Question on Flats

Dave,

This is something I have studied extensively due to my imaging work!

Theoretically, flats should be taken either before or immediately after a filter change so that you will be using the filter in its current position as well as the current focus point. Although filter wheels do a very good job at replicating the positioning of the filter wheel, they are not perfect.

To this end, we have a double issue involving the location of the filter as well as the possibly variable position of the focuser.

Unfortunately you must assume that the slight change in focus which is quite common during the evening will not impact the integrity of your flats. This is a poor assumption if you have large donuts on your subs and where the focuser changes during the evening WILL impact those large donuts. If you have a fairly clean CCD window, then the only real issue is the pixel to pixel sensitivity and which is somewhat independent of focuser position (ex. look at the dome flats taken by professional observatories and where the flat panel is not at infinity so as to match the focus point but a few meters away).

On the issue of target ADU, I always aim for about one-third (ie. 20,000 - 22,000 ADU) and which I have found to produce my best results. I have tried in the past one-half as well as two-thirds of maximum but have since settled on the one-third rule.

Anthony.

ayiomamitis wrote: Theoretically, flats should be taken either before or immediately after a filter change

Wow - that would be very time consuming and presumably, the only way to do it at night would be with artificial illumination. I can't imagine me going out with the lightbox every time I change filter...

I wonder what the pros do?

Dave,

It is my understanding that this what the pros do ... every filter change is accompanied by flats either immediately with the filter change or following the actual exposures for that filter.

It only makes sense since the only way you will confirm that the filter is at precisely the same position as when the subs were taken.

This would suggest the flat box is either moved into the field of view for the flats to be taken or the scope slewed TO the flat box. I think the former makes sense where the dome filter is rotated into view for flats to be taken.

Anthony.

I'd imagine doing that for the twin Kecks would eat into extremely valuable (and expensive) imaging time. Slewing those puppies to the interior of the dome to shoot a flat is not somthing I'd imagine is done in a minute or two...

I wonder what they did with Hubble?

I'd rather imagine that some bright sparks in Hawaii/Chile etc figured out that when taking flats, they would do them in one go (presumably in non-imaging time). They could then do the analyis and figure out how many focal distance they would need to take their flats at in order for them to still be "valid".

For example, in my case, my focuser travels 7000 steps from start to end. Assuming nothing else in the imaging train changes, I know that my focus in the clear filter goes from about 1500 steps to about 3500 steps when I'm shooting light frames (and that would cover all the temperature ranges we have here in Ireland).

I could take a series of images at say 250 step intervals. Then take flats at the same 250 step intervals. And then test the flats by using them with mismatched images - e.g use a 1500 flat on a 2000 image (and also a 1500 flat on a 1500 image etc) and do a subtract. If the resultant image looks ugly, I know my flats needs to be closer to focus than a 250 difference.

I could then figure out the point at which the focus point of my flats has no bearing on the quality of the image. For instance, that might be 100 steps - so I could use a flat of 2000 on images taken at a focus of 1900 to 2100 with no discernable difference on the calibrated image.

You could called it a scaled flat I guess - i.e. good for a defined range of focus positions...

I wonder if the pros do it like that? Or should I copyright the above process in case they haven't thought of it?