You will need AVR GNU Toolchain to compile. Let’s consider an example of native testing of the firmware functionality. Nevertheless, our firmware has functionality which it is necessary to test (converting ADC value into temperature and then displaying it on the LED screen). The function of communication with the internal and external periphery can be performed via operating the microcontroller registers. Temperature measurement is performed by reading the voltage on the voltage divider (which includes a thermistor), converting the ADC value into temperature values, and displaying it on the 1602 ( hd44780) LCD screen.
This approach allows me to automate test execution and firmware upload to the target system.įor example, let’s consider the firmware for the Atmega1284 microcontroller that implements the functionality of a simple thermometer. The project build process (compile, link, firmware) is done with the make utility. I rely on VSCode on Ubuntu for source code, AVR GNU Toolchain for compile, and link firmware, gcc for compiling tests and simulator. Unfortunately, this environment is not cross-platform and is only available for Windows.įor a clear comprehension of examples in this project, I use open-source tools. In this article we will analyze the first two methods as the most promising for automation.įor AVR MCU the most convenient and productive development environment is Atmel Studio.
It is very difficult to automate testing using this method. The test cycles will take a long time due as you will have to constantly reflash the MCU. You can program your own environment for the MCU.ĭisadvantages: limited accuracy of MCU and environmental simulations, difficulty in creating and configuring such an emulator.Īdvantages: it is possible to work with the periphery of the MC, the firmware will work out the same way as in production.ĭisadvantages: you need to have a ready-made device with all peripherals and electronic components. You can do this by using a small logger program on your unprocessed sensor data.Īdvantages: no device needed for testing. It’s much better to prepare a dataset in advance. The testing of such an algorithm with a real data source would be very inconvenient. A good example of using this method is the testing of a platform-independent computational algorithm that requires a dataset from sensors. It will not work with external peripherals. Let’s list them all:Īdvantages: tests start and work quickly, no need to access the device.
There are three main options for running tests on embedded systems.
0 kommentar(er)
