| 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 <2019> 2020 2021 2022 2023 2024 | Index | 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 <2019> 2020 2021 2022 2023 2024 |
| <== Date ==> | <== Thread ==> |
|---|
| Subject: | Re: [EXTERNAL] Beckhoff BK9000 support survey |
| From: | Jeong Han Lee via Tech-talk <[email protected]> |
| To: | Torsten Bögershausen <[email protected]>, Hinko Kocevar <[email protected]>, "Dunning, Michael" <[email protected]>, "[email protected]" <[email protected]> |
| Date: | Mon, 1 Jul 2019 10:59:38 -0400 |
Hi All, Thanks Torsten for this clear answer.Still, not for the production, but I think, it is enough to see the ethercat master (open source) performance to match a PLC level.
I am sure that test results are limited due to "kernel version, RT configuration, OS type, and Intel hardware".
It is a good subject which we can present to the EPICS community this Fall, or next Spring meeting.
FYI, the test platforms are* CentOS 7 (RH 3.10 kernel + CERN RT packages) with ethercat generic and native e1000e driver
* Debian 9 (Vanilla 4.9 kernel + Debian RT packages) with ethercat generic and native e1000e driver
* On several old Intel NUC PCs, and Concurrent CPU (old, discontinued)We saw the performance of the native driver (e1000e) is much better than a generic one, and Debian is better than CentOS. :)
HTH, Han On 7/1/19 7:01 AM, Torsten Bögershausen wrote:
Hej Hinko, the short answer: No. The longer answer: The BK9000 is part of a series of terminals that are from the pre-EtherCAT time. Still useful, still in use, terminals are named EKxxxx. What ESS is heading for is EtherCAT as a standard, and that does exclude the BK9000. The general question is, why do you (or anybody else) want to use the BK9000 together with EtherCAT sustems (the ELxxxx terminals) So in theory the ECMC can be enhanced to talk to the BK900. but we don't see a use case for this. (And feel free and invited to visit us in the motion lab). Another question; If somebody want to use ADS (instead of modbus) to connect the EPICS IOC with the Beckhoff world, please let me know. /Torsten On 01/07/19 11:30, Hinko Kocevar via Tech-talk wrote:Hi Han,would the asyn driver that ESS has work with the coupler in subject - BK9000?Thanks, //hinko ________________________________________From: [email protected] <[email protected]> on behalf of Jeong Han Lee via Tech-talk <[email protected]>Sent: Sunday, June 30, 2019 4:26:26 PM To: Dunning, Michael; [email protected] Subject: Re: [EXTERNAL] Beckhoff BK9000 support survey Hi Mike, ESS has a bit unique and generic EPICS support based on asyn. Please look at the following talkhttp://accelconf.web.cern.ch/AccelConf/icalepcs2017/talks/mocpl05_talk.pdfWe've called it as ecmc (EtherCat Motion Control), but we extend itto generic I/O supports. The current support modules in the enclosed file.Unfortunately, they are within the closed repository, but we are willing to share it within EPICS community. HTH, Han On 6/27/19 5:38 PM, Dunning, Michael via Tech-talk wrote:Thanks everybody for the responses so far. I should add that, as Davide mentioned, for some modules we need to access the Beckhoff "hidden" registers which can only be accessed through other registers. This makes the modbus "driver" approach necessary if we want to cover all modules. In our case we've needed to access these registers for changing configuration parameters, e.g. changing thermocouple types or ADC scaling, or for motor configuration. Mark - thanks for pointing out the new C++ version of modbus. It sounds like we could get rid of our custom modbus driver code and instead call modbus functions directly from our asynPortDriver. This should simplify things and make maintenance a bit easier. This definitely sounds like something we should pursue. Thanks again to all who have responded. Mike Michael Dunning SLAC National Accelerator Laboratory 2575 Sand Hill Road Menlo Park, CA 94025 (650) 926-5200On Thu, Jun 27, 2019 at 11:52 AM Mark Rivers <[email protected]> wrote:I would only do this if the application is complex enough that you need to move logic from the database down into C++ code. I'd still make use of the standard modbus support module and use it like a library to handle the modbus protocol (since it's been around for 10+ years and is well tested, that would be my starting pointIn the new asynPortDriver branch you can do the following: drvAsynIPPortConfigure("Koyo1","camaro:502",0,0,1); asynSetOption("Koyo1", 0, "disconnectOnReadTimeout", "Y"); modbusInterposeConfig("Koyo1", modbusLinkTCP, 5000, 0); // Use absolute addressing, modbusStartAddress=-1.drvModbusAsyn *pModbus = new drvModbusAsyn("K1", "Koyo1", 0, 2, -1, 256, dataTypeUInt16, 0, "Koyo");// Write 10 bits at address 2048 memset(data, 0, sizeof(data)); data[0] = 1; data[2] = 1; data[4] = 1; data[6] = 1; data[8] = 1; printf(" Writing [1 0 1 0 1 0 1 0 1 0] to adddress 2048\n");/* asynStatus doModbusIO(int slave, int function, int start, epicsUInt16 *data, int len); */ pModbus->doModbusIO(0, MODBUS_WRITE_MULTIPLE_COILS, 2048, data, 10);Because the drvModbusAsyn constructor was called with startAddress=-1 doModbusIO can write to any address with any function code.Mark Sent from my iPhoneOn Jun 27, 2019, at 12:23 PM, Pearson, Matthew R. via Tech-talk <[email protected]<mailto:[email protected]>> wrote:Hi,One version uses only the epics modbus module. This has no driver so requires less maintenance, but makes setting up IOCs more time consuming and results ina lot of code duplication.If the only problem with this method is the code duplication in each IOC then it's best to move the database into templates in a support module. Then all the IOC has to do is include them in a substitutions file. You could have one template per module type. And if there is common database code between modules then separate that out into common template files that are included in the per-module templates.Then in the IOC startup script you may have a large list of calls to drvModbusAsynConfigure in order to setup the modbus ports for different address ranges. You can also put this in the support module, and just include it in the IOC st.cmd, passing in macros as necessary. For example, this is what I do for one of my applications for the Moxa ioLogik modules:#E1214 Unit (6 DI and 6 Relay) epicsEnvSet("IP_ADDR","192.168.200.177:502") epicsEnvSet("IP_PORT","m1ip") epicsEnvSet("PORT","m1") < $(MOXA)/st_scripts/st_common.cmdWhere st_common.cmd is just a list of calls to drvModbusAsynConfigure. You may or may not need a different list for each type of module, depending on the modbus registry maps.Another version uses asynPortDriver and some custom modbus code. This is designed to make setting up IOCs easier, but requires maintenance of the drivercode.I would only do this if the application is complex enough that you need to move logic from the database down into C++ code. I'd still make use of the standard modbus support module and use it like a library to handle the modbus protocol (since it's been around for 10+ years and is well tested, that would be my starting point).Another uses custom epics device support. This requires writing device support for each bus terminal, and has resulted in a pretty ugly codebase. This is ourleast favored solution going forward.There's no need to write support this way anymore, since Asyn gives you the device support for free.Cheers, Matt