Probleme RMS avec Alpy600

[Robin Leadbeater]

Hi Peter,

It is normal to need adjust the pixel size from the actual value in ISIS. The effective pixel size depends on manufacturing variables in the spectrograph. (The exact l/mm and exact focal length of the lens.) Shelyak and Christian Buil recommend this.

Yes you sometimes see differences in the ratio of Ne/Ar. ISIS will still work with this provided the lamp spectrum is well exposed (for example averaging several short exposures to give good SNR without saturation. What Etienne is seeing is different. There are extra lines in the spectrum. No software can work if these lines overlap as they have here. You would need to build in a new set of known lines for calibration using the new lamp. (This could be done in ISIS using the "file mode" calibration but the "ALPY predefined mode" needs a known lamp with known lines .) I suspect there has been a change in the gas formula in the RELCO starters that Shelyak use. This is a problem for Shelyak.

Robin

[Robin Leadbeater]

Also would be nice to increase the 4 order poly and the ability to specify even more lines.

I do not agree that a higher order fit is needed. An order 4 (or even an order 3) gives a very good fit with the ALPY provided the lines are reliable and cover the full calibration range of interest. With a high order fit you get larger errors as the fit bends between the lines and is completely unreliable even a short distance above and below the highest and lowest lines. (The fit looks better but it is in fact worse).
The biggest problem with ALPY wavelength calibration is in the blue beyond where the RELCO lamp has reliable lines (many lines in the RELCO lamp look good but they are blends at low resolution so cannot be used.) I do not have list of lines that ISIS looks for but I suspect Christian has selected all the reliable lines that can be used for automatic calibration (ie all lines which are from known elements, are not blends and are well enough separated for the program to automatically detect their position). This is why the entered pixel size has to be tuned to prevent the program locking onto the incorrect lines. How ISIS does this is described on this page
http://www.astrosurf.com/buil/isis/new/release.html
V5.4.0

Cheers
Robin

[Robin Leadbeater]

Peter,

The procedure for calculating the effective pixel size is described here
http://www.astrosurf.com/buil/isis/guid ... ration.htm
(Method 1)
You must still check your calibration and the individual line errors though to make sure the calibration is correct. In this case a slightly different value from the calculated 6.246 was needed (6.3) because of problems ISIS had identifying the correct lines. If you rerun the calibration slowly increasing the pixel size up from 6.25 it is obvious when the fit is correct because all line errors should fall to less that 1A (one line does not for the reason I have already mentioned.) This calibration is stable over a small range of pixel size where the program correctly locks onto the lines. As the pixel size is increased, lock is lost on some lines and the calibration deteriorates again. Any incorrectly identified lines have obviously large errors. It is important to always check this. The ISIS automatic system is not 100% reliable.

Robin

[etienne bertrand]

Helo Peter,
With Alpy and ISIS you must change the size of pixel ; pixel size = (3123 / (Ha pixel - Hb pixel)
(voir le tutoriel de Christian Buil).

[etienne bertrand]

Bon, je reviens aux nouvelles,
J'ai démonter l'Alpy, j'ai re-régler la mise au point sur le spectre du soleil, je penses que la mise au point que j'avais faites n'était pas optimum, là j'ai obtenue une mise au point plus fine et résultat j'obtiens une R=525 et des RMS acceptables 0.13 / 0.15

