Experimental Physics and
Industrial Control System

[email protected] (Bob Dalesio) · Mon, 17 Jun 96 16:18:08 MDT

----------
X-Sun-Data-Type: text
X-Sun-Data-Description: text
X-Sun-Data-Name: text
X-Sun-Content-Lines: 33

Dear All,

To distribute the meeting notes, the topic is listed, along with a brief
description, an e-mail address and a URL where possible. Please contact the
presenter directly if you would like more information on the topic.

List of Attendees also attached. This list was only passed around on the
last day. I tried to add everyone that missed the last day. I tried to not add
those casual attendees - that were not registered. My apologies for either 
including or excluding your name - contrary to your wishes.

Before going through the agenda, I would like to solicit comments about the
meeting. We started these meetings as work sessions for collecting requirements,
settting priorities, discussing design, reviewing problems and just having
an oppurtunity to talk about any technical issues. As the collaboration is
now over 50 projects large, these meetings have drifted away from this -
but not in any clear direction. I would greatly appreciate any advice or
observations about these meetings. What would you like to get from them -
any ideas about format or organization - any comment on length - any
ideas about content. In this last meeting, we covered a broad spectrum
of topics - from advanced control to I/O bit manipulation. We had two
workshop type sessions - one to discuss the problems associated with detectors 
and one to discuss archiving requirements. All attendees should view these
meetings as an oppurtunity to discuss problems that exist, form work groups
around similar interests, solicit opinions from the community, give feedback
to help improve the usefulness of existing tools, offer new approaches and
solutions that others in the community could use, have a little fun, or
solicit training. Should we have smaller/ parellel sessions? Thanks for taking
the time to consider these questions. As always, we will attempt to build
future meeting agendas based on your input.
	Regards,
	Bob

----------
X-Sun-Data-Type: default
X-Sun-Data-Description: default
X-Sun-Data-Name: 596mtgnotes
X-Sun-Content-Lines: 454

==========================================================================
Brief Notes on the topics covered during the meeting:
--------------------------------------------------------------------------
Communication/API services in the EPICS environment

*Portable CA Server - Jeff Hill - [email protected]
	The portable CA server is nearing completion and available in
	Alpha release. It is currently running under UNIX. It is being
	incorporated into the Gateway being developed at ANL. The
	discussion covered the API for making any data store available
	to the EPICS channel access protocol - and therefore available
	to all clients. This is one method for extending the EPICS
	architecture to incorporate legacy systems and/or data stores
	other than the EPICS database.	
*Gateway Update - Jeff Hill (for Jim Kowlakowski - [email protected])
	The Gateway is being developed to provide a single connection
	into the principal control network for clients that do not
	require real-time connections - and whose traffic has a lower
	priority than the operation of the machine. The requests of
	all low priority clients are connected through this gateway.
	Many connection to the same pararmeters will result in a single
	request to the control network.
*CDEV Update - Chip Watson - [email protected]
	CDEV is an API being developed to provide a higher level view
	to the flat database architecture of EPICS. In addition, CDEV
	provides access to other data stores by adding services to the
	CDEV API. Currently, CDEV clients include some high level analysis
	tools. The services include: channel access, ACE, and the SLAC
	data access. This API provides a decoupling of the source of the
	data from the high level applications to monitor, analyze and
	control the process.
	off of - http://www.cebaf.gov/cdev
--------------------------------------------------------------------------
*WWW Documentation
	A brief dicsussion was held to determine the state of the documentation
	and in particular - what trans-oceanic issues needed to be handled.
	It was stated that the mirror site in England was not always a good
	alternative in the rest of Europe. Other mirror sites will be supported.
	We will wait to see where these will be. 
	The other major issue discussed here is paper size. Roland Mueller
	of BESSY has provided a proposal - the authors need to support it...
	http://www.bessy.de/comput/FramePrintOuts.html
--------------------------------------------------------------------------
*CVS Bug Fix releases
	A brief disucssion was held to derive a policy for bug fix releases.
	It is the position that major releases of EPICS base will continue
	to go through rigorous test and be released about once per year. Bug
	fixes will be made available between major releases with some cursory
	testing. The problems will be documented so that any site wanting a
	release can choose between the rigorously tested version and the
	more recent "fixed" version. Bug releases will be made no more
	frequently than once per month. Any more pressing fixes can be
	obtained and built at your local site.
---------------------------------------------------------------------------
High Level Client Applications

*MultiKnob Application - Walt Akers - [email protected]
	MutliKnob is a CDEV application that allows users to connect to a named 
	set of CDEV devices using a Compound Device service. Compund Device 
	service allows this service to be connected to itself - allowing nested 
	MultiKnob applications. The value of the devices can then be adjusted 
	using a single control while maintaining a defined relationship.

