This is the documentation for the latest development branch of MicroPython and may refer to features that are not available in released versions.

If you are looking for the documentation for a specific release, use the drop-down menu on the left and select the desired version.

esp32 — functionality specific to the ESP32

The esp32 module contains functions and classes specifically aimed at controlling ESP32 modules.



Configure whether or not a touch will wake the device from sleep. wake should be a boolean value.


Configure whether or not the Ultra-Low-Power co-processor can wake the device from sleep. wake should be a boolean value.

esp32.wake_on_ext0(pin, level)

Configure how EXT0 wakes the device from sleep. pin can be None or a valid Pin object. level should be esp32.WAKEUP_ALL_LOW or esp32.WAKEUP_ANY_HIGH.

esp32.wake_on_ext1(pins, level)

Configure how EXT1 wakes the device from sleep. pins can be None or a tuple/list of valid Pin objects. level should be esp32.WAKEUP_ALL_LOW or esp32.WAKEUP_ANY_HIGH.


Configure whether non-RTC GPIO pin configuration is retained during deep-sleep mode for held pads. enable should be a boolean value.


Read the raw value of the internal temperature sensor, returning an integer.


Returns information about the ESP-IDF heap memory regions. One of them contains the MicroPython heap and the others are used by ESP-IDF, e.g., for network buffers and other data. This data is useful to get a sense of how much memory is available to ESP-IDF and the networking stack in particular. It may shed some light on situations where ESP-IDF operations fail due to allocation failures.

The capabilities parameter corresponds to ESP-IDF’s MALLOC_CAP_XXX values but the two most useful ones are predefined as esp32.HEAP_DATA for data heap regions and esp32.HEAP_EXEC for executable regions as used by the native code emitter.

The return value is a list of 4-tuples, where each 4-tuple corresponds to one heap and contains: the total bytes, the free bytes, the largest free block, and the minimum free seen over time.

Example after booting:

>>> import esp32; esp32.idf_heap_info(esp32.HEAP_DATA)
[(240, 0, 0, 0), (7288, 0, 0, 0), (16648, 4, 4, 4), (79912, 35712, 35512, 35108),
 (15072, 15036, 15036, 15036), (113840, 0, 0, 0)]


Free IDF heap memory in the esp32.HEAP_DATA region is available to be automatically added to the MicroPython heap to prevent a MicroPython allocation from failing. However, the information returned here is otherwise not useful to troubleshoot Python allocation failures. micropython.mem_info() and gc.mem_free() should be used instead:

The “max new split” value in micropython.mem_info() output corresponds to the largest free block of ESP-IDF heap that could be automatically added on demand to the MicroPython heap.

The result of gc.mem_free() is the total of the current “free” and “max new split” values printed by micropython.mem_info().

Flash partitions

This class gives access to the partitions in the device’s flash memory and includes methods to enable over-the-air (OTA) updates.

class esp32.Partition(id, block_size=4096, /)

Create an object representing a partition. id can be a string which is the label of the partition to retrieve, or one of the constants: BOOT or RUNNING. block_size specifies the byte size of an individual block.

classmethod Partition.find(type=TYPE_APP, subtype=0xff, label=None, block_size=4096)

Find a partition specified by type, subtype and label. Returns a (possibly empty) list of Partition objects. Note: subtype=0xff matches any subtype and label=None matches any label.

block_size specifies the byte size of an individual block used by the returned objects.

Returns a 6-tuple (type, subtype, addr, size, label, encrypted).

Partition.readblocks(block_num, buf)
Partition.readblocks(block_num, buf, offset)
Partition.writeblocks(block_num, buf)
Partition.writeblocks(block_num, buf, offset)
Partition.ioctl(cmd, arg)

These methods implement the simple and extended block protocol defined by vfs.AbstractBlockDev.


Sets the partition as the boot partition.


Do not enter deepsleep after changing the OTA boot partition, without first performing a hard reset or power cycle. This ensures the bootloader will validate the new image before booting.


Gets the next update partition after this one, and returns a new Partition object. Typical usage is Partition(Partition.RUNNING).get_next_update() which returns the next partition to update given the current running one.

classmethod Partition.mark_app_valid_cancel_rollback()

Signals that the current boot is considered successful. Calling mark_app_valid_cancel_rollback is required on the first boot of a new partition to avoid an automatic rollback at the next boot. This uses the ESP-IDF “app rollback” feature with “CONFIG_BOOTLOADER_APP_ROLLBACK_ENABLE” and an OSError(-261) is raised if called on firmware that doesn’t have the feature enabled. It is OK to call mark_app_valid_cancel_rollback on every boot and it is not necessary when booting firmware that was loaded using esptool.



Used in the Partition constructor to fetch various partitions: BOOT is the partition that will be booted at the next reset and RUNNING is the currently running partition.


Used in Partition.find to specify the partition type: APP is for bootable firmware partitions (typically labelled factory, ota_0, ota_1), and DATA is for other partitions, e.g. nvs, otadata, phy_init, vfs.


Used in idf_heap_info.