Dispersion primaire :
Coefficient a4 : -2.716891E-13
Coefficient a3 : 7.276048E-09
Coefficient a2 : -6.693474E-05
Coefficient a1 : 0.54818
Coefficient a0 : -993.520
--------------------------------------------------------------------------
RMS : 0.128035 (en pixels)
--------------------------------------------------------------------------
Ecart d'ajustement en longueur d'onde sur les points primaires
point #1 Lambda = 3770.630 px = 456.957 dx = -0.093
point #2 Lambda = 3797.300 px = 464.832 dx = 0.216
point #3 Lambda = 3835.390 px = 476.061 dx = -0.110
point #4 Lambda = 3889.050 px = 491.845 dx = -0.019
point #5 Lambda = 3970.080 px = 515.607 dx = 0.060
point #6 Lambda = 4101.750 px = 554.057 dx = -0.069
point #7 Lambda = 4340.480 px = 623.368 dx = -0.023
point #8 Lambda = 4861.340 px = 773.707 dx = 0.055
point #9 Lambda = 6562.850 px = 1273.857 dx = -0.098
point #10 Lambda = 6869.000 px = 1367.067 dx = 0.094
point #11 Lambda = 7605.000 px = 1595.671 dx = -0.013
--------------------------------------------------------------------------
Equation de dispersion inverse
Coefficient a4 : 7.593870E-11
Coefficient a3 : -3.984758E-07
Coefficient a2 : 4.818265E-04
Coefficient a1 : 3.26645
Coefficient a0 : 2855.872
--------------------------------------------------------------------------
RMS : 0.149279 (en angstroms)
--------------------------------------------------------------------------
Ecart d'ajustement en longueur d'onde
point #1 x = 322.289 lambda = 3946.141 dlambda = -0.041
point #2 x = 383.499 lambda = 4158.583 dlambda = 0.007
point #3 x = 484.574 lambda = 4510.695 dlambda = 0.035
point #4 x = 494.335 lambda = 4544.734 dlambda = 0.316
point #5 x = 526.856 lambda = 4658.141 dlambda = -0.241
point #6 x = 557.487 lambda = 4764.917 dlambda = -0.047
point #7 x = 615.003 lambda = 4965.164 dlambda = -0.084
point #8 x = 740.684 lambda = 5400.551 dlambda = 0.009
point #9 x = 872.632 lambda = 5852.431 dlambda = 0.059
point #10 x = 995.469 lambda = 6266.479 dlambda = 0.011
point #11 x = 1067.736 lambda = 6506.531 dlambda = -0.001
point #12 x = 1265.097 lambda = 7147.100 dlambda = -0.060
point #13 x = 1339.935 lambda = 7383.942 dlambda = 0.038
--------------------------------------------------------------------------
Coefficient a4 : 7.593870E-11
Coefficient a3 : -3.984758E-07
Coefficient a2 : 4.818265E-04
Coefficient a1 : 3.26645
Coefficient a0 : 2855.872
--------------------------------------------------------------------------
RMS : 0.149279

Fichiers intermédiaires effacés
----------------------------------------------------------------------------
Date début de prise de vue : 17/02/2017 15:06:09
Durée : 48.0 secondes
Date de milieu de prise de vue : 17.630/02/2017
Jour Julien du milieu de prise de vue : 2457802.1295
Pouvoir de résolution : 525.5
Ok.


A tester sur le CIEL.
Merci pour votre aide.

[etienne bertrand]

Bon, je reviens aux nouvelles,
J'ai démonter l'Alpy, j'ai re-régler la mise au point sur le spectre du soleil, je penses que la mise au point que j'avais faites n'était pas optimum, là j'ai obtenue une mise au point plus fine et résultat j'obtiens une R=525 et des RMS acceptables 0.13 / 0.15

