Last month I found a new telescope (200 F/5 Newtonian), in order to benefit from my ALPY600 spectrograph (very restricted on my usual 305 mm Dobsonian), and I had a good night on April 14th so I decided to observe some stars to calibrate it. I was still using the Star Analyzer 100. The reference stars where Lam UMa, 10 UMa, Bet LMi and Psi UMa, so spectral type from A to K. During the session I observed three variable stars on the region: RS CNc again, mu UMa and Y CVn (La Superba).
I decided to perform my usual synthetic photometry method, but the results where... not bad, but not quite close to the tracks or the expected values. In the meantime, as always I did experiment the flux method (as described by Christian Buil) and to my suprise, the results looked much better... Having four different stars as references, I decided to take the time to make a careful analysis, including calculating the color terms of my system. Only the calculation of the extinction coefficient is missing (but all stars were observed at 1 airmass). Certainly on next time.
I present here the results obtained on the red giant mu UMa - Tania Australis, who is classified as NSV (from V 2,9 to 3,3. However in the AAVSO data it is not variable?)
The sensitivity curve has been built from two stars, Lam UMa (A1IV) for the blue part and Psi UMa (K1III) for the red part - in order to have the best quality at both end. I do not have the real spectra of the stars so I built the standard with the Pickles spectra. If we talk about normal stars not far away (to avoid reddening), it must work correctly.
The expected value for the Johnson-Cousin system are BV 4,64 / 3,05 and RI 2,15/1,28 (the R is calcultated with an equation starting from the Rj band, and the I comes from the Hipparcos V-I index)
Here is the graph. I find V for that night to be 3,02, and then B 4,57, R 2,12, I 1,20. Considering that the star looks to have been slightly brighter than average on that night, the values are coherent (B probably too bright).
Then I decided to try Sloan magnitudes. This is the fun part of synthetic photometry: you can use whatever filter you want, re-calculate old data, even modify the passband of the filters if needed. Almost like if you had the star at home in a laboratory!
The reference values are taken from a work by Anthony Mallama (Sloan magnitudes for the brighter stars) :g' 3,83 r' 2,45 i' 1,81. I find respectively 3,87 / 2,42 / 1,75. So again very close, and sligthly brighter for r' and i'. However, g' is fainter and this clearly not in coherence. Graphic
I have upload the values to WebObs. On a next time I will try to use the Alpy and its photometric slit!
To me this is fascinating work, and I am surprised that there has been no comment. Note that my experience is in photometry, not spectroscopy.
The procedures described seem to me not unlike the spectrophotometric methodology employed by a group of Spanish astrophysicists who have published a new photometric system for CMOS RGB cameras, which Brian Skiff alerted the Photometry Forum to on 1st April. Jaime Zamorano, the second author of the paper, has contributed discussion and links on the system to that Forum.
What particularly fascinates me about Christophe's post, if I understand him correctly, is the ability to derive synthetic magnitudes in various photometric systems from a spectrum obtained with a Star Analyzer. It looks like a lot of work, and presumably reasonably accurate photometry would be limited to bright stars with modest amateur equipment. It would be interesting to know how many amateur astronomers concentrate on this type of observation.