[LEAPSECS] UTC and Telescope Control.

[Peter Bunclark]

On Thu, 13 Nov 2008, M. Warner Losh wrote:


> methods used today.  Already +/- 1s isn't anywhere near good enough to

> aim larger telescopes with the precision necessary to do good science.


I've been worrying for some time that we have been overstating the
telescope problem.

Prior to 1990 (very loosly) telescopes were built on an equatorial axis,
in which the polar axis is aligned parallel to earth's axis. Notionally,
they just need to be driven at a constant sidereal rate to track the
stars. After that, large telescopes were built on alt-azimuth axes,
basically like an overgrown theodolite, because computer technology made
the complex real-time axis control possible, and the mount was more
compact, less flexible, required smaller building == much cheaper.

Alt-az telescopes can't use UTC as an approximation to earth-rotation
angle, for .1 arcsec tracking you need hand-wavingley 1/150 second clock
accuracy.  So they always have to apply DUT1 (no matter how big it gets).

Equatorial telescopes get off a little lighter, because a clock error
amounts to a simple offset in the RA axis. Such telescopes also, because
of their vintage, often have axis encoders which don't initialise to a
precise zero point; so part of the nightly startup procedure tends to be
to acquire an unmistakeable bright star, centre it on the TV, and tell the
TCS to reset the zero point.  This removes both encoder and clock error.

The only related issue I've been able to find is at least one
user-interfaces which traps the operator from entering a DUT1 > 0.9s. So
there would be a cost in fixing telescope software; I think we should be
careful not to overstate it.

Pete.