I had a close call Sunday morning with the little rain that we got. I started the system about 10:30 and went to bed, set my alarm for 4:40 am. For some reason I awoke at 4:20 (God looking out for me?) and decided to get up. When I checked the inside computer, I saw the latest images were starless so I peeked outside - clouds. I decided to go ahead and shut the observatory (I usually process what data I have first) and found little droplets of moisture as I walked out. Ten minutes later it was drizzling quite a bit - not enough to help the garden (we got a total of 0.05 inch) but I wouldn't have been happy with that much rain on the telescope, computer and camera.
So I went back to researching cloud detectors. There are several interesting, affordable ideas using an IR detector and Arduino but I kept thinking that a CCD camera makes a pretty good cloud detector if I could just figure out how to read it in more-or-less real time. Every time I slew to a new target I take a 10-second exposure for Elbrus to plate solve and update the telescope. Elbrus writes a status file after every analysis attempt that has a code for success or failure. I already knew how to read this file and I figured if Elbrus couldn't find a solution for whatever reason (clouds, telescope lost, telescope not tracking) that was bad enough to warrant waking me up to check it out.
So I added a line to my script to ship the sync image into the house computer where I had set up Elbrus to respond to new images in that folder. I then wrote a little VBScript that reads the status file every three seconds in an endless loop. When it detects the not-solved code, it sets off an alarm in the computer that I calculate to be strong enough to wake me (or Yvonne). The alarm is kinda cute. I found a simple way to play one of the many Windows sound files (like the shut down melody, etc) and put it in a loop to play it 10 times in succession. I tested it and it certainly works if the inhouse Elbrus can't solve the image - it also ignores the successful solves.