Our STX-16803 camera has "created" a super hot pixel during the summer break. It shows saturated values even on the bias frame and affects the whole column. It can be fixed with a bad pixel map, but I am bit worried about the photometric measurements if the bad column hits the aperture. Is there anything else that can be done? arto
Sorry to see that. A "super hot pixel" would be somewhere around 56000-65535 depending on the egain/offset of the camera. It's a characteristic of the CCD device structure, and they do happen. Have you got the cooler dialed up to -20 to -35C ? Bias/Dark subtraction and dithering may help in your situation - it's an unfortunate location. You could send over a set of bias, dark, and light frames - eg full frame FITS image binned 1x1, unprocessed for us to take a look. Likely you will have to use Dropbox/WeTransfer/Google Drive due to the size of the images.
It's simply radiation damage, which inevitably accumulates over time. The main culprit is cosmic ray neutrons. The rate of damage accumulation is worse at higher elevations, where you have less protection from the atmosphere. Unfortunately that is the best place to put an observatory. The damage will not compromise your photometric results. Typically the hot column will subtract out cleanly. You will want to avoid putting your target star on the hot pixel if possible. As you might imagine, the Hubble cameras are getting seriously sketchy by now. They have no atmosphere to protect them from cosmic rays, and there is also the South Atlantic Anomaly adding more radiation. You wouldn't believe the amount of work that is going on into calibrating out all the damage artifacts. And yet the cameras still produce excellent science data.
Yes, the column seem to calibrate out quite well just by making new bias frames. I will use Bad Pixel Map to remove the pixel itself. Should the Bad Pixel Map be applied on the master dark calibration frame(s) or does it matter? I enclosed here a subframe of a bias frame if you want to take a look. The camera was cooled to our standard -25C. arto
You only need to apply the Bad Pixel Map to the final result. In fact you can associate a map with your Dark Frame in Set Calibration, and it will apply it automatically as the last step during calibration. I'd try to avoid putting your target star on the hot pixel. I wouldn't worry about the column.
I noticed that the trail of the super hot pixel is different in subframes and we use subframes extensively. So simple bias subtraction does not work and I have enabled Bad Pixel Map for the hot pixel column. Unfortunately the hot pixel is close to the CCD center so avoiding it with automated observing (ACP) is rather difficult or impossible. arto
Bad Pixel Map interpolates across the column, taking the average of the two pixels on either side. It's really a cosmetic tool rather than a science data tool. I really can't recommend using it for photometry. It's probably okay if the background annulus had some repaired pixels in it, but it would be bad if the column crossed through the measurement aperture. I think you would be better served by avoiding it. Probably your odds of landing on it are low anyway, especially with how ACP can use PinPoint-based slew refinement. If you see your target or reference stars landing near the target, you could tweak your target coordinates to avoid the column. Set it up so you never use a subframe that includes the column. Another option would be to not use subframes, so the calibration works properly. I imagine you're either doing that to reduce download time, or to reduce data storage requirements. For the latter, you could try using our Compressed FITS feature. The files aren't compatible with other packages unless uncompressed first (e.g. via Batch Save and Convert), but it will typically reduce storage requirements 2-3X (the algorithm is lossless but was optimized for star images with lots of dark background). Yet another thought - perhaps make matching subframe darks. Time consuming but could be done via script. Unfortunately replacing the sensor is no longer an option. And it would have been a rather expensive option anyway. Ultimately you're going to have to find a way to cope with the problem, because ALL sensors eventually get cosmic ray damage. I have a 2001 vintage 6303 chip that had like 1 warm-ish pixel when it was new, and now has nearly a dozen columns. They all subtract out fortunately.
Yes, I agree with all you said, but the issue with full frame download is mainly the download time (both from the camera and from the remote observatory to own computer). What I usually do is time series and the down time between the images should be as short as possible. Luckily the stars are quite large (or the pixels are small enough) so averaging one column should not be a big problem even if it happens to be on the target star or an reference star. Thank you for your help. arto