Dispersion primaire :
Coefficient a4 : -2.716891E-13
Coefficient a3 : 7.276048E-09
Coefficient a2 : -6.693474E-05
Coefficient a1 : 0.54818
Coefficient a0 : -993.520
--------------------------------------------------------------------------
RMS : 0.128035 (en pixels)
--------------------------------------------------------------------------
Ecart d'ajustement en longueur d'onde sur les points primaires
point #1 Lambda = 3770.630 px = 456.957 dx = -0.093
point #2 Lambda = 3797.300 px = 464.832 dx = 0.216
point #3 Lambda = 3835.390 px = 476.061 dx = -0.110
point #4 Lambda = 3889.050 px = 491.845 dx = -0.019
point #5 Lambda = 3970.080 px = 515.607 dx = 0.060
point #6 Lambda = 4101.750 px = 554.057 dx = -0.069
point #7 Lambda = 4340.480 px = 623.368 dx = -0.023
point #8 Lambda = 4861.340 px = 773.707 dx = 0.055
point #9 Lambda = 6562.850 px = 1273.857 dx = -0.098
point #10 Lambda = 6869.000 px = 1367.067 dx = 0.094
point #11 Lambda = 7605.000 px = 1595.671 dx = -0.013
--------------------------------------------------------------------------
Equation de dispersion inverse
Coefficient a4 : 7.593870E-11
Coefficient a3 : -3.984758E-07
Coefficient a2 : 4.818265E-04
Coefficient a1 : 3.26645
Coefficient a0 : 2855.872
--------------------------------------------------------------------------
RMS : 0.149279 (en angstroms)
--------------------------------------------------------------------------
Ecart d'ajustement en longueur d'onde
point #1 x = 322.289 lambda = 3946.141 dlambda = -0.041
point #2 x = 383.499 lambda = 4158.583 dlambda = 0.007
point #3 x = 484.574 lambda = 4510.695 dlambda = 0.035
point #4 x = 494.335 lambda = 4544.734 dlambda = 0.316
point #5 x = 526.856 lambda = 4658.141 dlambda = -0.241
point #6 x = 557.487 lambda = 4764.917 dlambda = -0.047
point #7 x = 615.003 lambda = 4965.164 dlambda = -0.084
point #8 x = 740.684 lambda = 5400.551 dlambda = 0.009
point #9 x = 872.632 lambda = 5852.431 dlambda = 0.059
point #10 x = 995.469 lambda = 6266.479 dlambda = 0.011
point #11 x = 1067.736 lambda = 6506.531 dlambda = -0.001
point #12 x = 1265.097 lambda = 7147.100 dlambda = -0.060
point #13 x = 1339.935 lambda = 7383.942 dlambda = 0.038
--------------------------------------------------------------------------
Coefficient a4 : 7.593870E-11
Coefficient a3 : -3.984758E-07
Coefficient a2 : 4.818265E-04
Coefficient a1 : 3.26645
Coefficient a0 : 2855.872
--------------------------------------------------------------------------
RMS : 0.149279

Fichiers intermédiaires effacés
----------------------------------------------------------------------------
Date début de prise de vue : 17/02/2017 15:06:09
Durée : 48.0 secondes
Date de milieu de prise de vue : 17.630/02/2017
Jour Julien du milieu de prise de vue : 2457802.1295
Pouvoir de résolution : 525.5
Ok.


A tester sur le CIEL.
Merci pour votre aide.

[Peter Somogyi]

Thank you Robin and Etienne, I will try all with recalculating the pixel size.
(In fact, never had any problem with 6.45 or 12.9 at bin2x with the LHires/150 - there I had to care of A/pix value in the list file, however it worked perfectly as designed).

Be aware that RMS doesn't tell the real error - just the spline fit error to the given points, but it tells nothing about the real spectra fit.

I'll get back later when looked by details.

Cheers,
Peter

[Olivier GARDE]

Si tu as changé de lampe c'est "normal" d'avoir des différences car entre 2 lampes, il n'y a pas forcément les mêmes raies en emissions, je me suis aperçu de ce problème en identifiant les divers raies d'une lampe à l'autre. Et il y a des variations.

[Robin Leadbeater]

Hi Peter,

Thank you Robin and Etienne, I will try all with recalculating the pixel size.
(In fact, never had any problem with 6.45 or 12.9 at bin2x with the LHires/150 - there I had to care of A/pix value in the list file, however it worked perfectly as designed).

The pixel size entered in the General tab is not used if you use the "file mode" as you have specified the mean A/pixel in the file. (You can try entering any pixel value and it will still work). In the "ALPY predefined mode" the entered pixel size is used to calculate the A/pixel internally.

Cheers
Robin

[etienne bertrand]

Bon ce soir sur le ciel la résolution et le RMS marchent, par contre j'ai une bizarrerie dans le flux de l'étoile. J'ai l'impression qu'il y a comme des interférences, le flux oscille sur la longueur du spectr et il n'est pas uniforme, on le voit très bien dans le spectre traité.
Avez vous des idées ?
Avec la 414EX il y avait des problèmes sur le Lhires avec le résaux 2400, mais sur l'Alpy ça ne m'avait jamais fait ça

En jouant sur la réponse instrumentale j'arrive à avoir le continuum d'une A3V mais le spectre est assez sale quand même.