*Correlation Plot - Dave Wetherholt - [email protected]
	This application (in conjunction with X-Act) allows users to setup
	data taking experiments. It gives control over an arbitrary number
	of nested controllers and takes data after each change. The data
	can then be analyzed (plotted) as a function of time - or a function
	of other data. The data taken engine is a CDEV client.
	related doc - coming soon to the CEBAF WWW site

*SDDS Update - Claude Saunders - [email protected]
	Self-Describing Data Sets is a file format that is used at APS for
	passing data between modular data collection, analysis and visualization
	tools. An argument rich set of command line utilities are used to
	aid the accelerator physicists in trouble shooting and optimizing the
	operation of the accelerator. As useful scripts are created - they
	are converted to operational tools by providing a tcl/TK GUI to the
	script for ease of use. The file format also facilitates modular
	extensions.
	related docs @:http//www.aps/anl.gov/asd/oag/aoghome.html

*Bumpless Reboot - Bob Dalesio - [email protected]
	A set of routines were presented that save ascii save sets - and
	then either restore them into running databases through channel
	access or into newly loaded databses before reboot using the
	static database access library. These are raw routines with no
	user interface. The configuration file and savefile are straight
	ascii.
	[email protected] http://www.atdiv.lanl.gov/doc/epics/training/Slides.html

*Framework for an Online Data Flow Prototype - applied to data acquisition 
- Joe White - [email protected]
	This package provides a flexible, transparent transport mechanism among 
	sections of user code. It presents a state machine API, a set dynamic 
	of configuration blocks, and a set of monitoring blocks to any external 
	systems. The state machine API should be driven by a run control system 
	since the framework does not have any internal automatic state machine 
	model running. The configuration and monitoring blocks are intended for 
	user parameters such as calibration parameters and counter limits. 
	Internally, the framework allows the user code to access a set of data 
	flow services, such as event buffer management. The kind of transport 
	(called a link) connecting blocks of user code can range from a simple 
	direct call forward to a socket connection across the network. Since 
	the kind of link is transparent to user code, a given block can be 
	located in a separate process or moved to another machine by only 
	changing the setup, recompiling, and reloading. A link can also begin a 
	chain, for example by accepting an interrupt from hardware, or end a 
	chain, for example by dumping the data to tape and freeing the 
	associated event buffer.
	As part of this project, a run control and configuration driver 
	application (called Orc for Online Run Control) was developed to 
	interface with the Odf framework. This application uses several snl 
	state machines with EPICS database components for flow control and gui 
	control.
        http://www.slac.stanford.edu/BFROOT/doc/www/Computing/Online/Online.html

*Feedback System for Tristan Light Source Study - Mimashi Toshihiro 
- [email protected]
	The beam orbit feedback system was designed and built for the TRISTAN 
	Light-Source Study. Some of the experiments which use TRISTAN Main Ring
	as a synchrotron-light source require beam position and slope stability 
	at the light source. The beam orbit feedback system eliminate the 
	vibration of the beam at the center of undulator. The feedback system 
	was composed of 2 single path beam monitors, 4 of each horizontal and 
	vertical steering magnets. The control system was built with EPICS. The 
	frequency of the feedback loop was set to 100 Hz. And the feedback 
	system satisfied the tightest requirement from the experiments. They 
	are +- 1500 um position stability and +- 15 urad slope stability in 
	horizontal direction, and +-50 um and +-5 urad in vertical direction.

