Thursday, 27 June 2013

An Example of the Usual Maintainence Insanity

It's hard to get across just how hard it is to maintain the BRT and keep it running smoothly. The system is so immensely complex that when something starts acting funny, its almost impossible to work out what's gone wrong. I'd like to share what we did today to try and diagnose a big problem with the telescope.

The most pressing problem we've had recently has been the fact that on some nights, the telescope will point half a field/image away from where it is supposed to be pointing in the dec axis.

Our initial thought was that, since it is something that comes one night, stays the night, and fixes itself the next, it could be something to do with homing. The mount has within it a small optical sensor against its gears. It rotates the gears until a protrusion blocks the sensor and it then knows where it is. All later movements are done relative to this calibration. The mount never actually knows where exactly it is pointing as such except that it is so far in ra and dec from where it started when it last homed.

Dan assumed this problem might be caused by grease, which might move into a position during some days so that it would block the optical sensor somehow that night, melting again the next day. So, the mount was stripped down and the homing sensor meticulously cleaned. This took the best part of two days.

The pointing problem continued.

There are a number of other things that could cause a declination error. It could be some flop in one of the telescopes two mirrors, some movement of the camera or the various things between the telescope and the camera. Even the telescope itself could be moving. It only takes 1/6 of a degree of movement to get the image from the centre to the edge of the frame. We spent an hour or two trying to visualise which if any of these could be causing the error we were seeing, by all pretending to be equatorial mounts with our arms and spinning round a lot (spacial intelligence is harder at 75% of normal Oxygen levels). We looked ridiculous but hopefully nobody was watching the webcams.

We eventually gave up and decided to retest the homing idea, to see if that really was the cause. It's difficult to see 1/6 of a degree in error on a 30cm diameter wheel. We needed to amplify the angle. We attached a laser pointer to the telescope with gaffer tape, which would shine onto the wall beneath the dome. This effectively gave us a 4m diameter circle. A small change in angle became a large and visible change in the position of the dot on the wall. The laser pointer was just a cheap one built into the back of a pen. It needed some zip ties to hold down the on button.

Then the batteries in the laser pointer ran out.

We didn't have any more button cells. The nearest shop is over an hour away and at this point it was already 6pm.

So Dan cobbled together some bits of bolts and washers and wires and a voltage regulator and within not long at all, we had a mains powered laser. Dan had padded a bolt with washers to exactly the diameter of a button cell, which fit nicely into the battery compartment of the laser, which was then soldered to some wires and a power converter. And only very rarely did it give you an electric shock. We plugged this into the galaxy camera mains plug so that we could turn the laser on and off, by telling the system to turn galaxy camera on and off and re-stuck it to the telescope.

We re-homed the telescope a few times, slewing away in between. Each time, we drew a dot on the wall with a board marker to mark where it had homed to.

It quickly became obvious that the pen was not stuck down well enough as it was moving erratically by several degrees each time we re-homed, which would not be mount error. We toyed with several other methods of attaching a circular laser pen to a circular telescope.

Eventually, we found a 90 degree crevice in the mount that the laser pointer fit into snuggly. 15 minutes of gaffer taping and later and the laser was once again attached. We got a nice blank piece of paper and stuck it to the wall. After a while we found a biro that actually worked for more than a single line and we were ready to go.

This time, the laser stayed in place. We re-homed the mount over and over. It was always homing to one of two points, slightly apart in the dec axis by about the amount we were expecting. It looked like we had found our error. Since the wall was not perpendicular to the laser beam, it took some more faffing to accurately work out what angle 6mm of dot movement actually equated to on the mount so that we could write a detailed email to Software Bisque asking for advice (as we were still no closer to knowing why it wasn't homing right).

By the time we actually managed to leave the mountain the sun had set and by the time we were back to sea level almost everywhere that did food was shut (and we hadn't eaten since breakfast). We found a Chinese restaurant which had already turned off the kitchen for the night but was kind enough to turn the cookers back on to make us a meal.

Just another normal day in the running of a fully robotic telescope 8000ft up a mountain.

The wires disappearing into the gaffer tape is the
precision instrument use in this experiment
Precision Measurements
When working with lasers, protective eye-gear is vital.
We care a lot about health an safety here at the BRT