I once spent a summer working at the Winchester Science Centre. I provided the electronics for an installation sponsored by the Electrical and Electronic Engineering department at Southampton University. The installation aimed to explore using electric vehicle batteries to help stabilize power supply during periods of high demand. Although the concept was straightforward, explaining it to the public was more challenging. One requirement of the research grant was to inform the public about the research goals and potential benefits to the UK. The Winchester Science Centre was tasked with creating an educational experience for the public and delivering the information.
I was responsible for creating the electronics for a basic grid-balancing game. The game features a large circular table with a five-foot metal electricity pylon. At the pylon’s base, four distinct segments are marking industrial and residential zones. Simple images represent these segments. Surrounding these segments are eight active zones showcasing different types of power-generating plants, such as coal and gas. Interspersed among these are models of residential areas with clearly defined car parking spaces. The user interface for the power stations consists of large metal spinner wheels. The residential areas have spaces where a model car can be parked. Each is equipped with an optical reflector sensor to detect the presence of the model car. Additionally, the spinner wheels incorporated a basic rotation sensor created from a hall-effect switch with a magnet attached to the spindle.
The input devices were used for the game. At the top of each pylon, there was a rotating “clock” showing “peak” and “off-peak” demand times. There were four identical units, each with an addressable LED tape below the clock. This was used to indicate the degree to which the power grid was out of balance. The game involved turning power stations on or off to see the effect on the grid balance. If the LED lit up red and was too far to the left, the grid was under-supplied, and another power station needed to be turned on. If it was too far to the right, there was an over-supply, causing the lights in the centre to flicker. The goal was to keep the central green LED lit, which showed that demand was balanced by supply.
The clock ticked slowly the entire time. When the pointer switched from ‘peak’ to ‘off-peak’ or vice versa, the grid demand changed, affecting the balance. This required power stations to shut down or start up.
The educational aspect was improved by placing one or more cars in the parking spaces. These cars would provide power to the grid when there was not enough and receive power from the grid when there was too much. The cars could be added or removed depending on the demand and the activity of power stations. It turned out to be quite an intense game, somewhat similar to the work of operators who manage our electrical grid!
The power stations were illuminated by small aluminium PCBs filled with high-power white LEDs. When a power station was activated, the light intensity increased gradually. If the associated spinner wheel wasn’t spun for a period of time, the light faded. Each optical-sensing car parking spot contained an addressable LED strip that showed power flowing to or from the car based on the grid status.
The electronics used for this installation were mostly off-the-shelf, with a few custom-made parts. The main controller was a Raspberry Pi 3, and an Arduino Mega 2560 was used as an auxiliary controller. The Arduino controlled the stepper motors for the clocks and sent data to the four addressable LED strips concealed inside the top of the pylon. This setup simplified connectivity, as data and power for the Arduino were transmitted through a long USB cable. A separate power cable supplied power for the stepper motor driver board and LED strips. Each clock was equipped with a discreetly hidden homing sensor, using the same optical reflector setup as before. Upon powering up, the Arduino would first home each indicator dial before being ready to receive timing and indicator commands from the main controller.
A second auxiliary controller, another Arduino 2560, was used for user input. Each power station spinner had a Hall effect switch, and each car parking spot had an optical sensor. These inputs were connected to the user input controller, which sent events to the main controller via USB.
I used a cheap multichannel PWM driver board, connected to the RPi3 via I2C, to supply gate signals to a handful of NFETs to drive the main lighting. Data and clock signals for the car LED arrays were driven directly from the RPi GPIO pins via a few 5V level shifter chips mounted upon some veroboard. I could do this because the LED strips I chose used APA102 LEDs, not the more common WS2811. Because the former has a discrete clock input, they are far more tolerant of timing issues. This meant driving them directly from the GPIO pins was reliable.
I secured an off-the-shelf PC power supply to a custom bracket and mounted it underneath the system. I used a Snappit ATX power board to connect it to the rest of the system. This setup made it easy to power the installations and was mounted directly onto a Snappit protoboard. I used Snappit mounting plates for the RPi3 and Arduino and quickly laser-cut a new mounting plate for the PWM board. I screwed the Veroboard circuits for the level shifter and power transistors directly to the protoboard. This created a single assembly that could be mounted underneath the woodwork for the V2G chassis. Due to logistical issues, I couldn’t turn the chassis over to work on it face-down. So, I had to lay on my back and work on it above my face with a head torch. This was challenging and took many days, but with some help from others at critical moments, I could finish the construction. It’s an experience I hope not to repeat!