*Fast Feedback for CEBAF	- Mahesh Chowdhary - [email protected]
	A 60 Hz prototype feedback system to regulate the beam energy 
	has been beta tested at CEBAF. The 60 Hz loop was run over
	ethernet through a high priority dedicated vxWorks task. The 
	optimal control algorithm for this system was developed in 
	state space formalism of Modern Control Theory. The closed loop
	bandwidth of this system running at 60 Hz rate is 7 Hz. 
	This feedback system will be implemented to run at a rate around
	1.2 kHz in order to suppress all primary components of disturbances.
	The highest noise frequency is 180 Hz.
	http://www.cebaf.gov/~mahesh contains papers on this system

*The accelerator modelling codes Structured Accelerator Design/Strategic 
 Accelerator Design (SAD) - Mimashi Toshihiro - 
 (for Noboru Yamamoto [email protected])
	These codes are developed in the KEK accelerator department. SAD treats 
	many of beam dynamics calculations, ie. beam tracking, optics 
	calculation, etc. We used SAD for TRISTAN accelerator operation. The 
	global orbit correction, local orbit correction, optics generation, 
	bump creation, tune adjustment, chromaticity correction etc, have been 
	done with SAD. Since SAD will take a very important role in the KEKB 
	accelerator control, it was connected to EPICS. Usually SAD input is 
	ASCII file. For example when we correct the beam orbit, SAD read beam 
	position monitors data and current steering magnet kick angle from 
	ACSII file, write corrected steering magnet kick angle and expected 
	orbit in the file. Now SAD was modified to read steering magnet kick 
	angle and beam position monitors value directly from EPICS channels and 
	it can set steering magnet kick angle directly. The graphical user
	interface provides the operators with easy access to these modelling 
	codes.

---------------------------------------------------------------------------
*PCs - in EPICS (hosted)
	The purpose of this session was to determine what directions people
	were considering with regard to PC technology - mass market - low
	price hardware and software. The question was posed - hoping to form
	some teams to work together or at least avoid overlap...

	channel access client ported to Windows NT and windows '95
		fully operational for a long time
		LBL - Chris Timassi - [email protected]
		benchmarks at APS beamlines reported to show ~Sparc10 thoughput
	port of IOC code to the 486/Pentium running vxWorks
		Build ioc core using EPICS make files - largely complete
		Run the database on the Pentium/vxWorks - working
		Portable Channel access server to the Pentium - just starting
		Performance of database roughly equivalent to the 68060
		LANL - [email protected]
		APS will reintegrate the build/install changes needed
	port of IOC code to the PowerPC
		KEK - Noboru Yamamoto [email protected] 
		- scheduled to test this year
		APS - Jim Kowlakowski - [email protected]
		- board in hand - needs some time
		BNL (Brahms) - Karim Ashktorab - [email protected]
		- board in hand - needs WRS delivery
	video processing with Pentium and video boards on the PCI bus
		DESY - Matthias Clausen - [email protected]
	port IOC database functiuonality to NT (very small/informal effort)
		LANL - [email protected]
	X-display running on a PC from a UNIX platform
		Many projects are already doing this....
	move display manager to windows NT running X
		Again - please announce interest (ability to do the work)
	create a development environment under WNT for databases
		LANL - Capfast - should already work - untested
	have a WNT display manager
		APS / LANL / Tate - want to start work here any others?
		[email protected]
	port the portable ca server to the NT environment
		MIT - Coles Sibley - [email protected]
	Native displays on NT/W96
		Using Delphi, IDL, Visual Basic, PV-Wave
		APS/CATs - Mark Engbretson - [email protected]

	In the long term - the portable CA server, display tools under NT,
	and a database port to NT (or use of a cohabitating vxWOrks and NT)
	will lead to the ability to run a single box solution that can be
	scaled up or integrated into a large installation.
-------------------------------------------------------------------------
*EPICS software distribution homepage - Bakul Banerjee - [email protected]
	Bakul presented the organization of the distribution page, the
	documents and the verification process for obtaining a password.
	A very nice web site is set up to support distribution at
	http://epics.aps.anl.gov/xfd/WWW/xfd/SoftDist/Welcome.html
--------------------------------------------------------------------------
*Applications framework - Bjorklund, Kraimer, Wright, Lewis, Tang
	EPICS distributions do not include a standard environment for developing
	applications. These frameworks are crucial to the productive use
	of these tools by those responsible for getting the applications done.
	This decision should be made in an informed manner with those 
	responsible for the maintenance of the EPICS and project application
	codes fully involved.
	Four prominent environments were represented. They include:
	CEBAF - scalable - independent versions for IOCs fast fallback
		Johnny Tang - [email protected]
		http://devserve.cebaf.gov/~epics/cebafEpicsAppArch.html
	BaBar - installations distributed over many sites - small apps at each
		Joe White -  [email protected]
		http://www.slac.stanford.edu/BFROOT/doc/www/Computing/Online
		/Reference.html
	UAE - flexible - in use on many medium size sites (~4K signals)
		Nick Reese - [email protected] 
		William Lupton - [email protected]
		Eric Bjorklund - [email protected]
	APS - large installation - 
---------------------------------------------------------------------------
Solutions to application problems at various labs

