Starship Asterisk*

After much discussion, the consensus is that the bad images from the MK CONCAM3 like this:

were caused by the Sun hitting the camera.

Now Alan Holmes, the president of Santa Barbara Instrument Group (SBIG), says that the power to the camera does not even have to be on for Sun damage to a CONCAM3 to occur. He speculates that the camera lost power when the shutter was open, and so the shutter remained open past sunrise the next day when the Sun was then imaged onto the CCD and effectively melted some pixels. CONCAM2s seem to be immune to this.

We therefore need a way to minimize the risk that this will happen again in the future.

After looking over our system, there are three areas that can be changed that can reduce the chance of Sun exposure.

1. The unix script file "clockdaily.script" has a fault mode where the date clock overwrites the hardware clock. This script is called before every CONCAM exposure. This will occur if the Internet connection is down. The data clock will lose time during exposures, so it is the hardware clock that must remain pristine. So clockdaily.script needs to be rewritten.

2. All of the CONCAMs are slaved through the "rdate" command to reset their time to the time of concam.net mother computer. We found this works better than Network Time Protocal (NTP). But this leaves the "single fault mode" of someone (like me) accidentally offsetting the
time on the concam.net mother computer, and then all of the CONCAMs all over the world will follow suit, and could start imaging during the day. Then we would like the CCD chips for ALL of the CONCAM3s. Yuck. So we need a way that the concam.net mother computer checks its own time against some standard and alerts a human if everything is not as expected. This could be a time check to a nearby computer and if the two times are not within 60 seconds of each other, then an email could be sent out to several people.

3. If a good and smart Uninterruptable Power Supply (UPS) is added to the field CONCAMs, this UPS should be able to signal the local CONCAM computer that power has gone out and the system is now running on the UPS's backup power. When the system gets this message, it should begin a graceful shut down, making sure the shutter of the CONCAM is closed until a human can bring the whole system back up. That way, it should not accidentally expose to the Sun when power goes out.

We will work on implementing these for all of the CONCAM stations in the near future.

- RJN

1. How about adding some simple heuristics to check what the difference is between the two clocks. If they're out of sync by more than five minutes, keep the hardware clock time and don't reset it. -ves: "Simple heuristics" are usually a good idea for avoiding disaster, although they tend to increase overall system complexity.

2. What exactly's wrong with NTP? I use it to make sure the clock my my VCR doesn't slip (otherwise it slips by as much as a minute/month) and it works ok for that. A form of the heuristics above are partly built into NTP. Perhaps any clock time change of more than five minutes should be done manually?

3. Don't know much about UPSs. Perhaps a change to the mechanics of the shutter could also help? (dead-man's switch)

Boldra

I assume (for lack of looking) that the same motor-driven wheel is used in the CONCAM3 cameras as SBIG used in older models like the STV. If so, one could feasibly modify the shutter drive for those systems to allow an external circuit (photodiode looking at the outside sunlight + opamps) to run the wheel around to "closed" (& keep it there independent of malfunctioning soft/hardware) off of a small battery, though you'd probably want to make that mod before you built up the system. The wheel would have to be re-initialized it powered back up, but presumably other things would have gone wrong before this anyway.
Not a trivial fix, but not a trivial problem either...

mfw