{include 9_1.doc|9.0	ELECTRONICS, COMPUTING AND DETECTOR INFRASTRUCTURE

9.1.	Electronic equipment
	G.C. Harper, A.W. Myers and T.D Van Wechel
	Projects undertaken by the electronics shop this year included the 
following:
	a.  Vertical and horizontal steering power supplies, with a -600 volts 
to +600 volts range, were designed and constructed for the terminal ion source 
(see Section 10.3).
	b.  A circuit was designed and constructed for the cluster experiment to
 drive the voltage on the inflector plates with a calibrated ramp voltage 
generated by the multi-channel analyzer.
	c.  A voltage to frequency converter circuit was constructed for the 
cluster experiment to provide an output frequency proportional to the current 
measured by a Keithley picoammeter.
	d.  A portable data acquisition rack, including three NIM bins and a 
CAMAC crate, was assembled for weak interaction studies.
	e.  Preliminary design of the clock generator circuit for the 13/12 high
 energy buncher (see Section 10.7) and design of modifications to be made to the
 low energy buncher controller for phase locking to the 13/12 high energy 
buncher were started.
	f.  The current preamp for the SNO neutral current detector (see Section
 2.3) has been proto-typed and tested.  Ten preamps assembled from discrete 
components are presently being constructed and will be sent to Los Alamos.  The 
major part of the circuitry of the 112 preamps for the neutral current detector 
will be fabricated on hybrid circuits.
	g.  A prototype 30 nsec delay line for the SNO neutral current detector 
has been constructed.  Some development work is still required as there are some
 wave form aberrations on the present prototype.
	h.  An Analog Devices AD606 log amp evaluation board has been tested for
 use with the SNO neutral current detector.  Originally, the rise time of the 
log amp was approximately 400 nsec which is unsatisfactory for this application.
  We were able to modify the circuit to improve the rise time to approximately 
25 nsec.
	i.  Two Ortec 113 preamps were modified.  Ten 270 nSec delay cables and 
several signal and power cables were constructed for the new electron detector 
for the "Mass-8" experiment.
	j.  A Sun microcomputer workstation and CADENCE electronic design system
 software were acquired to facilitate board layout and circuit analysis for the 
SNO and EMIT projects.
}
{include  9_3_4.doc|9.2	VAX-based acquisition systems
	M.A. Howe, R.J. Seymour and D.W. Storm
	Our three data acquisition systems consist of Digital Qbus-based 
VAXStation 3200s running VMS v4.7a.  We use VWS/UIS as the "windowing'' 
software.  Each VAXstation supports a BiRa MBD-11 controlled CAMAC crate.  Our 
principal VAXStation BA-23's cabinet is cabled into a BA-23 CC expansion 
cabinet.  It has an MDB-11 DWQ11 Qbus to Unibus converter driving our old PDP 
11/60's Unibus expansion bay.  That system's peripherals include a CMD 
CQD-220/TM SCSI adapter, a Seagate ST41650 1.38 gigabyte disk, a TTI CTS-8210 
8mm tape, a DEC IEQ11 IEEE-488 bus controller, and a DEC DRV11-J.  The Unibus 
bay contains a DR11-C, our Printronix lineprinter controller and a Unibus cable 
to the MBD-11.
	Our main CAMAC system contains interface modules for our dozen Tracor 
Northern TN-1213 ADCs.  Those ADCs and other CAMAC modules are gated by an 
in-house built synchronization interface, which includes routing-Or 
capabilities, and 32 10-digit 75 MHz scalers.
	Additional CAMAC space is available for our LeCroy 2249s, 2228s and 
2551s.  We have two FERA 4300B ADCs, and two Phillips 7186 TDCs.
	The other two acquisition systems consist solely of the VAXstation's 
BA-23 using an Able Qniverter to cable directly to a stand-alone MBD-11.  Unlike
 the "principal" system, these do not control non-CAMAC-based equipment.
	All three systems run acquisition software based upon TUNL's XSYS, with 
major modifications to their DISPLAY program.  
	TUNL's XSYS software includes an EVAL language compiler which generates 
VAX-native code.  Our "version" of that compiler is limited to 1024 longwords of
 VAX sorting code per MBD channel.  One Mass-8 experiment required far more 
space, so we completely replaced the EVL-generated code section by a 
pre-compiled Fortran subroutine set.  Interface subroutines were created to 
provide easier access to the raw incoming event buffers.  We still require an 
EVL routine to be used, but only for histogram storage coordination.

