micropython/ports/esp32/README.md

11 KiB

MicroPython port to the ESP32

This is an experimental port of MicroPython to the Espressif ESP32 microcontroller. It uses the ESP-IDF framework and MicroPython runs as a task under FreeRTOS.

Supported features include:

  • REPL (Python prompt) over UART0.
  • 16k stack for the MicroPython task and 96k Python heap.
  • Many of MicroPython's features are enabled: unicode, arbitrary-precision integers, single-precision floats, complex numbers, frozen bytecode, as well as many of the internal modules.
  • Internal filesystem using the flash (currently 2M in size).
  • The machine module with GPIO, UART, SPI, software I2C, ADC, DAC, PWM, TouchPad, WDT and Timer.
  • The network module with WLAN (WiFi) support.

Development of this ESP32 port was sponsored in part by Microbric Pty Ltd.

Setting up the toolchain and ESP-IDF

There are two main components that are needed to build the firmware:

  • the Xtensa cross-compiler that targets the CPU in the ESP32 (this is different to the compiler used by the ESP8266)
  • the Espressif IDF (IoT development framework, aka SDK)

If you are on a Windows machine then the Windows Subsystem for Linux is the most efficient way to install the ESP32 toolchain and build the project. If you use WSL then follow the Linux instructions rather than the Windows instructions.

The ESP-IDF changes quickly and MicroPython only supports certain versions. The git hash of these versions (one for 3.x, one for 4.x) can be found by running make without a configured ESPIDF. Then you can fetch the required IDF using the following command:

$ cd ports/esp32
$ make ESPIDF=  # This will print the supported hashes, copy the one you want.
$ export ESPIDF=$HOME/src/github.com/espressif/esp-idf  # Or any path you like.
$ mkdir -p $ESPIDF
$ cd $ESPIDF
$ git clone https://github.com/espressif/esp-idf.git $ESPIDF
$ git checkout <Current supported ESP-IDF commit hash>
$ git submodule update --init --recursive

Note: The ESP IDF v4.x support is currently experimental. It does not currently support PPP or wired Ethernet.

Python dependencies

You will also need other dependencies from the IDF, see $ESPIDF/requirements.txt, but at a minimum you need pyserial>=3.0 and pyparsing>=2.0.3,<2.4.0.

You can use Python 2 or Python 3. If you need to override the system default add (for example) PYTHON=python3 to any of the make commands below.

It is recommended to use a Python virtual environment. Even if your system package manager already provides these libraries, the IDF v4.x is currently incompatible with pyparsing 2.4 and higher.

For example, to set up a Python virtual environment from scratch:

$ cd ports/esp32
$ python3 -m venv build-venv
$ source build-venv/bin/activate
$ pip install --upgrade pip
$ pip install -r path/to/esp-idf/requirements.txt

To re-enter this virtual environment in future sessions, you only need to source the activate script, i.e.:

$ cd ports/esp32
$ source build-venv/bin/activate

Then, to install the toolchain (which includes the GCC compiler, linker, binutils, etc), there are two options:

  1. Using the IDF scripts to install the toolchain (IDF 4.x only) ================================================================

Follow the steps at the Espressif Getting Started guide.

This will guide you through using the install.sh (or install.bat) script to download the toolchain and add it to your PATH. The steps are summarised below:

After you've cloned and checked out the IDF to the correct version (see above), run the install.sh script:

$ cd $ESPIDF
$ ./install.sh    # (or install.bat on Windows)

Then in the ports/esp32 directory, source the export.sh script to set the PATH.

$ cd micropython/ports/esp32
$ source $ESPIDF/export.sh   # (or path\to\esp-idf\export.bat on Windows)
$ # Run make etc, see below.

The install.sh step only needs to be done once. You will need to source export.sh for every new session.

Note: If you get an error about --no-site-packages, then modify $ESPIDF/tools/idf_tools.py and make the same change as this commit.

  1. or, Downloading pre-built toolchain manually (IDF 3.x and 4.x) =============================================================

Note: while this works with 4.x, if you're using the 4.x IDF, it's much simpler to use the guide above, which will also get a more recent version of the toolchain.

You can follow the 3.x guide at:

You will need to update your PATH environment variable to include the ESP32 toolchain. For example, you can issue the following commands on (at least) Linux:

$ export PATH=$PATH:$HOME/esp/crosstool-NG/builds/xtensa-esp32-elf/bin

You can put this command in your .profile or .bash_login, or do it manually.

