Flat not "flattening?"

[alanrock]

Working on an RGB image of M81 & M82. My green master has some streaks through it, which can be traced back to a smudge on the filter--when I look at the flat, it's clearly there. I used BPP to produce the master light. My question is, why didn't the flat subtract out the streaks?
I've attached the (cropped) master and the master flat. Any ideas?

Upon further inspection, I see that my green filter is actually perfectly clean, so now I'm really mystified as to where these artifacts have come from.

[ngc1535]

It isn't in your flats... it is in your chip.
You are seeing the pattern of the silicon that was chip was cut from.
This often shows up when there are fluctuations in the amplifier or a pre-flash is used.
In my experience, it often isn't apparent in all images and can change through time (which is why it doesn't necessarily calibrate out).
Below is an uncalibrated image that shows the same effect. It does not calibrate out.

[pfile]

you should see this pattern in your darks... check the master dark with a "super-STF".

there is still hope though, if it is in your darks, then you can try calibrating without dark optimization. this assumes then that your dark duration exactly matches the light duration - if not, you need new darks that do match.

when you use RBI preflash this pattern is in every dark and every sub, and generally it will calibrate out, but only if your darks match time-wise (and are collected around the same time as the lights because as adam mentions this pattern can change over time.)

rob

[alanrock]

Thanks guys, that’s incredibly helpful. Not necessarily good news, but helpful! No idea what RBI Preflash is. My darks are a little dated. Do you think I should take new darks, try taking some new subs, or both?

[pfile]

i guess it is worth a try to get some new darks of the same duration of the lights and give it a try without optimization.

RBI means "residual bulk image" and is a problem where some photoelectrons get trapped in the substrate (or bulk) of the imaging chip. it tends to get worse the colder the die temperature. to avoid having actual subexposures getting trapped in the substrate, you 'preflash' the chip with IR light and then read out the bogus image before starting your real sub. because the charge-trapping capability is related to the structure of the substrate, you usually see these circular rings which are a byproduct of growing the silicon ingot from which the wafer (and eventually the die) was made.

rob