Modbus Controller Number

The modbus_controller platform creates a Number from a modbus_controller. When the Number is updated a modbus write command is created sent to the device.

Configuration variables

register_type (Optional): the type of register to write. One of holding (default), coil, or custom.
address (Required, int): start address of the first register in a range (can be decimal or hexadecimal).
value_type (Optional): datatype of the modbus register data. Defaults to U_WORD, a 16 bit integer in big endian format (MSB first):
- U_WORD (unsigned 16 bit integer from 1 register = 16bit)
- S_WORD (signed 16 bit integer from 1 register = 16bit)
- U_WORD_S (unsigned 16 bit integer from 1 register with bytes swapped within the register = 16bit)
- S_WORD_S (signed 16 bit integer from 1 register with bytes swapped within the register = 16bit)
- U_DWORD (unsigned 32 bit integer from 2 registers = 32bit)
- S_DWORD (signed 32 bit integer from 2 registers = 32bit)
- U_DWORD_R (unsigned 32 bit integer from 2 registers low word first)
- S_DWORD_R (signed 32 bit integer from 2 registers low word first)
- U_QWORD (unsigned 64 bit integer from 4 registers = 64bit)
- S_QWORD (signed 64 bit integer from 4 registers = 64bit)
- U_QWORD_R (unsigned 64 bit integer from 4 registers low word first)
- S_QWORD_R (signed 64 bit integer from 4 registers low word first)
- FP32 (32 bit IEEE 754 floating point from 2 registers)
- FP32_R (32 bit IEEE 754 floating point - same as FP32 but low word first)

WARNING

U_WORD_S and S_WORD_S are a rare, non-standard configuration: the two bytes within one 16-bit register are reversed (LSB first on the wire within that register). Modbus registers are big-endian (MSB first) per the specification. The _S suffix is not the same as _R (word order reversed / low word first across multiple registers).

min_value (Optional, float): The minimum value this number can be.
max_value (Optional, float): The maximum value this number can be.
step (Optional, float): The granularity with which the number can be set. Defaults to 1.
multiply (Optional, float): multiply the new value with this factor before sending the requests. Ignored if lambda is defined.
use_write_multiple (Optional, boolean): By default the modbus command Function Code 6 (Preset Single Registers) is used for setting the holding register if only one register is set. If your device only supports Function Code 16 (Preset Multiple Registers) set this option to true.
skip_updates (Optional, int): Deprecated and no longer has any effect — every register range is polled on each update_interval. The key is still accepted (logging a config-time warning) and will be removed in 2027.3.0. To poll some registers less often, place those sensors on a second modbus_controller with the same address: and a slower update_interval:.
response_size (Optional): Size of the response for the register in bytes. If unset, defaults to the size implied by value_type (2 bytes per register).
reuse_previous_range (Optional, boolean or auto): How this item relates to the register range built just before it (same register type, ascending address order). auto (default) joins when the addresses are adjacent and the item’s position in the reply is exact; true joins unconditionally, reading across address gaps and past registers with a non-standard response_size; false always starts a new range here (later items may still extend it). See Register ranges. Replaces the deprecated force_new_range and register_count options, which will be removed in 2027.3.0 — the linked section covers migration.
offset (Optional, int): Offset from start address in bytes (only required for uncommon response encodings). If more than one register is written in a command this value is used to find the start of this datapoint relative to start address. The component calculates the size of the range based on offset and size of the value type.
custom_pdu (Optional, list of bytes): The modbus PDU (function code + data) for a custom command. This allows using non-standard commands. The device address (taken from the controller’s address:) and the CRC are added automatically, so do not include a leading device-address byte. If custom_pdu is used, address and register_type can’t be used. Renamed from custom_command, which is now removed and raises a config-validation error. See Using custom_pdu for how to use custom_pdu
lambda (Optional, lambda): Lambda to be evaluated every update interval to get the new value of the number.

Parameters passed into the lambda
- x (float): The parsed float value of the modbus data
- data (std::span<const uint8_t>): span containing the complete raw modbus response bytes for this number note: because the response contains data for all registers in the same range you have to use data[item->offset] to get the first response byte for your number.
- item (ModbusNumber *): The number entity, a SensorItem subclass, so the configured address, offset, bitmask, etc. are accessible.
Possible return values for the lambda:
- return <FLOATING_POINT_NUMBER>; the new value for the number.
- return NAN; if the state should be considered invalid to indicate an error (advanced).
write_lambda (Optional, lambda): Lambda called before send. Lambda is evaluated before the modbus write command is created.

Parameters passed into the lambda
- x (float): The new float value being set
- item (ModbusNumber *): The entity as its own modbus device: the write helpers and queue_pdu() are available - for example item->write_single_register(0x1234, 42) or item->queue_pdu(...) - regardless of the number’s own register type. Also a SensorItem subclass, so the configured address, offset, bitmask, etc. are accessible.
- payload (RegisterValues &payload): Deprecated, removed in 2027.3.0 - use item instead. An empty buffer; if the lambda fills it (push_back/clear/assign - not the full std::vector API), it is sent as a raw modbus frame packed from the 16 bit words, and must include all required bytes, including the device address; the CRC is added automatically, and the return value is ignored. This is the number platform’s historical buffer meaning; it differs from select and output, whose buffers take plain register words.
Possible return values for the lambda:
- return <FLOATING_POINT_NUMBER>; the value to be written to the device. The number still reports the requested value x; use this to translate between the ESPHome value and the modbus transmitted value.
- return {}; if the lambda handled (or deliberately skipped) the write itself, for example via item->write_*().
All other options from Number.

Example

number:
  - platform: modbus_controller
    modbus_controller_id: modbus1
    id: battery_capacity_number
    name: "Battery Cap Number"
    address: 0x9001
    value_type: U_WORD
    multiply: 1.0

  - platform: modbus_controller
    modbus_controller_id: modbus1
    id: battery_capacity_number_tenths
    name: "Battery Cap Number (device stores tenths)"
    address: 0x9002
    value_type: U_WORD
    lambda: "return x / 10.0;"
    write_lambda: |-
      ESP_LOGD("main","Modbus Number incoming value = %f",x);
      // the device stores tenths: write the scaled value through the entity;
      // the number still reports x
      item->write_single_register(0x9002, (uint16_t) (x * 10));
      // the lambda must return something - return empty, the write is already done
      return {};

Modbus Controller Number

Configuration variables

Example

See Also