Skip to content

InspireHornets/p-net

 
 

Repository files navigation

p-net Profinet device stack

Web resources

Build Status CodeQL

Converting examples for your own usage

Convert p-net to use it with your own datatypes and GSDML file.

  1. Copy the GSML file that is found in the p-net/samples/pn_dev/GSDML-V2.4-RT-Labs-P-Net-Sample-App-20220324.xml Note: I tried to create my own profinet device with TwinCAT or with CODESYS and export a GSDML file. But I never got this to work and I don´t know why.
  2. Manually adjust the GSDML file as needed, for example:
    • the company name, name of the device etc.
    • size of IO, datatypes of IO and number of modules
    • also make sure all submodules have a unique ID (not sure if needed).
  3. Duplicate the samples/pn_simple_example folder and content and rename the pn_simple_example folder, for example to pn_my_app
  4. Then open the CMakeLists.txt in your pn_my_app folder and rename all pn_simple_example to pn_my_app
  5. Open app_gsdml.h and add the sizes and (sub)module IDs of the modules you added
  6. In app_gsdml.c:
    1. Add static const app_gsdml_module_t for each module. Check that all members (.id, .name, etc.) are correct
    2. Add static const app_gsdml_submodule_t for each submodule and check that all the members are correct
    3. Add the added modules and sub modules to the supported lists: app_gsdml_modules[] and app_gsdml_submodules[]
  7. Open app_data.c:
    1. add a static struct where in/output data can be stored into.
      • static uint8_t inputdata[APP_GSDML_INPUT_DATA_DIGITAL_SIZE] = {0};
      • Floats are saved as unsigned integers, because of endian conversion. Profinet data has network endianess (Big endian), whereas C uses little endian ?
  8. add a typedef struct to store I/O data

Endianess = The attribute of a system that indicates whether integers are represented with the most significant byte stored at the lowest address (big endian) or at the highest address (little endian)

p-net

Profinet device stack implementation. Key features:

  • Profinet v2.4
    • Conformance Class A and B
    • Real Time Class 1
    • Multiple Ethernet ports
  • Easy to use
    • Extensive documentation and instructions on how to get started.
    • Build and run sample application on Raspberry Pi in 30 minutes.
  • Portable
    • Written in C.
    • Linux, RTOS or bare metal.
    • Sources for supported port layers provided.

The RT-Labs Profinet stack p-net is used for Profinet device implementations. It is easy to use and provides a small footprint. It is especially well suited for embedded systems where resources are limited and efficiency is crucial. The stack is supplied with full sources including porting layers and a sample application.

Also C++ (any version) is supported for application development.

The main requirement on the platform is that it can send and receive raw Ethernet Layer 2 frames.

Features:

  • Multiple Ethernet ports (for Linux only, so far)
  • TCP/IP
  • LLDP
  • SNMP
  • RT (real-time)
  • Address resolution
  • Parameterization
  • Process IO data exchange
  • Alarm handling
  • Configurable number of modules and sub-modules
  • Bare-metal or OS
  • Porting layer provided
  • Supports I&M0 - I&M4. The I&M data is supported for the device, but not for individual modules.

Limitations or not yet implemented:

  • This is a device stack, which means that the IO-controller/master/PLC side is not supported.
  • No media redundancy (No MRP support)
  • Legacy startup mode is not fully implemented
  • No support for RT_CLASS_UDP
  • No support for DHCP
  • No fast start-up
  • No MC multicast device-to-device
  • No support of shared device (connection to multiple controllers)
  • Supports only full connections, not the limited "DeviceAccess" connection type.
  • No iPar (parameter server) support
  • No support for time synchronization
  • No UDP frames at alarm (just the default alarm mechanism is implemented)
  • No ProfiDrive or ProfiSafe profiles.

This software is dual-licensed, with GPL version 3 and a commercial license. If you intend to use this stack in a commercial product, you likely need to buy a license. See LICENSE.md for more details.

Requirements

The platform must be able to send and receive raw Ethernet Layer 2 frames, and the Ethernet driver must be able to handle full size frames. It should also avoid copying data, for performance reasons.

  • cmake 3.14 or later

For Linux:

  • gcc 4.6 or later
  • See the "Real-time properties of Linux" page in the documentation on how to improve Linux timing

For rt-kernel:

  • Workbench 2020.1 or later

An example of microcontroller we have been using is the Infineon XMC4800, which has an ARM Cortex-M4 running at 144 MHz, with 2 MB Flash and 352 kB RAM. It runs rt-kernel, and we have tested it with 9 Profinet slots each having 8 digital inputs and 8 digital outputs (one bit each). The values are sent and received each millisecond (PLC watchdog setting 3 ms).

Getting started

See the tutorial in the documentation: https://rt-labs.com/docs/p-net/tutorial.html

Note that you need to include submodules when cloning:

git clone --recurse-submodules https://github.com/rtlabs-com/p-net.git

Dependencies

Some of the platform-dependent parts are located in the OSAL repository and the cmake-tools repository.

Those are downloaded automatically during install.

The p-net stack contains no third party components. Its external dependencies are:

  • C-library
  • An operating system (if used)
  • For conformance class B you need an SNMP implementation. On Linux is net-snmp (BSD License) used http://www.net-snmp.org

Tools used for building, testing and documentation (not shipped in the resulting binaries):

Contributions

Contributions are welcome. If you want to contribute you will need to sign a Contributor License Agreement and send it to us either by e-mail or by physical mail. More information is available on https://rt-labs.com/contribution.

About

PROFINET device stack for embedded devices

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • C 84.7%
  • C++ 13.6%
  • CMake 1.3%
  • Shell 0.4%