Document Sample
downloading Powered By Docstoc
					Techniques for Increasing the Reliability of Accelerator Control System Electronics
                                 (ten years development)

                                    Angel H. Angelov
                     Institute for Nuclear Research and Nuclear Energy
                          Bulgarian Academy of Sciences, Sofia

       An important part of the job of the accelerator physicist or engineer involved in
the accelerator control system project is to estimate the quality of the used electronics.
The estimation of the software service of this electronics has a similar importance.
       Ten years ago Jeff Utterback from FNAL reported on the gained experience as
regards the reliability of the accelerator control system components (PAC’1993). It seems
useful to make a review on the same matter today - ten years later. We did it as a
comparison between the problems discussed ten years ago and the nowadays system
design, the system components and the appropriate parameters. Some disputable ideas,
that have emerged as a result, are reported in the light of the modern tendencies in the
automated control.

       The history of the automated accelerator control systems (ACS) started about 40
years ago, when the computers became relatively reliable. At the very beginning the
accelerator control had an important role for the development of the computing
technologies. The sophisticated tasks of the on-line accelerator commissioning needed
high computing power and a reliable electronics performance. Today the computers are
far ahead in their development thus offering new possibilities to the ACS designers. They
can be applied in new accelerator projects, as well as to upgrade old ACS. The latter
becomes more and more actual because the major part of the accelerator equipment has
much longer life than the ACS's one. It seems normal to ask: “Why should we upgrade
something which is doing its job?" Except for the above-mentioned reason and the
inevitable pressure of progress, there exists the human factor. The operator on the
accelerator control console is usually a person, who is very close to the modern IT tools
and he is eager to complete his job as well as possible. Of course, the funding is decisive
for the degree and the velocity of the upgrading process. We shall discuss this moment
wherever reasonable in our remarks on the use of new components. Anyway, both cases -
a new ACS project and an ACS upgrade project require to improve the reliability of the
ASC components as much as possible.
       Nowadays many companies offer a full or a partial design and elaboration of the
ACS. A comparison between the experience with the industrial and “home-made”
systems was reported recently [1]. It seems now too early to make decisive statements
about the priority of one of these two types of the ACS design, but anyway the ACS
designer must be able to estimate the quality of the components on the basis of a good
understanding of their electronics. This is much more important for the so-called
“industrial” accelerators and of course for the accelerators used in the Accelerator Driven
Systems (ADS) where the data security needs a special attention.
       The ACSs, commercial or ‘home-made’, are an object of precise estimation and
all possible elements, leading to the their increased reliability are in the scope of the
designer’s attention. Ten years ago J. Utterback published his report on the techniques for
increasing the reliability of ACS electronics [2]. The fast development of the
technologies makes it useful to review the same matter today - ten years later. We did it
as a comparison between the problems discussed ten years ago and the nowadays system
design, the system components and the appropriate parameters. Some disputable ideas,
that have emerged as a result, are reported in the light of the modern tendencies in the
automated control and the new moments in the accelerator security.

