Friday, 11 December 2009

Murphy's Law, or how to break stuff

We call it the "top-end" but it's really the secondary mirror with the associated mechanical, electronic and optical components. Unlike secondary mirrors in most amateur and consumer telescopes, the top-end is a rather complicated beast. During night-time observations it "vibrates" at up to 100 Hz (100 times a second) to compensate for distortions in the optical path of light from a star, galaxy or whatever we're observing. The distortions are due to turbulence in our atmosphere. This system, called tip-tilt, can improve our delivered image quality by up to 100% compared to a telescope that doesn't compensate for the atmosphere.

On top of this we also move the secondary mirror to compensate for "flexure". Just about all astronomical instruments that are attached to a telescope suffer flexure. As the telescope slews around the sky the effect of gravity bends the instrument (and telescope) ever so slightly. The amount of bending is minute but it's enough to impact image quality - the detectors in the various instruments are no longer exactly perpendicular to the wavefront (the beam of light that the telescope sends to the instruments) and the result is small aberrations that can be seen in the images of the night sky. These can be corrected in software to a certain amount but always result in a loss of sensitivity, i.e., the faintest objects we detect might just be too noisy to extract a scientific result or the image of a star isn't quite round enough to to see if it's a point source or slightly extended which means it might have some interesting circumstellar material around it - perhaps even a new solar system!

Our top-end is one of my responsibilities and I spent a lot of time coming up with a way to compensate for flexure in our current survey instrument. The secondary is moved in both the translational axis (i.e., from side-to-side) and in the rotational axis (i.e., it tilts) to correct for the flexure. The work resulted in a significant improvement in our delivered image quality, or as some might say, the "seeing" (although that would be a bit of an inaccurate way to describe things).

Unfortunately, the top-end is being asked to do a lot of work and occasionally just seems to give up and stops moving. The tip-tilt still works but the flexure compensation stops (they are controlled by different systems). It's an annoying problem although it's quickly fixed at night so rarely loses much time, but it does often mean we have to stop observing for a couple of minutes to reset things.

At the start of this week one of our engineers played around with a spare controller for the top-end and then put the original back in place. That's always a bit of a pain because it means things get reset during the day and the mechanisms that move the top-end are sensitive to temperature. During the day things are warm and at night it's cold, so any daytime work on the system means we have to reset things at night when the temperatures have settled.

Well, it turned out that swapping the controller and then putting the old one back in place fixed things. For two and-a-half nights we experienced no problems at all with the top-end whereas normally we'd see the problem two or three times each night. I was so happy!

Being confident things had been fixed albeit not sure how, I sent a message through our fault system describing what had happened and that it appeared things had been fixed even if it was serendipitous.

I believe it was within one hour of my happy and relieved response in the fault system that the top-end stopped working again.


Protege said...

I think I need to read this again.
This is somewhat too scientific for me.;)

Andrew Cooper said...

All to true! BTDT!!

I have learned not to make any pronouncements when faced with a problem like this. When I think it is fixed I might make some guarded comment like "try it now, it may be working, but let me know if you have trouble with it". I then let everyone else come to the conclusion that I fixed it when it does not break again. Much safer this way without a direct challenge to the great god Murphy.

Tom said...

Protege - I apologise. I started off trying to keep it relatively simple because I know most people haven't a clue about how pro telescopes work, but probably kept too much jargon in it. Sorry about that!

Andrew - I've BTDT as well! I'll admit to a little poetic license in my post because if I'd really said the problem was fixed I know what hell I'd be letting myself in for!

In my fault response and after detailing and describing the problem and what happened, this is how I ended my response:

"Probably the kiss of death, but I think this is likely significant."

(I was referring to the fact that replacing the controller, and then replacing it again with the original, seemed to have made a difference).

As you can see, I even predicted (prophesied?) another instance of Murphy's Law! I was just stunned it happened so quickly...