Held at CERN 5./6. May 1998

AGENDA:

5.5.

• 14:00 welcome and introductory remarks
• 14:10 short presentation of the existing systems (10-15 Min every group)
• 15:40 break
• 16:00 short presentation of the planned systems as far as there is already something planned (10-15 Min each)
• 17:00 LabVIEW introduction and demonstration

• 09:00 summarising weak points of the existing systems that should be avoided in the next generation of control systems.
• 09:30 Discussion: demands a future system should fulfil
• 10:30 Discussion: How to go for a common experiment control system?

After the welcome by the organiser it was emphasised that the topic to be discussed in this workshop is a system which allows both experiment control and online evaluation of the collected data.

The introduction was followed by the presentation of the existing systems of the groups attending the workshop.

First Dietrich Beck presented the ISOLTRAP system. This system was intended to be a control system, which is almost independent of the experiment. It is based on a motorola processor controlled VME bus running OS-9 and a PC running Windows (3/95/NT) while the programs are written in C and C++, respectively. The experiments having used it or still using it are located at the University of Mainz (e.g. the Cluster experiment), at ISOLDE (ISOLTRAP, RexTrap) and in Stockholm (SmileTrap). Then he presented strong and weak points as for example the high flexibility on one hand and the long learning time it takes to step into the system on the other hand.

Thomas Johansson from SmileTrap had nothing to add, since SmileTrap is using basically the same system as ISOLTRAP.

Manuel Vogel from the Cluster experiment presented the layout of their system that is a VME-PC based system, which has the same origin as the ISOLTRAP system. In order to simplify their system they want to get rid of their CAMAC crates and the VME bus system. Trying this they face two problems which is the cycle frequency of 10Hz and an older CAMAC controlled Transient Recorder.

Friedhelm Ames reported on the RexTrap system, which is again basically the same as the ISOLTRAP system. Because there are 3 different HV platforms involved in RexTrap and RexEBIS one needs to have for the "nonintelligent" devices also a bus, which can easily be linked via optical links. So they chose in addition to GPIB used for the "intelligent" devices the PROFIBUS (RS485) for the "nonintelligent" devices. The PROFIBUS mastercard will be connected directly to the VME bus.

Hartmut Häffner described the control system for the g-factor measurements at Mainz, which is based on a PC driving a GPIB-Bus and a few DAC and digital I/O channels. The software is based on self-written programs using Turbo Pascal 7.0.

Christian Toader talked about the computer system used to control the Mistral experiment. It is based on a SUN workstation for the graphical user interface and a VXWORKS realtime kernel on a motorola processor driving a VME. He emphasised that they need to have a real time system because they want to measure very short-lived isotopes and they have to react on a trigger from outside (ISOLDE).

None of the experiments within the EXOTRAP and EUROTRAPS networks that are still in the planning phase could report on their control and online evaluation systems because either there was no representative or the experiment is in a too early stage (for some perhaps both).

Gary Rouleau introduced the planned Athena control system. Athena will consist of at least 4 traps, which have to be controlled and a position-sensitive detector with a big burst of data when the event occurs. Thus the control and data acquisition system will be large and complex. They decided to go for LabVIEW as the programming environment for this system. The hardware is made up of PCs for the user interface and VMEs controlled via a PCI-VME interface with the PCs. For data acquisition there is a VME controller with a self-made real time kernel linked to LabVIEW on a PC.

Then he proceeded with a short glance on LabVIEW. During this presentation a lot of small questions came up what LabVIEW is able to do and what not.

Then we had a look onto the small project Gary has created already by using LabVIEW. Basically he controls some devices via GPIB for which he has made a surface. But he also showed that a 2-dim scan is possible to program with LabVIEW. After a look on his famous Paul trap demonstration the first day was finished.

During the second day we started to evaluate weak points to be avoided in the next generation of systems. This sounds like an excerpt of a textbook but sometimes we obviously didn't read these books before working on the current systems.

Weak points to be avoided:

• too complicated system for simple problems, resulting in long learning time and sometimes causing problems changing and adding things 
• exotic hardware and/or difficult programming of it
• LACK of DOCUMENTATION

List of demanded "features":

• DOCUMENTATION (Since we all are often a bit lazy writing documentations this has a very high priority. Even for LabVIEW where people claim that it is rather intuitive one should spend enough time on documentation. The reason is that one will have with an increasing project size an increasing number of sub-VIs. To keep up with their function one has to document each of it extensively and spend some time to create "self-explaining" icons.)
• The user interface and the control of the hardware (which implies the data acquisition) should be divided into at least different programs but favourably onto different computers.
• Some of us cannot do without real-time capabilities (Mistral, ISOLTRAP would like to have it, g-factor) and it is therefore highly desirable to find a LabVIEW-supported real time kernel (there is one project in this direction by National Instruments, but not yet on the market).



Dietrich Beck raised the question if it is possible with LabVIEW to link to a device dynamically without knowing all the possible types at "compiler time", as it is possible in the ISOLTRAP system now. Gary promised to talk to some people about this problem.

Though it is not yet clear if one can reach with a LabVIEW-based program the same flexibility and performance we have for instance in the ISOLTRAP system, we agreed that the system of choice for a new experiment is LabVIEW. The smaller an experiment is the easier is this choice. But we also agreed that the knowledge of C/C++ is essential, because for complicated algorithms one will have to use the LabVIEW interfaces to the C/C++ world.

We found two modules, which will be needed in almost all of our experiments, and which can be the start of our "software collaboration".

1. Vacuum control module. This is an excellent task for LabVIEW; this is the kind of task LabVIEW is designed for. Dirk Peter van der Werf will start programming this within the Athena collaboration in the next weeks, taking care that it will be as modular as possible to ease the application in other projects within our networks.
2. X-Y-Z scan module. Most of us are going to scan frequencies and voltages either during the measurements or in the setup phase of e.g. the ion optics. This module should take as input the devices and functions to scan (as flexible as possible) and the scan range. For instance, one time I would like to scan the frequency of the AFG nr.1, but another time a DAC channel controlling a power supply. The output should be a graphical visualisation of the scan result. Ideally this result is updated after each step to react on trends. Out of this one should be able to deduce (either via mouse pointer or analytically) the optimal values be aware of possible strong noise and multiple maxima.
It was decided that further meetings on a common control system are useful, if not too often. The next meeting will be held during the EUROTRAPS/EXOTRAP meeting in Portugal by the end of October.

Frank Herfurth (ISOLTRAP)

  List of participants in alphabetical order:
 

name	experiment	institute
Ari Jokinen	JYFL	University of Jyväskylä
Celine Monsanglant	Mistral	CSNSM Orsay
Christian Toader	Mistral	CSNSM Orsay
Dave Lunney	Mistral	CSNSM Orsay
Dietrich Beck	weak interaction studies	K.U. Leuven
Dirk Peter van der Werf	Athena	University College London
Frank Herfurth	ISOLTRAP	GSI
Friedhelm Ames	RexTrap	CERN
Gary Rouleaux	Athena	CERN
Hartmut Häffner	g-factor of HCl	University of Mainz
Manuel Vogel	Cluster experiment	University of Mainz
Peter Seibert	g-factor of HCl	University of Mainz
Thomas Johansson	SMILETRAP	MSI-Stockholm