Renewal of the 3.5m telescope control system
Karl Zimmermann, Rainer Wolf, Josef Fried
Max Planck Institut für Astronomie, Heidelberg
Reason for renewal
The 3.5 m telescope went into operation in the early 1980's.
While the mechanics is still
working nicely, many electronics and computer parts are at the end
of their lifetimes. Changes in technology have made replacement of parts
increasingly difficult in the last years: many parts - neither original
nor equivalent ones - are simply not available anymore. In order to avoid
losses of observing time due to technical failures it was decided to renew the
telescope control system completely, i.e. to replace computer,
bus system and practically all electronics hardware.
Concept of new control system
The original control system consisted of one central
LSI 11 computer which was connected to the various drives and control units
with an extended DMA bus system.
The new concept
of the telescope control system, devised by R.Wolf, is drastically different.
It uses a workstation from SUN Micro-systems, Inc with standard
Solaris operation system as a host for 5 VME computers which control
the drives. The workstation also serves as a router to the public Calar Alto network. A
linux PC is used for the communication with the operator
through a graphical user interface. All computers communicate with each other by a
private ethernet network, not accessible directly from the outside.
The VME's are located at different places close to the corresponding telescope electronics.
These are: main electronics rack (second floor), operation desk (control
room), yoke mounting
(yoke structure), tube (inside of the telescope tube),and coudé mirror S5 rack
(floor of the coude plant). The linux PC is located in the main desk.
A second communication between VME and SUN is established by Nport
servers (bidirectional protocol converters between ethernet and
the RS232 console port of the VME's) which is useful for software
development and debugging.
Additionally each VME has a real time clock interface board (RTC)
fed from a central GPS clock. The vme system installed in the
control desk is shown
here.
The cycletime of the IO-system has been set to 20 Hz for a sufficient
realtime response to changes and events triggered from hardware.
The software is organized in several layers. The
first layer is the interface to the observer/staff (GUI), which
allows the selection of 5 operation modes:
- IDLE this is the state after restart of the software.
The only allowed command is the selection of an new operation mode
- OPERATOR the only allowed mode for observers to operate the
telescope.
- SYSTEM CHANGE the staff uses this mode for changing the
configuration of the telescope in the automatic-mode (cass. focus, prim. focus
I, prim. focus II, IR-system)
- BALANCE allows the balanceing of the telescope manually or in
automatic mode
- MAINTAINANCE in this operation mode the staff is able to change the telescope
configuration in manual mode. This mode also allows the use of
functions with the instrument protection (see below) disabled
The second layer controls the telescope drives which are right ascension,
declination, focus, cassegrain flange, coude S5 mirror azimuth and elevation
and dome setting.
This task computes 4 times per second all necessary astronomical and internal
used data derived from the drivers input like encoder values, universal time,
airmass, updates the pointing correction, the refraction etc.
The third layer acts as instrument protection. It controls limit
switches, stop positions and current state of the drives. It also builds the
interface to the drivers, which are found in the layer below the
instrument protection. A watchdog monitors all five vme-systems and the tasks running
on them and programs running on the SUN by executing
a live check every 4 seconds. If all systems respond within the
expected time, the watchdog is resetted and repeats the system check 4 seconds
later. If not, the electronics of the drives and system power of
the vme's is shut down. The system is shut down also in case the watchdog
process itself crashes.
A basic part of the software is the telescope database which is built
from the input data of all 5 vme-systems and which contains values computed
from the input data by the drives control task.
As mentioned earlier this telescope database is availabe in each of the
5 vme-systems and the SUN-workstation. The update rate is 20 Hz.
The most important data of the telescope database are mirrored to an
epics database. This database is used as interface to the
observers GUI and also as interface to the instrumentation
software. The epics interface enables the observing instrument to
interact with the telecope, in detail: move
the telecope to new positions, execute
small offsets, move the telescope focus absolute or relative, turn
the cassegrain adapter, read telescope data like hour angle, delta, universal
and sideral time,
airmass, focus positions, cassegrain adapter angle.
At the moment the telescope is controlled by a Graphical User Interface
similar to the one which was in use before renewal of the telescope control
system.
This GUI (see Fig. 3) offers all the functions which an observer needs to
control the telescope. The observer can
select an operation mode, turn the main drives on/off, enter coordinates
for new positions and move the telescope to the new destination, move the
focus, control the telescope tracking, control the dome tracking
(manually or automatically), control the mirror covers and light baffle and
display all neccessary data.
Staff and maintainance operations
Operations which are reserved for the staff are:
- change the telescope configuration manually
- balance the telescope manually or automatically
A lot of functions, which have been built into the new version of the
GUI, have currently to be executed by entering a scriptname,
e.g. balanceStart. These functions will finally be incorporated in the GUI.
Realization and current status
Since the telescope control system is very complex, we decided to
split its renewal into 2 phases. Phase I comprises replacement of the
computer and bus system, phase II replacment of the motor controls
including the closed loop control of the drives in hour and delta.
The problems of phase I are mainly to establish a stable operation
of the computer system and furthermore to supply all functionalities
of the system.
After careful preparation, installation of hard and software for phase I
was done in April and May 2004. It became clear very quickly that
the concept is a sound one, the losses in observing time due to problems with
the telescope control system were on the order of less than an hour per night
in the beginning. Logging of telescope actions and analyzing errors and
failures quickly led to enormous improvements.
As of December 2004, phase I is now in a stable state,
all functionalities required for astronomical observations are working.
To terminate phase I, we need to incorpate read out of temperatures at several
positions in the telescope (these
are needed for temperature compensation of focus) and pressures in the
hydraulics. The GUI is preliminary and will be replaced by
one which includes observational as well as system service
functionalities. Change of the front ring, which is currently done
manually, also has to be automized. We intend to finish this early 2005.
Further items are improved diagnostics and error recovery, new observer
interface to telescope software, independent of epics.
A major task is instruction/education of the staff
on Calar Alto and especially documentation. One of us (K.Z.) will spend some time to
introduce the Calar Alto personnel to the new system. It is obvious that
providing a good documentation means a major effort, but will
shorten telescope down time and so pay off finally.
Some documentation of the new system already exists, and we
intend to finish documentation with the help of Calar Alto staff next year.
Phase II comprises replacement of the motor control electronics for the
drives of the Cassegrain flange, the focus and hour and declination. Usage
of modern digital output stages should lead to relatively straight forward
solutions for the drives of cassegrain and focus. Problematic are the hour and
declination drives which each have a closed loop control to ensure tracking
capabilities. Each of these control systems actually consists of
3 nested loops. It is obvious that such systems are prone to unwanted
oscillations and require careful finetuning. We are currently analyzing
where and how to replace old analogue electronics with modern digital
units without changing the logic and structure of the loop system.
Karl Zimmermann
Josef Fried
December 2004