*BESSY Status
	-EPICS experiences @ BESSY (Europe): Quick start up, no serious 
	 problems so far
	-Graphical Server + API for GUI generation using Motif (BESSY~II)
	 Any required Widget can be embedded, abstact, event driven API with
	 no knowledge of the presentation layer needed, code is independent 
	 of the graphical paradigm. It uses full cdev functionality.
	 generic graphics client allows to execute scripts containing cdev calls
	-Configuration tool based on ora_tcl
	-CAN device Support/Driver 
	 Multi-drop industrial standard I/O bus using the VME-CAN2 interface 
	 card

*NASA Wind Tunnel Applicaitons - Steve Syrett - [email protected]
	A short presentation was given on the control of wind tunnels.
	Most of this discussion focused on the plan at NASA to port
	the portion of EPICS currently running under vxWorks to LnyxOS. 
	Connections will be made with Kazuro Furukawa, who took a look
	at this port while at CERN. The dicsussion also lead to a plan
	to first port channel access and the database into a single
	memory space (i.e. no thread protection) and limit the scope to
	determine what functions rely on the linking loader.

*DESY Status - Matthias Clausen - [email protected]
	-Common Driver Interface - http://herao1.desy.de/heras1/cdi_intro.html
	 This common driver interface simplifies the integration of new drivers
	-CAN-CDI driver - http://herao1.desy.de/heras1/CAN_Driver.html
	 This is an industrial standard multi-drop I/O bus that uses the CDI 
	-Power supply record - http://herao1.desy.de/cc/www/doc/pscdoc.html
	 for controlling and synchronizing control of various types of power 
	 supplies
	- Fermilab Internet Rack Monitor as an IOC
	 http://herao1.desy.de/cc/www/doc/irm.html
	 a low cost IOC with a 68040/64 16-bit AI/8 12-bit AO/64 binary in or 
	 outs/arrays on all 64 AIs/latched clock events

