Up: AiC Report to NOT
Previous: Down Time
The last mirror cleaning (in May) was less than satisfactory. It is
clear that a wet wash is required ASAP and realuminisation will be
necessary next summer.
A swedish undergraduate student, Anders Magnussen, joined NOT for two months
to work on this project under Graham Cox's supervision. Graham reports:
The object of the project was to re-design the active primary mirror damping
system for the NOT, in order to remove the significant oscillations measured
at 12 and 14.5Hz. The signals from the load cells of the primary mirror are
filtered, amplified, inverted and fed back using loudspeaker voice coils,
mounted on the edge of the mirror.
An old mirror damping system using analogue filtering exists, but has not
been operational on the telescope since the active optics upgrade. The new
system uses a digital filter and a prototype was implemented on a Texas
Instruments development board, the TMS320VC5402 EVM. The digital filter of
type IIR band pass only passes the desired frequencies of interest and
attenuates the unwanted noise.
Tests have demonstrated that this new filter works satisfactory though further
work is required before the active damping system can be re-installed on the
Ingvar Svardh reports:
The new TCS will function as a server that performs all kinds of
operations with the telescope. The goal for the new TCS (written in c,
the old one in Pascal) is that the source code shall be very easy to
understand for any professional real time programmer. I estimate a few
minutes for most of the solutions, compared to many hours for the old
system. Only simple solutions are used, all well explained. The size
is expected to be a third of the old one with the same
functionality. The switch from the old to the new system is expected
to take place during a few hours in day time by moving about 25
connectors to a new mating contact. All functions will have been fully
tested before this. A few hours night time adjustments will be needed
for final servo and autoguiding adjustments. The I/O is arranged so
that only one program controls an I/O section which speeds up and
simplifies things. It is my intention that the new TCS should be as
reliable as the old one immediately after commissioning.
32% done, finishing date stable at July 2003.
- Master, initializes all data used in the system.
- CPU load calculation.
- Weather. Fully verified with the real weather station.
- Building power.
- Building servo.
- Rotator power.
- Rotator servo.
- Support slow I/O (controls slow hardware).
- AltAz power.
- AltAz servo.
- The calculation of the position using the coarse Stegmann and
fine Inductosyn is done with a much better method, the old one
was extremely difficult to understand.
- Reference position calculation with a few statements instead
of 20 A4 pages as in the old system. It should be noted that the
old system was designed for unreliable communication between
master and subunits.
- Deceleration in advance to avoid overshoots has been made with
a very elegant formula:
vRamp = c * abs(PositionError) where c = 2a/vMax
instead of a separate program as before.
- GPS time client.
All programs have been tested in realistic simulation.
All server functions defined, amounted to 115. The definition file
will be used by the GUI developers.
Small standalone GPS receiver installed for tests, had some faults
but looks promising as a general time server placed in the service
Good progress has been reported by Carlos for the hardware part so
we hope that tests on the telescope systems can be performed soon.
Project on hold.
Occasional unexplained building crashes (without telescope/building
misalignment) suggest that all was still not well at the beginning of
the reporting period. Eventually, they were found not to be building
crashes at all, but arose from anomolous responses from the safety
system, causing immediate power downs that appear like building
crashes. The problem was traced and repaired. There have been no
building crashes since.
Up: AiC Report to NOT
Previous: Down Time
Tim Abbott, AiC