       We can summarize the main problems with electronics, reported in the J.
Utterback’s paper [2] in two groups:
       1. “Do not use”: chip sockets, polarized capacitors, mechanical calibration
components and new components.
       2. “Recommended to use”: built-in diagnostic aids, protected I/O and new
        Many of these “old” problems can be resolved by using modern techniques and
components, produced by new technologies. Of course, all Utterback’s prescriptions are
valuable now, too.
       A. Chip sockets.
       The corrosion of the chip sockets is a factor, very often decreasing the reliability
of the electronics [2]. It seems valuable to look deeper at the scope of influence of the
corrosion effect. This effect has behavior, depending on the ambiance and the
commissioning conditions. The very high quality connectors and electrical contacts can
be damaged by the small pollution in the atmosphere or by the current load parameters,
etc. Sometimes the corrosion makes the connector unusable. If the ‘old standard’
connector is no more available it could be very expensive to replace the cable because in
order to use a new type connector, we must replace the module’s connector, too. (The use
of the complex ‘male/female’ of two different manufacturers can increase the corrosion
probability if the contact surfaces are covered by inappropriate metals).
The full elimination of this effect can be done by the implementation of the wireless
connections. Sometimes the use of this technology can be cheaper than a cable
connection. J.Gandowitz in his article ‘Adding Bluetooth connectivity to existing system
and platforms’ [3] wrote about the advantages from the removal of the wired serial
connections and their replacement by the Bluetooth [4] wireless connections. There exists
an Internet Protocol Web server running on a Bluetooth chip [5]. This is valuable for the
temporary connections or for the connections with long periods of inactivity such as
transfer of sensor data, system status messages or downloading of software and data.
Except for the PAN (Personal Area Networking) Bluetooth applications, the wireless
networking can be made by using LAN and WAN wireless communication standard
equipment, i.e. GPRS (General Packed-switched Radio System) [6].
B. Polarized capacitors
The use of the polarized capacitors is most dangerous in the power supply units [2]. The
quality of the modern IC converters with frequency up to 750 KHz offers the possibility
to avoid the use of these capacitors. In case the use of the high capacitance is inevitable,
one can use Multi-layer Ceramic Chip Capacitors (MLCCs). They are better than
tantalum and have a better electrical performance over a wider frequency range. Even ten
times rated voltage will not destroy a MLCC capacitor.
C. Calibration components
The calibration has two aspects that create unreliability. The first is the component by
which we perform it and the second is the calibration procedure [2]. While the
mechanically variable potentiometer can now be replaced by a Digitally Programmable
Potentiometer/Resistor DPP/DPR to make the procedure of calibration reliable, it should
be automatic. The electronic components market offers a good resource for the
implementation of the automatic calibration where necessary.
D. Built-in Diagnostic Aids
The positioning of the LED local diagnostic indicators does not eliminate fully the
difficulties of the maintenance [2]. We may position in parallel to the LEDs the Bluetooth
local channels that will send independently of the other channels remote information
about the modules’ status from the most crucial positions.
E. Protected I/O
In case of the wireless connection there is no need of electrical I/O protection.
Nevertheless the new I/O protection components that have been offered are of very good
quality, e.g. low capacitance (5 pF input), 1 nA supply current and up to 15 KV
electrostatic discharge protection.
F. New components implementation problem
One can see that this category has a place in both groups, distinguished at the beginning
of the paragraph. This means that it is still necessary to use the new electronics
components with caution. In comparison with the ‘old’ situation the use of a new
component is now better supported by the electronic information and the increased
possibilities of the e-commerce, thus, the problem with the “Vapor Component” is not so
           The Software problems, discussed by Utterback [2] are in connection with the
Programming style of the MCU on the electronics boards. He recommends to USE a
selftest, a sign of life (for example ‘heartbeat’ or ‘watchdog’), a full service of all
interrupt vectors and the diagnostic aid used at Fermilab called Age (the time elapsed
since the last reset). All these recommendations are of current importance.
       With the era of the wireless connections and the Internet, the new software
problems appeared. The Internet networking is often used in the ACS. During the second
half of the last decade of the previous century the size of the Internet doubled every six
months. The addresses of the old version IPv4 (about 4 billion) are going to run out in the
next years. We must put the stress on the fact that the new IP (Ipv6) has an enormous size
of addresses - 2128 (roughly 6.67x 1023 per square meter). Another fact is that IPv6 will
become a requirement very soon for the U.S. companies. Now on the MCU market it is
possible to buy IPv6 networking with the resident silicon software. All this must direct
the attention of the ACS designer to the implementation of this type of system
       The interrupt service will not be a problem if a qualitative RTOS is implemented.
More than 100 vendors offer the RTOSs for MCUs. The prices vary from some hundreds
to about USD 50,000. Many of them include a technical support and a training course [7].
       Nowadays, the tracking, recommended in [2] can be more intensive and more
detailed. It seems not a bad idea to make a pool (a forum) for the users’ community on
the industrial modules (as EPICS team does [9]). The test station [2] is not needed if
wireless networking is used.

       The problem with the wireless PAN and LAN networking on the first place
appears in the EMI compatibility. The Wi-Fi gives a possibility for 3 channels with
‘sin(x)/x’ shape to be used, while the Bluetooth has 79 channels (1 MHz each one), both
at 2.4 GHz band and the simultaneous use must be implemented very carefully. The
WAN has another opportunity, i.e. GPRS (General Packed-switched Radio System).
       The second problem with the wireless networking is the data security. To avoid
the bad-will intervention different techniques as fast SRAM [8] can be useful.
       The staff problem is always present when a new technique is in use.
       The most prominent conclusion is that the implementation of the recent
technology achievements makes the ten-year old problems very often obsolete. New
technologies generate new structures with a new class of problems. The electronics, due
to the improved quality of components and the almost full digitalization has a satisfactory
level of reliability. The today’s problems of the ACS have been moved to the sphere of
the Communication Reliability and the Data Security.

1. M. Pieck, K. Kasemir, ”Comparison between an in-house vs. a
commercial control system for beam line control”, PAC2001, Chicago, 2001.
2. J. Utterback, ”Techniques for increasing the reliability of accelerator control system
electronics”, PAC1993, Washington, D.C., May, 1993.
3. J.Gandowitz, “Adding Bluetooth connectivity to existing system and platforms”,
Embedded Systems, June 2003, p.12.
5. L. D’Arcy, “Benefiting from IP over Bluetooth”, Embedded Systems, June 2003, p.14.
6. N. Robins, “ Integrating wireless communications”, Embedded Systems,
February 2003, p. 33-34.
7. M. Barr, “Choosing an RTOS”, Embedded Systems, February 2003, p. 40-41.
8. K. Self, “Fast SRAM and MCUs improve data protection”, Embedded Systems, April

Shared By:
wang nianwu wang nianwu http://
About wangnianwu