Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Wed, 7 Nov 2007 08:31:55 -0700

Ron,

There is, in fact, some work in progress on an IOC based (server based)
event filtering upgrade for EPICS - which will have the potential to be
based on beam flavors, pulse counters, and any other site or device specific
parameter stored on the event queue.

http://ics-web4.sns.ornl.gov/icalepcs07/WPPA24/WPPA24.PDF

Also, several client based time correlating support library implementations
already exist for EPICS buffering the subscription updates and returning a
complete set of time correlated data based on some time synchronization
criteria. The most mature implementation is, I suspect, supplied with the
XAL package developed and in use at the SNS, and elsewhere. It's true that
the RF pulse count might be converted to time, but for a general purpose
system like EPICS event correlation out of a buffered event stream using
only time stamps has been demonstrated to be a flexible client based
approach.

Jeff

> -----Original Message-----
> From: [email protected] [mailto:tech-talk-
> [email protected]] On Behalf Of Ron Rechenmacher
> Sent: Wednesday, November 07, 2007 7:42 AM
> To: [email protected]
> Cc: [email protected]
> Subject: synchronizing data to an event
> 
> Hi,
> 
> I did a search of tech-talk for: timing event synchronization
> The results along with my experience to date leads to the conclusion
> that synchronization by 8 byte time stamp is the only synchronization
> that
> CA currently provides. (please correct if wrong)
> 
> The question:
> 
>   I was wondering if there has ever been any (perhaps behind the
> scenes)
>   discussion of adding additional synchronization information to the
>   Channel Access protocol?
> 
> 
> Perhaps, in addition to the timestamp, a "synchronizing event"
> structure,
> which would be NULL if not used, but would otherwise have an event type
> and
> event count.
> 
> At some linear accelerator facilities, there has been a move toward
> counting
> RF pulses and associating data collection triggered by the RF pulse
> with that
> pulse count.  Of course there would be a pulse count to time
> correlation also,
> but many people seem to embrace the main data association (in archived
> data)
> being with an event count.
> 
> I've heard that there has been work on "beam flavors" at SLAC. (sorry
> if this
> is bogus information). But if in addition to a timestamp, data also has
> "flavor" information, then this could be considered a start to
> additional
> synchronization information?  Has there been work on this? and if so,
> has any
> progress reports been published?
> 
> Thanks,
> Ron

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System