Penny's Computer Watch (from Inspector Gadget)

Here's how I made a replica of Penny’s video watch from Inspector Gadget. I 3D printed the case, buttons, and strap, and wired up the circuit using an Arduino Nicla Voice.

Here are all the parts and tools I used to make this project:

Parts list:

• Arduino Nicla Voice
• USB cable (micro B)
• Adafruit 1.69" 280x240 Round Rectangle Color IPS TFT Display
• 30AWG stranded wire
• Lipoly battery
• Switch
• 3D printer filamentPLA matte dark blue
• PLA red
• PLA neon green
• Watch stand

Tools:

• Soldering tools and supplies
• Multimeter
• 3D printer
• Tweezers
• Kapton tape

I made a build video for this project, check it out!

To put together this circuit, I used really thin 30 gauge silicone-coated stranded wire. This is a very small project, and more difficult because of it. I used tweezers to help me solder the wires in place. 

I followed a diagram I made for myself to be sure I was hooking everything up correctly, and not trying to juggle pin numbers in my head.

The TFT display needs a bunch of wires to communicate with it and its onboard SD card, so I couldn’t test it out until I connected all of them, only to find out that I had broken my display. These things are more fragile than I thought. So use my experience as a word of caution, don’t press on these screens to get them into or out of an enclosure, and don’t clamp them in your helping hands. I was more careful with the next one.

I touched up some screengrabs from the show for the images on the watch and loaded them onto the SD card.

To model the watch in 3D, I used the free browser-based software Tinkercad.  I created the main body shape to fit the components and found a watch band design on Printables that I incorporated into the design. I made a smaller version of the watch band too, because the default size was too big for me.

The dark blue case is made of two pieces, then a neon green set of buttons fits inside the front piece. The red band is attached to the body using a piece of 3D printer filament of just the right length.

I used hot glue to both affix the Arduino Nicla in place and separate it from the TFT board so nothing shorts out.

The watch is meant to be battery-powered, and the Nicla docs recommend a lipoly battery with a capacity of at least 200mAh. I carefully stripped and tinned the leads of my battery and soldered it to a small switch, using heat shrink tubing to insulate the connections before attaching the battery and switch to the Arduino Nicla.

I used a piece of Kapton tape to insulate the battery from the display contacts. Mostly any tape would work here, but Kapton is designed for this purpose and I happen to have some on hand.

Over in the Arduino program, there are a few extras required for the Nicla to work how we’ve got it. First, the Nicla system must be included, even if you are just testing some other sketch, or the board will go to sleep after a few seconds. Second, the battery charging must be enabled in the setup.

With those small additions, I could test the library sample code for displaying images from the SD card and have a looks-like prototype.

Next up it’s the voice control. I followed the online instructions to update the firmware and upload the Alexa Demo via the command line, then uploaded the AlexaDemo Arduino sketch, and was able to get the board to successfully recognize the wake word.

So I moved on to creating a custom wake word– I want Penny’s “come in, Brain” to summon Brain’s image to the screen. I followed Arduino’s IoT Cloud instructions to create a new model using the Edge Impulse machine learning service. Basically, you connect a recording device, either your phone, computer, or the Arduino Nicla loaded up with the right sample sketch, and input a series of audio samples that represent your wake word. I used my own voice to say the phrase, and also recorded Penny’s sample from my computer book project from various angles and positions. Then I classified these samples with the label for my wake word, and recorded another set of audio samples that do not contain the wake word, and labeled them noise.

Then I was able to train my model and deploy the model by downloading the files and setting them up on my Arduino Nicla. One gotcha on Mac that isn’t heavily detailed in the instructions is the need to involve your security preferences when running apps via the command line. I had to execute most things twice, once to trigger the security feature, then I’d click “Allow Anyway”, and run it again.

Here’s the finished project. It charges over USB, and loads a new picture of Brain whenever it recognizes Penny’s iconic phrase.

This project was on display at the Digi-Key booth at the 2023 Maker Faire Rome, along with Penny’s Computer Book. Head over to my other post for an overview of the whole costume including details on the cosplay wig and costume elements.