*Message Based functionality in a Value Based System 
- Andy Foster RGO [email protected]
	The hierarchical structure of the Gemini software system implies that
	commands will be sent between EPICS databases running in separate IOC's.
	It is a requirement of the software system that we are able to
	download complete observational configurations between the top-level
	Observatory Control System and lower-level systems. In order to meet 
	these requirements, 4 new 'soft' records have been written. These are 
	the APPLY, CAD, CAR and SIR records. 
	CAD - Command Action Directive
	CAR - Command Action Response
	SIR - Status Information Record
	Typical use form a commanding system would be
	(1) Write the data associated with each command to the appropriate CAD.
	(2) Send the directive START to the APPLY record. 
	(3) The commanding system monitors the VAL field of the APPLY to see if 
	   each command has bee accepted or rejected by the CAD records.
	(4) CAR record is the state machine that reflects the status of the 
	   command. Typically this is a hierarchy of CAR records.
	For further info: ftp.ast.cam.ac.uk:/pub/ajf in the file: cadcar.tar.Z.
---------------------------------------------------------------------------
*Physics detector workshop - Pabien Perriollat & Helfried Burckhart
	-This workshop was an oppurtunity to talk about where EPICS
	 fits and does not fit into large detector run control.
	Areas cited:
	-Hierarchical state control - sequencer provides this functionality
	-Exception Handling - 3rd alarm layer needed - an easy modification
	-Idea of namespace for large arrays of similar items -
		This provokes the most discussion. There were several approaches
		mentioned - database array records - large name space - CDEV
		groupings - this problem dominates the issue of applying EPICS
		to large detector controls. 
	-Handling Name space and dynamic connections for 1 million signals
		This was discussed as part of SSC earlier. Distributed name
		server - name caching in the client - were some of the solutions
		for handling the runtime issue. Managing the large name space
		is also an issue to address.
	-It would be worthwhile to convene a group to address these issues 
		directly and in more detail.
----------------------------------------------------------------------------
*3.13 Changes for 3.13 - Marty Kraimer [email protected] and 
  Bob Dalesio [email protected]
	-The database and record support are now loaded as ASCII.
	-All address fields are available over channel access as strings to
	 read and modify. NOTE: not all device support will react 
	 propoerly to this modification - but most should - TEST!
	 Also NOTE: lockset will be locked for 5 msec (100 records 25MHz 040) 
	 during the change
	-New address link options:
	 FWDLNK 's to channels in other IOCs that reference teh .PROC field will
	  cause the record to be processed
	 OUTLINK now supports a -CA which makes the link a channel access link
	  even if the record is in the same IOC.
	 INLINK now support CA and CP - CP says process when a monitor occurs
	- Some impacts
	 Database addresses that are channel access links are > 2x faster
	 Channel access client access to the database is 20% faster
	 Database fast links are 15% slower
	 The link fields were reduced in size. Less memory usage.
	-Locally developed record support will need to use the new
	 database access routines for processing address fields
	off of: http://epics.aps.anl.gov/asd/controls/epics/EpicsDocumentation
		     /WWWPages/EpicsFrames.html in RelNote.html
------------------------------------------------------------------------------
*Static Database Tools - Rozelle Wright [email protected] and 
  Tatsuro Nakamura - [email protected]
	http://www.atdiv.lanl.gov/doc/epics/epicshm.html 
	- dbt -- A command line tool (optionally interactive)
	 to invoke static database functions to check the validity of databases 
	 and create additions to edb.def (a capfast configuration file)
       	 from new records in .db format.
	- db2capsym -- a command line tool to create capfast graphical symbols 
	 from new records in .db format.
	- e2db -- An upgrade of the e2sr tool to transform capfast edif files to
	 database files in .db format.
	-An alpha version of these available by anonymous ftp in the very near 
	 future.	
------------------------------------------------------------------------------
*How to handle an output that could change from another source
	This is just a note to point out that the all output record/device
	support does not read the value back periodically to determine if it was
	changed by a local controller. To implement this - either some database
	logic - or changes to the record device support are required. As long
	as the changes are made through the database - it is OK.
------------------------------------------------------------------------------
Data Archiving - Note that a working group is forming to address this
- send interest in participating to [email protected]

*Archiver Requirements - Matt Bickley - [email protected]
	Requirements for data archiving were presented. These requirements will
	be the functional requirements for an archvier that is being designed
	for CEBAF. It is the goal to make this a general useful tool and have
	a group from several laboratories contribute to the effort. Please
	review and comment on the talk taht is posted on the web. If you have 
	resources to work on this - please contact Matt. The document will be
	available for final comment soon.
	http://devserve.cebaf.gov/external/talks/ArchiverReq.html

*Archiving at DESY - Matthias Claussen - [email protected]
	An archiver is developed in a modular fashion that includes pieces
	for archiving in the IOC with an archive record as a local buffer, 
	archiving any channels (including the archive record) to files that 
	keep a channel per file. The archiver runs on both VMS and UNIX. 
	The archive retrieval routines provide data as a service over CORBA 
	to a CORBA client that provides a archive viewer to the operator. The 
	archive viewer provides for channel selection, plotting, rotation, 
	and a rich collection of 2-D and 3-D plots written in IDL. It is
	currently archving 2900 @ 1 Hz.
	docs @ http://herao1.desy.de/cc/www/info/archive.html
	off of http://herao1.desy.de/epics_add_ons.html
------------------------------------------------------------------------------
*CAMAC device support improvements - Eric Bjorklund [email protected]
	(Johnny Tang - [email protected]) (Mark Rivers - [email protected])
	waveform record support is added
	Hytec VSD2992 serial driver and KSC 2917/3992 parrallel driver supported
	cclnk routine added, LAMs supported, diagnostics added 
	available over anonymous ftp: thoth.atdiv.lanl.gov /pub/epics/camac.tar
-----------------------------------------------------------------------------

