Skip to main content

A Python Ethernet/IP library for communicating with Allen-Bradley PLCs.

Project description

pycomm3 is a Python 3 fork of pycomm, which is a native python library for communicating with PLCs using Ethernet/IP. The initial Python 3 translation was done in this fork, this library seeks to continue and expand upon the great work done by the original pycomm developers. This library supports Ethernet/IP communications with Allen-Bradley Control & Compact Logix PLCs. pycomm has support for SLC and MicroLogix PLCs, but they have not been ported yet. The module still exists in the package, but is broken and will raise a NotImplementedError on import. pylogix is another library with similar features (including Python 2 support), thank you to them for their hard work as well. Referencing pylogix code was a big help in implementing some features missing from pycomm. This library is only supported on Python 3.6 and up.


PLCs can be used to control heavy or dangerous equipment, this library is provided ‘As Is’ and makes no guarantees on it’s reliability in a production environment. This library makes no promises in the completeness or correctness of the protocol implementations and should not be solely relied upon for critical systems. The development for this library is aimed at providing quick and convenient access for reading/writing data inside Allen-Bradley Control/Compact Logix PLCs.


The Logix5000 Controller Data Access Manual, available from the Rockwell Developer How-to Guides, was used to implement the Ethernet/IP features in this library. Features like reading tags/arrays, writing tags/arrays, getting the tag list are all implemented based on the Data Access Manual. The Rockwell KB Article CIP Messages References 748424 lists many useful KB Articles for using the MSG instruction to perform various Ethernet/IP services. The Rockwell Knowledge Base Article 23341 was used to implement feature for getting the program name of the controller. Article 28917 was used for collecting other controller information.


The package can be installed from


pip install pycomm3

Basic Usage

Connect to a PLC and get some basic information about it. The path argument is the only one required, and it has 3 forms:

  • IP Address Only ( - Use if PLC is in slot 0 or if connecting to CompactLogix
  • IP Address/Slot ( - Use if PLC is not in slot 0
  • CIP Routing Path ( - Use if needing to route thru a backplane
    • first 2 examples will be replaced with the full path automatically, they’re there for convenience.
    • enet/backplane (or bp) are for port selection, standard CIP routing but without having to remember which port is what value.
from pycomm3 import LogixDriver

with LogixDriver('') as plc:
    # OUTPUT:
    # Program Name: PLCA, Device: 1756-L83E/B, Revision: 28.13

    # OUTPUT:
    # {'vendor': 'Rockwell Automation/Allen-Bradley', 'product_type': 'Programmable Logic Controller',
    #  'product_code': 166, 'version_major': 28, 'version_minor': 13, 'revision': '28.13', 'serial': 'FFFFFFFF',
    #  'device_type': '1756-L83E/B', 'keyswitch': 'REMOTE RUN', 'name': 'PLCA'}

By default, when creating the LogixDriver object, it will open a connection to the plc, read the program name, get the controller info, and get all the controller scoped tags. Using the init_tags kwarg will enable/disable automatically getting the controller tag list, and init_info will enable/disable program name and controller info loading. By reading the tag list first, this allows us to cache all the tag type/structure information, including the instance ids for all the tags. This information allows the read/write methods to require only the tag name. If your project will require program-scoped tags, be sure to set the init_program_tags kwarg. By default, only the controller-scoped tags will be read and cached. Calling plc.get_tag_list(program='*') will also have the same effect.

Symbol Instance Addressing is only available on v21+, if the PLC is on a firmware lower than that, getting the controller info will automatically disable that feature. If you disable init_info and are using a controller on a version lower than 21, set the plc.use_instance_ids attribute to false or your reads/writes will fail.

Default behavior is to use the Extended Forward Open service when opening a connection. This allows the use of 4KB of data for each request, standard is only 500B. Although this requires the communications module to be an EN2T or newer and the PLC firmware to be version 20 or newer. To use standard the Forward Open service set the large_packets kwarg to False.

Reading/Writing Tags

Reading or writing tags is as simple as calling the read and write methods. Both methods accept any number of tags, and will automatically pack multiple tags into a Multiple Service Packet Service (0x0A) while making sure to stay below the connection size. If there is a tag value that cannot fit within the request/reply packet, it will automatically handle that tag independently using the Read Tag Fragmented (0x52) or Write Tag Fragmented (0x53) requests. Other similar libraries do not do this automatically, this library attempts to be as seamless as possible.