The RMT (Remote Control) module, specific to the ESP32, was originally designed to send and receive infrared remote control signals. However, due to a flexible design and very accurate (as low as 12.5ns) pulse generation, it can also be used to transmit or receive many other types of digital signals:

import esp32
from machine import Pin

r = esp32.RMT(0, pin=Pin(18), clock_div=8)
r  # RMT(channel=0, pin=18, source_freq=80000000, clock_div=8, idle_level=0)

# To apply a carrier frequency to the high output
r = esp32.RMT(0, pin=Pin(18), clock_div=8, tx_carrier=(38000, 50, 1))

# The channel resolution is 100ns (1/(source_freq/clock_div)).
r.write_pulses((1, 20, 2, 40), 0)  # Send 0 for 100ns, 1 for 2000ns, 0 for 200ns, 1 for 4000ns

The input to the RMT module is an 80MHz clock (in the future it may be able to configure the input clock but, for now, it’s fixed). clock_div divides the clock input which determines the resolution of the RMT channel. The numbers specified in write_pulses are multiplied by the resolution to define the pulses.

clock_div is an 8-bit divider (0-255) and each pulse can be defined by multiplying the resolution by a 15-bit (1-PULSE_MAX) number. There are eight channels (0-7) and each can have a different clock divider.

So, in the example above, the 80MHz clock is divided by 8. Thus the resolution is (1/(80Mhz/8)) 100ns. Since the start level is 0 and toggles with each number, the bitstream is 0101 with durations of [100ns, 2000ns, 100ns, 4000ns].

For more details see Espressif’s ESP-IDF RMT documentation..


The current MicroPython RMT implementation lacks some features, most notably receiving pulses. RMT should be considered a beta feature and the interface may change in the future.

class esp32.RMT(channel, *, pin=None, clock_div=8, idle_level=False, tx_carrier=None)

This class provides access to one of the eight RMT channels. channel is required and identifies which RMT channel (0-7) will be configured. pin, also required, configures which Pin is bound to the RMT channel. clock_div is an 8-bit clock divider that divides the source clock (80MHz) to the RMT channel allowing the resolution to be specified. idle_level specifies what level the output will be when no transmission is in progress and can be any value that converts to a boolean, with True representing high voltage and False representing low.

To enable the transmission carrier feature, tx_carrier should be a tuple of three positive integers: carrier frequency, duty percent (0 to 100) and the output level to apply the carrier to (a boolean as per idle_level).

classmethod RMT.source_freq()

Returns the source clock frequency. Currently the source clock is not configurable so this will always return 80MHz.


Return the clock divider. Note that the channel resolution is 1 / (source_freq / clock_div).

RMT.wait_done(*, timeout=0)

Returns True if the channel is idle or False if a sequence of pulses started with RMT.write_pulses is being transmitted. If the timeout keyword argument is given then block for up to this many milliseconds for transmission to complete.


Configure looping on the channel. enable_loop is bool, set to True to enable looping on the next call to RMT.write_pulses. If called with False while a looping sequence is currently being transmitted then the current loop iteration will be completed and then transmission will stop.

RMT.write_pulses(duration, data=True)

Begin transmitting a sequence. There are three ways to specify this:

Mode 1: duration is a list or tuple of durations. The optional data argument specifies the initial output level. The output level will toggle after each duration.

Mode 2: duration is a positive integer and data is a list or tuple of output levels. duration specifies a fixed duration for each.

Mode 3: duration and data are lists or tuples of equal length, specifying individual durations and the output level for each.

Durations are in integer units of the channel resolution (as described above), between 1 and PULSE_MAX units. Output levels are any value that can be converted to a boolean, with True representing high voltage and False representing low.

If transmission of an earlier sequence is in progress then this method will block until that transmission is complete before beginning the new sequence.

If looping has been enabled with RMT.loop, the sequence will be repeated indefinitely. Further calls to this method will block until the end of the current loop iteration before immediately beginning to loop the new sequence of pulses. Looping sequences longer than 126 pulses is not supported by the hardware.

static RMT.bitstream_channel([value])

Select which RMT channel is used by the machine.bitstream implementation. value can be None or a valid RMT channel number. The default RMT channel is the highest numbered one.

Passing in None disables the use of RMT and instead selects a bit-banging implementation for machine.bitstream.

Passing in no argument will not change the channel. This function returns the current channel number.



Maximum integer that can be set for a pulse duration.

Ultra-Low-Power co-processor

This class gives access to the Ultra Low Power (ULP) co-processor on the ESP32, ESP32-S2 and ESP32-S3 chips.


This class does not provide access to the RISCV ULP co-processor available on the ESP32-S2 and ESP32-S3 chips.

class esp32.ULP

This class provides access to the Ultra-Low-Power co-processor.

ULP.set_wakeup_period(period_index, period_us)

Set the wake-up period.

ULP.load_binary(load_addr, program_binary)

Load a program_binary into the ULP at the given load_addr.

Start the ULP running at the given entry_point.



Selects the wake level for pins.

Non-Volatile Storage

This class gives access to the Non-Volatile storage managed by ESP-IDF. The NVS is partitioned into namespaces and each namespace contains typed key-value pairs. The keys are strings and the values may be various integer types, strings, and binary blobs. The driver currently only supports 32-bit signed integers and blobs.


Changes to NVS need to be committed to flash by calling the commit method. Failure to call commit results in changes being lost at the next reset.

class esp32.NVS(namespace)

Create an object providing access to a namespace (which is automatically created if not present).

NVS.set_i32(key, value)

Sets a 32-bit signed integer value for the specified key. Remember to call commit!


Returns the signed integer value for the specified key. Raises an OSError if the key does not exist or has a different type.

NVS.set_blob(key, value)

Sets a binary blob value for the specified key. The value passed in must support the buffer protocol, e.g. bytes, bytearray, str. (Note that esp-idf distinguishes blobs and strings, this method always writes a blob even if a string is passed in as value.) Remember to call commit!

NVS.get_blob(key, buffer)

Reads the value of the blob for the specified key into the buffer, which must be a bytearray. Returns the actual length read. Raises an OSError if the key does not exist, has a different type, or if the buffer is too small.


Erases a key-value pair.


Commits changes made by set_xxx methods to flash.