Goodnight Pi: Reducing Raspberry Pi Power Consumption
At a high level the design consists of four major hardware components. These are the Raspberry Pi itself, the ATMEGA328P, the RTC, and the Power Control Circuit. The Raspberry Pi wakes up when the ATMEGA328P provides it with power, and performs some action or series of actions based on why it was woken up. It then shuts itself down and alerts the ATMEGA that it can turn off power to the Pi. The Raspberry Pi and ATMEGA communicate with a custom protocol using SPI. The ATMEGA tracks the state of the Raspberry Pi and determines if it should be powered on or turned off. It determines this based on the time provided by the RTC and the state of its interrupt pins. The ATMEGA and the RTC communicate using I2C. The time interval and interrupts that should be able to trigger the ATMEGA to wake up the Raspberry Pi can be configured using a GUI on the Raspberry Pi. The ATMEGA then sets the state of the Power Control Circuit to control the power to the Raspberry Pi. Once configured and running, the system continuously cycles through a sequence of turning the Raspberry Pi on, letting it complete whatever application is desired, and shutting down the Pi again until the time interval expires or an interrrupt arives. A high level view of the system components is caputred in the block diagram below.
The first step in designing the project was choosing a microcontroller to control the power. The ATMEGA328P was used to control the power of the Raspberry Pi. This microcontroller was chosen because it can operate with far less power than a Raspberry Pi, is relatively inexpensive, and has extensive support and documentation because of its uses in the Arduino. Even though a stand-alone ATMEGA328P was used, all existing Arduino libraries could be used, making development much simpler. The ATMEGA328P also offers a diverse range of peripherals. It has external interrupts and is capable of communicating using both SPI and I2C, all of which were all necessary functions for this project. Below is a picture of the ATMEGA integrated into the system.
Next, the supporting hardware for the micrcontroller had to be designed. The ATMEGA328P was run at 3.3V and 8 MHz. In an Arduino it is typically run at 5V and 16 MHz, but this requires significantly more power. The lower frequency and voltage sacrificed performance to conserve energy, but the performance of the ATMEGA was still more than sufficient for this application. The ATMEGA only drew 3 mA while running, a significant improvement over the 210 mA that the Rasbperry Pi drew when idle. A schematic for the setup of the ATMEGA can be seen below.
Next, we had to determine how the Pi and the ATMEGA would communicate. Initially we planned to use I2C, but it proved impossible to have the Raspberry Pi act as a slave on an I2C bus and setting up a multi-master bus proved challenging. Instead, SPI was chosen for the link between the Raspberry Pi and the ATMEGA, so that I2C could be used exclusively for the RTC. The ATMEGA communicated with the Raspberry Pi using SPI, serving as the slave. The ATMEGA also connected to a GPIO pin on the Raspberry Pi that was configured to always output a high signal. When this pin went low, it indicated that the Raspberry Pi had shutdown, and when it went high again it indicated that the Raspberry Pi had turned back on successfully. This prevented any premature cycling of power, allowing the ATMEGA to track the exact state of the Raspberry Pi. The diagram below illustrates the connections between the Raspberry Pi and the ATMEGA.