Configuring the MicroPython build

You then need to set the ESPIDF environment/makefile variable to point to the root of the ESP-IDF repository. The recommended way to do this is to have a custom makefile in ports/esp32 which sets any additional variables, then includes the main Makefile. Note that GNU Make will preferentially run GNUmakefile, then makefile, then Makefile, which is what allows this to work. On case-insensitive filesystems, you'll need to use GNUmakefile rather than makefile.

Create a new file in the esp32 directory called makefile (or GNUmakefile) and add the following lines to that file:

ESPIDF ?= <path to root of esp-idf repository>
BOARD ?= GENERIC
#PORT ?= /dev/ttyUSB0
#FLASH_MODE ?= qio
#FLASH_SIZE ?= 4MB
#CROSS_COMPILE ?= xtensa-esp32-elf-

include Makefile

Be sure to enter the correct path to your local copy of the IDF repository (and use $(HOME), not tilde (~), to reference your home directory).

If the Xtensa cross-compiler is not in your path you can use the CROSS_COMPILE variable to set its location. Other options of interest are PORT for the serial port of your ESP32 module, and FLASH_MODE (which may need to be dio for some modules) and FLASH_SIZE. See the Makefile for further information.

The default ESP IDF configuration settings are provided by the GENERIC board definition in the directory boards/GENERIC. For a custom configuration you can define your own board directory.

Any of these variables can also be set on the make command line, e.g. to set the BOARD variable, use:

$ make BOARD=TINYPICO

Note the use of ?= in the makefile which allows them to be overridden on the command line. There is also a GENERIC_SPIRAM board for for ESP32 modules that have external SPIRAM, but prefer to use a specific board target (or define your own as necessary).

Building the firmware

The MicroPython cross-compiler must be built to pre-compile some of the built-in scripts to bytecode. This can be done by (from the root of this repository):

$ cd mpy-cross
$ make mpy-cross

Then to build MicroPython for the ESP32 run:

$ cd ports/esp32
$ make submodules
$ make

This will produce binary firmware images in the build/ subdirectory (three of them: bootloader.bin, partitions.bin and application.bin).

To flash the firmware you must have your ESP32 module in the bootloader mode and connected to a serial port on your PC. Refer to the documentation for your particular ESP32 module for how to do this. The serial port and flash settings are set in the Makefile, and can be overridden in your local makefile; see above for more details.

You will also need to have user permissions to access the /dev/ttyUSB0 device. On Linux, you can enable this by adding your user to the dialout group, and rebooting or logging out and in again. (Note: on some distributions this may be the uucp group, run ls -la /dev/ttyUSB0 to check.)

$ sudo adduser <username> dialout

If you are installing MicroPython to your module for the first time, or after installing any other firmware, you should first erase the flash completely:

$ make erase

To flash the MicroPython firmware to your ESP32 use:

$ make deploy

This will use the esptool.py script (provided by ESP-IDF) to flash the binary images to the device.

Getting a Python prompt on the device

You can get a prompt via the serial port, via UART0, which is the same UART that is used for programming the firmware. The baudrate for the REPL is 115200 and you can use a command such as:

$ picocom -b 115200 /dev/ttyUSB0

or

$ miniterm.py /dev/ttyUSB0 115200

Configuring the WiFi and using the board

The ESP32 port is designed to be (almost) equivalent to the ESP8266 in terms of the modules and user-facing API. There are some small differences, notably that the ESP32 does not automatically connect to the last access point when booting up. But for the most part the documentation and tutorials for the ESP8266 should apply to the ESP32 (at least for the components that are implemented).

See http://docs.micropython.org/en/latest/esp8266/esp8266/quickref.html for a quick reference, and http://docs.micropython.org/en/latest/esp8266/esp8266/tutorial/intro.html for a tutorial.

The following function can be used to connect to a WiFi access point (you can either pass in your own SSID and password, or change the defaults so you can quickly call wlan_connect() and it just works):

def wlan_connect(ssid='MYSSID', password='MYPASS'):
    import network
    wlan = network.WLAN(network.STA_IF)
    if not wlan.active() or not wlan.isconnected():
        wlan.active(True)
        print('connecting to:', ssid)
        wlan.connect(ssid, password)
        while not wlan.isconnected():
            pass
    print('network config:', wlan.ifconfig())

Note that some boards require you to configure the WiFi antenna before using the WiFi. On Pycom boards like the LoPy and WiPy 2.0 you need to execute the following code to select the internal antenna (best to put this line in your boot.py file):

import machine
antenna = machine.Pin(16, machine.Pin.OUT, value=0)

Troubleshooting

  • Continuous reboots after programming: Ensure FLASH_MODE is correct for your board (e.g. ESP-WROOM-32 should be DIO). Then perform a make clean, rebuild, redeploy.