attendees:
Eric Bjorklund		[email protected]
Marty Kraimer		[email protected]
Bob Dalesio		[email protected]
Toshi Mimashi		[email protected]
Chip Watson		[email protected]
Matthias Clausen	[email protected]
Jeff Hill		[email protected]
Steve Lewis		[email protected]
Roland Mu"ller		[email protected]
Richard Sachtschale	[email protected]
Fred Carter		[email protected]
Don Dohan		[email protected]
Ed Desmond		[email protected]
Bret Goodrich		[email protected]
Andy Foster		[email protected]
Bakul Banerjee		[email protected]
Hamid Shoaee		[email protected]
Jeff McDonald		[email protected]
Terry Heuring		[email protected]
David Myers		[email protected]
Fabien Perriollat	[email protected]
Tim Mooney		[email protected]
Claude Saunders		[email protected]
Janet Anderson		[email protected]
Helfried Burckhart	[email protected]
Kagarmanov Albert	[email protected]
Graham Waters		[email protected]
Rolf Keitel		[email protected]
Frank DiMaio		[email protected]
Sebastien Fabbro	[email protected]
Xingyue Li		[email protected]
Charlie Briegel		[email protected]
Mark Lucas		[email protected]
Deb Kerstiens		[email protected]
Johnny Tang		[email protected]
Walt Akers		[email protected]
J. Frederick Bartlett	[email protected]
Valarie Bookwalter	[email protected]
Thomas Blumer		[email protected]
Anton Mezger		[email protected]
Stephanie Allison	[email protected]
Ric Claus		[email protected]
Joe White		[email protected]
Rozelle Wright		[email protected]
John Winans		[email protected]
Coles Sibley		[email protected]
Matt Bickley		[email protected]
Anna Estes		[email protected]
Steve Syrett 		[email protected]
Ying Wu			[email protected]
Tom Clifford		[email protected]
Don Shea		[email protected]
Dave Wetherholt		[email protected]
Bruce Bowling		[email protected]
Marty Knott		[email protected]
Dave Gurd		[email protected]
Mike Thuot		[email protected]
----------
X-Sun-Data-Type: default
X-Sun-Data-Description: default
X-Sun-Data-Name: 596attendees
X-Sun-Content-Lines: 59

attendees:
Eric Bjorklund		[email protected]
Marty Kraimer		[email protected]
Bob Dalesio		[email protected]
Toshi Mimashi		[email protected]
Chip Watson		[email protected]
Matthias Clausen	[email protected]
Jeff Hill		[email protected]
Steve Lewis		[email protected]
Roland Mu"ller		[email protected]
Richard Sachtschale	[email protected]
Fred Carter		[email protected]
Don Dohan		[email protected]
Ed Desmond		[email protected]
Bret Goodrich		[email protected]
Andy Foster		[email protected]
Bakul Banerjee		[email protected]
Hamid Shoaee		[email protected]
Jeff McDonald		[email protected]
Terry Heuring		[email protected]
David Myers		[email protected]
Fabien Perriollat	[email protected]
Tim Mooney		[email protected]
Claude Saunders		[email protected]
Janet Anderson		[email protected]
Helfried Burckhart	[email protected]
Kagarmanov Albert	[email protected]
Graham Waters		[email protected]
Rolf Keitel		[email protected]
Frank DiMaio		[email protected]
Sebastien Fabbro	[email protected]
Xingyue Li		[email protected]
Charlie Briegel		[email protected]
Mark Lucas		[email protected]
Deb Kerstiens		[email protected]
Johnny Tang		[email protected]
Walt Akers		[email protected]
J. Frederick Bartlett	[email protected]
Valarie Bookwalter	[email protected]
Thomas Blumer		[email protected]
Anton Mezger		[email protected]
Stephanie Allison	[email protected]
Ric Claus		[email protected]
Joe White		[email protected]
Rozelle Wright		[email protected]
John Winans		[email protected]
Coles Sibley		[email protected]
Matt Bickley		[email protected]
Anna Estes		[email protected]
Steve Syrett 		[email protected]
Ying Wu			[email protected]
Tom Clifford		[email protected]
Don Shea		[email protected]
Dave Wetherholt		[email protected]
Bruce Bowling		[email protected]
Marty Knott		[email protected]
Dave Gurd		[email protected]
Mike Thuot		[email protected]

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System