Menu

openCONFIGURATOR v1.4.1: PDO mapping with REAL32 or REAL64

Help
2015-11-20
2016-01-27
  • Romain Naour

    Romain Naour - 2015-11-20

    Hello,

    I'm using openCONFIGURATOR 1.4.1 and I'm trying to use a REAL32 variable mapped as PDO.

    I get this error while building the mnobd.cdc:

    Code:33 Msg:Node 1, (Sub)Index 0x1600/0x06: Mapping of datatype 'Real32' not supported. Supported data types: Integer8(16/32/64), Unsigned8(16/32/64).
    

    (see CheckAllowedDTForMapping())

    As far I understand, there is no restriction to Interger in the openpowerlink specification and Real32 and Real64 are allowed to be used in PDO mapping.
    Also, a collegue reported to be able to use REAL32 with a B&N cpu (MN).

    Is CheckAllowedDTForMapping() is too strict with allowed types ?
    Is the upcoming openCONFIGURATOR 2 will be able to use REAL as PDO ?

    Best regards,
    Romain Naour

     
  • Christoph Ruecker

    Hello Romain,

    You are right the openCONFIGURATOR 1.4.1 is too strict with the mappable datatypes. The upcoming 2.0.0 release of openCONFIGURATOR will support the Real32 and Real64 data as PDO. You can expect a release of the Eclipse framework based version soon.

    Best regards,
    Christoph

     
    • Romain Naour

      Romain Naour - 2015-11-23

      Hello Christoph,

      Thanks for your reply, I rebuilt the openCONFIGURATOR 1.4.1 library and generated correctly the mnobd.cdc.
      I had some isues with the latest boost version (1.59) and I finally used the one used initially (1.54).

      I can't do a runtime test for now due to another problem with a third parties tool but this is another topic :)

      Best regards,
      Romain

       
  • Romain Naour

    Romain Naour - 2016-01-25

    Hello Christoph,

    I'm trying the new eclipse pluging with my current XDC configuration using REAL32 as PDO.

    While adding a new CN with this XDC I get the following error:

    Errors occurred during the build.
    Errors running builder 'openCONFIGURATOR builder' on project 'SlaveBoard_A01'.
    [SlaveBoard_A01] Dynamic channel for datatype: 'REAL32' and direction: 'Rx' not found on MN.
    [SlaveBoard_A01] Dynamic channel for datatype: 'REAL32' and direction: 'Rx' not found on MN.
    

    Do you have any idea ?

    Also I tried to rebuild the eclipse pluging from the sources [1] and [2] but I get several errors from pom.xml:

    Error resolving version for plugin 'org.apache.maven.plugins:maven-site-plugin' from the repositories Plugin not found in any plugin repository
    
    Project build error: Unknown packaging: eclipse-plugin
    

    Do you have a step by step eclipse install procedure to rebuild the plugin ?

    [1] https://github.com/brrueckerc/openconfigurator-core.git (devel branch)
    [2] https://github.com/Ramakrishnan45/openconfigurator-core-plugin.git (master branch)

    Best regards,
    Romain

     
  • Christoph Ruecker

    Hello Romain,

    The OC 2.0.0 is a bit stricter than the former versions in terms of how the MN PI is auto generated. Therefore it checks for a dynamic channel supporting the Real32 datatype which is not existing in the MN XDD shipped with the stack.

    If you add a dynamic channel in the MN XDD with the address space you reserverd than it should be able to generate the configuration.

    e.g.
    <dynamicChannel accessType="readWriteOutput" addressOffset="0000" bitAlignment="0" dataType="0008" maxNumber="XXX" startIndex="XXXX" endIndex="XXXX"/>

    Could you provide us a dummy project with an example of the Real32 usage that we can intregrate in our tests?

    Concerning the build process I will have to check back with Ram what is happening here.

    Best regards,
    Christoph

     
  • Romain Naour

    Romain Naour - 2016-01-26

    Hello Christoph,

    Thank you for your quick reply !

    So, I added this line in the MN XDD

    <dynamicChannel accessType="readWriteOutput" addressOffset="0000" bitAlignment="0" dataType="0008" maxNumber="2032" startIndex="A6C0" endIndex="A6C7"/>
    

    Then I get the following error:

    [ERROR] BaseObject 'PReqActPayloadLimit_U16' with id '0x4' on node '1' exceeds highlimit '36'.
    [ERROR] Object 'PReqActPayloadLimit_U16' '0x1F98/0x4' on node '1' with actual value '0x013C' exceeds the highlimit.
    

    It seems that the default value 36 is used as highlimit. With OC 1.4.1 the upper limit was 1490.

    No problem for the dummy project, I'll create it ASAP.

    Best regards,
    Romain

     
  • Christoph Ruecker

    Hello Romain,

    It seems so. The upper limit should be 1490 so there has to be somethings else going wrong here. Can you provide me the project to reproduce this issue.

    Best regards,
    Christoph

     
  • Christoph Ruecker

    Hello Romain,

    I tried to reproduce the issue and I have an idea what might be the reason for the error. The default CN XDD has the low and hightlimit of the 0x1F98/0x4 object set to 36. Please check your XDDs and change the high limit accordingly. For your convenience I have attached a test project with mapped Real32 and Real64 data.

    Best regards,
    Christoph

     
  • Romain Naour

    Romain Naour - 2016-01-26

    Hello Christoph,

    Indeed, the 0x1F98/0x4 and (0x5) has a wrong hightlimit value.
    With that fixed, I'm able to build the mnobd.cdc.
    Now I need to do a runtime test.

    Thanks,
    Romain

     
  • Christoph Ruecker

    Hello Romain,

    Thx we will fix the encoding issues. Btw fetching from my develop branch might not always be a good idea because the commits can be unstable.

    br
    Christoph

     
    • Romain Naour

      Romain Naour - 2016-01-27

      Hello Christoph,

      Well, when something go wrong with the latest stable reslease it's sometime useful to test with the latest develop branch and try to reproduce the issue. That's why I tried to rebuild the eclipse pluging.
      Also it allow to verify if everything is fine with the next/upcomming release when a bug is fixed :)

      By the way, I think I found an issue with the PDO padding in OC 2.0.0.
      I'll create another thread.

      Best regards,
      Romain

       

      Last edit: Romain Naour 2016-01-27

Log in to post a comment.