Experimental Physics and
Industrial Control System

Hello Steven

Thanks for the feedback

For information, by selecting the hardware and the protocol (not the Siemens native one but the SLS one (thanks again, Dirk)), we measured an average transmission time of 25 ms with S7/300m, 12 ms with S7/400, for 8KB frames on a 100 Mbps Ethernet. 
But of course, average is not worst-case.

Still, I remember at CERN having control commands to Siemens S7/300 from a RT OS over a local switch (Linac pulsed power supplies) within a strict 200 ms time window. So it's feasible ... maybe too costly for most applications.

We'll continue tests and give new results when we have some. 

Cheers,
Franck
-----Original Message-----
From: Steven M. Hartman [mailto:[email protected]] 
Sent: 27 September 2011 23:32
To: Di Maio Franck
Cc: [email protected]
Subject: Re: [S7plc EPICS driver] REAL type record jumps to zero

Di Maio Franck wrote:

> But I also wonder if others with similar requirements (predictable traffic with PLC) are using a dedicated network interface to separate the PLC traffic from CA.

Hello Franck--

How predictable do you require, and how fast are you expecting for PLC to EPICS communications? For SNS, we are not relying on tight timing requirements between IOC and PLC (typically ~1 second is
adequate) and we can tolerate a long timeout (several seconds). 
Anything with lower tolerance is handled within the PLC (or within the IOC).  With this expectation, we do not see anything like your flicker, and when we have seen problems, it has generally been traced to a buggy firmware version on the PLC or bad hardware (e.g. 
damaged network cable). The SNS network is relatively flat and has lots of traffic. A dedicated network may be beneficial, but I think in our case, the inefficient network protocol to the AB PLCs is more of a limitation then the network capacity. I don't have experience with the S7 PLCs for comparison.

Regards,
--
Steven Hartman
[email protected]

References:

[S7plc EPICS driver] REAL type record jumps to zero Pavel Masloff

RE: [S7plc EPICS driver] REAL type record jumps to zero Di Maio Franck

Re: [S7plc EPICS driver] REAL type record jumps to zero Steven M. Hartman

Navigate by Date:

Prev: RE: Sequence monitor not getting callback Zelazny, Michael Stanley

Next: Fw: Simple ADC - Ethernet device? Mark Davis

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 

Navigate by Thread:

Prev: Re: [S7plc EPICS driver] REAL type record jumps to zero Steven M. Hartman

Next: looking for RFM2g driver for RTEMS Dan Eichel

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