Here is a description of my process to acquire reference star to compute instrumental response with a Lhires and a grating of 2400tr/mm. I have seen some people taking reference star spectrum after each object, looking for a reference star at same elevation. This is the recommended procedure for low resolution spectrograph, like an Alpy, but for a Lihres with a high resolution grating this is overkilling.
The wavelength domain of my lhires+ grating of 24000lp/mm with an Atik 460 is around 150 angstrom centered on H-alpha at 6563
I have a set of reference star depending of the period in the year: Altair for summer, castor for winter and regulus for spring. I can occasionnally use Vega.
As I image always the same star I have history I can refer too and check if the spectrum is correctly focused, with the expected flux count.
I acquire the reference star at the beginning of each night, and every nigth. Even if the setup was not dismounted. It is a kind of quality control, a routine check like a pre-flight inspection.
At this period of time, Regulus is my reference star. I acquire 5x120s exposures. I process them with ISIS, my atik 460EX is in binx2. I made a test of acquiring another set of exposures later in the night. Here is the ratio.
- regulus same night
- _regulus_20190322_831.png (4.88 KiB) Viewed 7630 times
I looked back at the spectrum of regulus over the last few weeks.
- regulus2
- _regulus_20190322_908.png (5.49 KiB) Viewed 7630 times
- regulus3
- _regulus_20190324_815.png (6.79 KiB) Viewed 7630 times
- yes, it is important to make a reference star sprectrum every night, good practice is to make it at the beginning of the session as an inspection of the setup before "flying trhough the night"
- no, it is not necessary to repeat acquiring reference star after each object
Obvioulsy this is valid for a high resolution spectrograph, and if you have some doubts, I encourage you to make the test and compare the spectrum of these reference stars at different time in the night and nigth after nigth. Do not extract the response curve, just divide one spectrum by another one and look at the ratio.
Hope this helps and answers some legitimate questions
Valerie