Build Instructions

Collapse

• 1

  Programming the NodeMCUs

  The writeups here are phrased more as project logs than directions for replication. I'm not using the logs section because I can't insert logs back in time to patch steps I forgot to document in the build process. Nevertheless, there should be enough detail here to make the buzzers yourself.
  

  The NodeMCUs arrived in about 14 days using AliExpress Standard shipping. Upon receiving them, I set up the Arduino IDE to be able to flash them using mainly this video tutorial. I read through this guide to get up to speed with its capabilities. The documentation and examples on github were quite clear. I also added the arduinoWebSockets library (Sketch > Include Library > Add .ZIP Library) so that the program could communicate with the desktop application I had previously coded over websockets. Websockets provide a persistent connection over TCP. Data sent over websockets theoretically has a lower latency than an HTTP request with the same payload, since a persistent connection is opened, though I haven't tested it for this application.

  My initial attempts to flash the NodeMCU were met with the error
  

  error: espcomm_open failed
  error: espcomm_upload_mem failed
  Some quick googling indicated that I had to enable flash mode on the board: hold the flash button, tap the reset button, and finally release the flash button.
  

• 2

  Breadboarding

  Annoyingly the NodeMCU required 2 breadboards to mount. I hear that this a problem with v3, as v2 has a smaller form.
  

  The first problem was finding the laptop's IP address so that the board could talk to it. I looked into a few protocols for device discovery on a local network, including mDNS, UPnP and SSDP, but couldn't find a drop-in library for a C# application. Instead, the desktop program multicasts UDP packets every 2 seconds to advertise its IP address and the port of the websocket server (inside the packet contents). When the NodeMCU receives a signal, it connects to the websocket at the sender's IP address and the port specified. Any subsequent messages are sent over this persistent connection. Afthe NodeMCU connects to the websocket server, the LED on the board stops blinking to indicate successful setup. Note that the PC and board have to be on the same network for this work.

  A push button was connected to GPIO pin 12 (D6) and configured to use an internal pullup resistor. Using a pullup meant that the button pressed state was logic low and released state logic high.
  So that the board could recognize both short and long presses, I opted to use polling logic over interrupts. This method is easily extensible to record double clicks as well. While interrupts would allow the ESP8266 to sleep, this application is unable to take advantage of the low power modes since all of them disconnect from wifi. Reconnect time is on the order of seconds. And polling is easy to implement.
  While a typical mouse click only records a click after the button has been released, the buzzer records one as soon as the button is down, to reduce perceived latency. A long press is recorded if the button is held down for more than 2 seconds.

  My goal was sub-100ms latency from a 10m distance so that teams further away don't appear on the screen after teams close by that buzzed in later (wireless mice seemed to have this problem). Preliminary tests in a WiFi-congested environment indicate a round trip of ~60ms, well within the project constraints.

  See project files for the NodeMCU code, and the github repo for the desktop app code.

• 3

  Designing the button cap

  An important part of the quiz experience is smashing the buzzer button when you figure out the answer. Typical 10mm diameter buttons weren't going to cut it because they're too small for the satisfaction. Sizeable arcade buttons weren't available cheap locally or online (except, of course, the trusted AliExpress). I didn't want to wait another 2 weeks for shipping, so I decided to 3D print a 60mm diameter button cap for a momentary push button I found in the lab.

  It was my first time 3D printing using a Makerbot Replicator+ at the university. I read several articles on the excellent 3dhubs.com to understand how FDM printing worked, as well as techniques and common pitfalls. Modeling using Autodesk Fusion 360 was fairly easy to pick up using this tutorial and simply playing around. It took 2 prints to get the tolerance for the cylinder fit right (0.2mm greater than button diameter).
  There fillet at the base of the inner cylinder makes it a little less prone to tear off. The other fillets and chamfers are just for aesthetics :)

  
  In the first prototype the cap height was about 6 mm, whereas the final version is about 10 mm. This is due to changes in the lid design, which I will detail in the next step.
  

View all 10 instructions