9.3	Data analysis and support system developments 
	J.G. Cramer, M.A. Howe, R.J. Seymour, D.W. Storm and T.A. Trainor
	All of our offline analysis VAXes are running VMS v5.5-2.  We have 
acquired several additional VAXstation 3200s which have been surplussed from the
 Physics Department.  These are becoming deskside systems, Qbus board-testing 
systems and an additional data acquisition system.  Our cluster now consists of 
five VAXstation 3100s, two VAXstation 3200s, a VAXstation 2000 and an AXP 
3000/400.  Three more 3200s are about to be added to the cluster itself.
	The AXP 3000/400 (Alpha) has 96 megabytes of memory, 1- and 4-gigabyte 
disks, a CDrom, a 19 inch color display, and occasional 8mm Exabyte tape drives 
attached to its external SCSI port.  It is currently running OpenVMS v1.5.  As 
members of the Digital's Campus Software License Grant program, we have 
implemented Digital's Fortran, C and C++ as its principal languages.
	We use TGV's Multinet to provide our cluster with TCP/IP access to the 
Internet.  Our primary Internet address is npl.washington.edu (128.95.100.10).  
The UW Campus dropped all BitNet support last summer, so our only access to 
BitNet systems is via off-campus gateways.
	Our 8 megabyte VAX 11/780 finally died last March.  It has been replaced
 by one of the VAXstation 3200s, with DHv11s now driving our connections to 
twenty-odd local rs232 terminals.  Since the 11/780's boot disk was already on a
 Qbus extension to its Unibus, migration of the disk drives was trivial.  The 
only major peripheral which did not make the transition to the Qbus was our 
9-track tape drive, but it can be driven by a controller attached to another of 
our 3200s.  There has been little demand for its services since the advent of 
8mm systems.
	Another VAXstation 3200 serves as the Linac's control and display 
system.
	Our NA35, NA49 and STAR work depends upon our two HP 9000/710s, which 
have been updated to HP-UX v9.03 and v9.05.  One (Mist) has been moved to 
Germany during John Cramer's sabbatical.  The splitting of the two 
cross-referencing systems was accomplished primarily by adding disks to the 
to-be remote system, and putting Mist's disks on the remaining system.  This 
allowed links and pointers to gradually be migrated to better locations while 
still maintaining the desired level of service to each.  Cramer's system then 
moved to CERN for the NA49 run in November, then back to Munich in December.  
Our remaining 9000/710 (Sand) is providing world-visible World Wide Web (WWW) 
service of programs and data related to those projects, as well as a growing 
library of other NPL documents, including our annual reports (try 
http://sand.npl.washington.edu/home.html).  We have added two HP 712/60 
workstations, also running HP-UX v9.05.
	The November NA49 run at CERN also marshaled all of our Tektronix xp338 
 X-terminals.  They were preconfigured here to ease their arrival in Geneva.  At
 CERN they were hosted by "Mist", which we had configured remotely.  
	The SNO group's arrival added a fleet of networked Macintoshes to this 
otherwise all-VMS (with a few PCs) laboratory.  Their impact on the cluster has 
been little more than traffic on the common ethernet and usage of the Alpha.  As
 a part of the SNO group support, we have installed a Sun SparcStation 20, 
running SunOS v4.1.4, to provide CADENCE circuit layout facilities to our 
electronics shop.  We are still early on the learning curve in providing 
seamless support to that system.  
	Our in-house PCs now include a Pentium P90 system for general lab 
AutoCAD duties.  We are gradually providing TCP/IP software to the PCs.
	We still provide some system management services for the Institute for 
Nuclear Theory and the Physics Nuclear Theory group.  Their three remote-site 
DECstation 5000/200s have been replaced with four AXP 3000/600s, two running 
OSF/1, and two running OpenVMS.  Principal system support for those machines is 
now handled by Physics, with our site providing fall-back assistance and 
additional expertise.  The two groups are still using three VAXstation 3200s 
which are located in our building.
}
{include  9_5.doc|9.4	SControl, an object oriented program for the control of 
large physics experiments
	J.G. Cramer, P.B. Cramer* and M.A. Howe
	SControl is an object-oriented control system construction set written 
to develop the slow control main console for CERN experiment NA49. It was 
written in C++ over a period of seven months and contains more than 250 classes.
  Developed on an HP 9000 series 700 system, SControl has an X11 based user 
interface that is graphic, mouse-driven, and "friendly".  Although developed for
 a particular experiment, SControl is a run-time configured program that could 
find general use. 
	SControl uses a book metaphor to present parameter data on an hierarchy 
of display pages, with each page showing the system or sub-system that is to be 
controlled or monitored.  Connections are made between the pages with 
hyper-links so the page hierarchy can be structured in as much detail as 
required.  Each page can represent the overall system, a sub-system, a 
sub-sub-system, or can be a composite of information drawn from a number of 
systems.  The hyper-links can be made to conform to any irregular shape and can 
be attached to custom bit-mapped images, buttons, or text. 
	Each page consists of a device layer and a logic layer.  The device 
layer contains objects which convey information to the operator.  A large number
 of devices are available and include digital and analog meters, switches, 
sliders, charts, and graphs.  The underlying  logic layer consists of objects 
which are used for data processing and include a wide selection for receiving, 
processing, altering, transmitting, and archiving slow control parameter data.  
Data display or processing objects are selected from pull down menus, placed on 
the screen, and connected into the desired configuration using the mouse.  
Custom built X11 bit map images can be added to the system by placing the image 
file into a file folder.  Images can be used as backgrounds on either the device
 layer or the logic layer. 
	Parameter data comes from network connected satellite processors via 
TCP/IP and is placed in a shared memory data base in the SControl host computer.
  SControl extracts parameter data from the data base as objects which are then 
processed by passing through the connected logic layer objects.  A data flow 
model is used by the processing objects.  Each object processes its data only if
 valid data is present on all of its inputs.  Because the data resides in shared
 memory, more than one SControl program can be running at any given time.
	SControl provides an integrated alarm system page which displays up to 
five levels of alarms from NOTICE to EMERGENCY.  Alarms are triggered as needed 
by connections made in the data processing networks on the logic layers.  All 
alarms display the parameter name, value, and time triggered.  Alarms can be 
acknowledged.  A log is kept of alarm acknowledgment activity.  Clearing a 
serious alarm requires that the operator's name be entered into the log also.  
Alarms can display additional help information that comes from a custom file 
that operators can edit.
	During the November 1994 NA49 test run SControl received data from IBM 
PC and Macintosh computers running LabView and custom software.  The program 
successfully prepared data-bank structures of slow control parameter information
 for the data acquisition system, and maintained a separate time-structured 
archive of operational system data parameters (see Section 7.4).
}
*Max -Planck-Institut für Physik, Föhringer Ring 6, D-80805 München, Germany.

{PAGE|57}