debounce, while the user buttons use a 250ms debounce.

This project overall was fairly successful. We managed to make a complete, 2 fully featured blasters with the intended 12 different modes of operation. The overall look and feel of the blaster exceeded our expectations. However, it was quite disappointing that our custom MCU circuit boards did not work, especially since we had invested quite a bit of time and money into them. The biggest consequence of moving away from the custom PCB was the improvised methods for mounting various components, especially the beam break sensors. The beam break sensors are truly meant to be mounted to a circuit board, there are no holes or intended mechanical methods to mount them. We used a clamping mechanism to try and secure them in place, however due to the extreme vibrations of the blaster the beam break sensors continually lift themselves out of alignment. This prevents the blaster from firing, since the sensors are required for actuation.

Additionally, the potentiometers used gave us quite a bit of trouble. On a test breadboard setup, the potentiometers would work perfectly. However, after soldering into the blaster, they would become unreliable or sometimes outright unreadable. We believe this may have been due to the heat of the soldering iron affecting the internals of the potentiometers. 

Overall, we are proud that all of the functionality works for our project, however it lacks the robustness for continued use or demonstrations.

The following references were exceptionally valuable in the software and electrical design of our project: 

Interfacing with WS2812 LEDs on STM32 - ControllersTech 

Using Motor Drivers to Drive Solenoids - Texas Instruments

BL Heli S Firmware Manual