Both methods will return Tag objects to reflect the success or failure of the operation.

class Tag(NamedTuple):
    tag: str
    value: Any
    type: Optional[str] = None
    error: Optional[str] = None

Tag objects are considered successful if the value is not None and the error is None. Otherwise, the error will indicate either the CIP error or exception that was thrown. Tag.__bool__() has been implemented in this way. type will indicate the data type of the tag and include [<length>] if multiple array elements are requested. value will contain the value of the tag either read or written, structures (read only) will be in the form of a { attribute: value, ... }. Even though strings are technically structures, both reading and writing support automatically converting them to/from normal string objects. Any structures that have only the attributes LEN (DINT) and DATA (array of SINT) will automatically be treated as strings. Reading of structures as a whole is supported as long as no attributes have External Access set to None (CIP limitation). Writing structures as a whole is not supported (for the time being) except for string objects.


with LogixDriver('') as plc:'tag1', 'tag2', 'tag3')  # read multiple tags'array{10}') # read 10 elements starting at 0 from an array'array[5]{20}) # read 20 elements starting at elements 5 from an array'string_tag')  # read a string tag and get a string

    # writes require a sequence of tuples of [(tag name, value), ... ]
    plc.write(('tag1, 0), ('tag2', 1), ('tag3', 2))  # write multiple tags
    plc.write(('array{5}', [1, 2, 3, 4, 5]))  # write 5 elements to an array starting at the 0 element
    plc.write(('array[10]{5}', [1, 2, 3, 4, 5]))  # write 5 elements to an array starting at element 10
    plc.write(('string_tag', 'Hello World!'))  # write to a string tag with a string
    plc.write(('string_array[2]{5}', 'Write an array of strings'.split()))  # write an array of 5 strings starting at element 2


Tag names for both read and write are case-sensitive and are required to be the same as they are named in the controller. This may change in the future. (pull requests welcome)

Tag Definitions

Tag definitions are uploaded from the controller automatically when connecting. This allows the read/writing methods to work. These definitions contain information like instance ids and structure size and composition. This information allows for many optimizations and features that other similar libraries do not offer. The old pycomm API does not depend on these, but the new read/write methods do. The tag definitions are accessible from the tags attribute. The tags property is a dict of {tag name: definition}.

Tag Information Collected:

    'tag1': {
        'tag_name': 'tag1',  # same as key
        'dim': 0,  # number of dimensions of array (0-3)
        'instance_id':  # used for reads/writes on v21+ controllers
        'alias': True/False,  # if the tag is an alias to another (this is not documented, but an educated guess found thru trial and error
        'external_access': 'Read/Write',  # string value of external access setting
        'dimensions': [0, 0, 0]  # array dimensions
        'tag_type': 'atomic',
        'data_type' : 'DINT'  # string value of an atomic type
   'tag2' : {
        'tag_type': 'struct',
        'data_type': {
            'name': 'TYPE', # name of structure, udt, or aoi
            'internal_tags': {
                'attribute': {  # is an atomic type
                    'offset': 0 # byte offset for members within the struct, used mostly for reading an entire structure
                    'tag_type': 'atomic',
                    'data_type:  'Type', # name of data type
                    'bit': 0   # optional, exists if element is mapped to a bit of a dint or element of a bool array
                    'array': 0,  # optional, length of error if the attribute is an array
                'attribute2': {  # is a struct
                    'tag_type': 'struct',
                    'data_type': {
                        'name': 'TYPE',  # name of data type,
                        'internal_tags' : {  # definition of all tags internal/hidden and public attributes
                            ... # offset/array/bit/tag_type/data_type
                        'attributes' : [...], # list of public attributes (shown in Logix)
                        'template' : {...}, # used internally




If running multiple clients, you can initialize all the tag definitions in one client and pass them to other clients by turning off the init_* args and setting plc2._tags = plc1.tags.

Unit Testing

pytest is used for unit testing. The tests directory contains an L5X export of the Pycomm3_Testing program that contains all tags necessary for testing. The only requirement for testing (besides a running PLC with the testing program) is the environment variable PLCPATH for the PLC defined.


Test coverage is not complete, pull requests are very much welcome to cover all combinations for reading and writing tags.


pycomm3 is distributed under the MIT License

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for pycomm3, version 0.5.2
Filename, size File type Python version Upload date Hashes
Filename, size pycomm3-0.5.2-py3-none-any.whl (72.6 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size pycomm3-0.5.2.tar.gz (68.2 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page