Difficulty with LhiresIII H-Beta calibration-2400 l/mm Grat

[James Foster]

I got the Vega H-Beta spectrum fully dispersion/IR corrected in Isis V5.9.1from the medium resolution Vega (normalized) spectrum in the Elodie database (R=10000); see:

The resolution came out to be R= 11227. Here is the .lst file figures I used:
2
0.15112
4788.93
4847.81
4879.86

Please note that the 4847.81 is very weak (short) on the continuum of the Ar-Ne calibration images. I still don't understand why the other major lines in the calibration file don't line-up with the Ar-Ne RELCO lines when H-Beta and the Fe lines are used to calibrate as Paolo indicated in his earlier post.

James

[Paolo Berardi]

Hi James, generally speaking, the wavelength calibration should always be performed with lamp lines. Using high resolving power many observed Doppler shifts are clear, i.e. radial velocity of the system, RV of the observed component in a binary, the RV due to the orbit movement of the Earth, etc. So you can't get a precise calibration using the target lines (at rest wavelengths).

I think you should investigate the lamp issue, why not, by repeating the H-beta range observation (also possible in daylight time). Can you rule out an overwriting of the lamp frame with that of another spectral region? If the H-beta lamp spectrum is actually what you showed us, perhaps it is worth changing the lamp with another Relco one.

If useful, I show you my H-beta range Ar-Ne lamp 2d spectrum (provided with the Lhires III upgrade kit). Please note that I used the 1200 l/mm grating (I don't have a spectrum with the 2400 l/mm one in that spectral range).



Another element for the lamp analysis could be the 2400 l/mm H-alpha spectrum (you should have it).



Look your spectrum like the one on the bottom? H-alpha line (blurred) is near the center. The top spectrum is related to the standard Lhires III neon lamp.

I hope you come up, the use of the lamp is very important for our activity.

Paolo

[Peter Somogyi]

Hello James,

3 lines is the bare minimum for order=2, you should include as much as possible!
3 lines also won't determine any error, RMS will always be "too nice" with (order+1).
Avoid such a condition whenever possible, and try to include as many lines adding one by one, till RMS doesn't jump suddenly! (a little combinatoric game, to include lines one by one from an external library, and experiment which makes the RMS wrong, of course works above order+2 lines only, wide spread evenly on the chip)
Your chip is bigger than mine, so you'll have even more usable lines. The more lines you add, you improve RV measurement quality the more (can't check visually)!
After some practice (and first hard effort), it will be reusable.

Here is my ISIS log snippet for the 5 lines (cal file already posted):
Wavelength fitting error
Line #1 x = 200.761 lambda = 4847.810 dlambda = 0.000
Line #2 x = 409.623 lambda = 4879.864 dlambda = -0.001
Line #3 x = 469.678 lambda = 4889.042 dlambda = 0.001
Line #4 x = 572.785 lambda = 4904.752 dlambda = 0.000
Line #5 x = 760.454 lambda = 4933.209 dlambda = 0.000
--------------------------------------------------------------------------
Wavelength file : C:\csill\428exm\20170401_lhires\g2400\30_bfcyg_Hb35\calib\g2400_Hb_after.lst
Coefficient a4 : 0.000000E00
Coefficient a3 : 0.000000E00
Coefficient a2 : -2.547079E-06
Coefficient a1 : 0.15504
Coefficient a0 : 4816.633
--------------------------------------------------------------------------
RMS : 0.001079

2400/mm grating, 35 micron slit, internal Ar/Ne bulb.

Here is my selection of the 4879.864 line (1 x 20 seconds shot):

cal_select.gif (51.42 KiB) Viewed 7717 times

My firm knowledge about tellurics, that we amateurs by default should never ever remove it from the final spectrum for a database (no chance for perfection unless taking a very good quality ref.star dedicated just only for tellurics), unless requested explicitly.

Cheers,
Peter

[Paolo Berardi]

Hi James, I think I've got it. The lamp spectrum should be correct but your lamp doesn't contain argon! I think you're using the standard Lhires III Ne lamp (with a long exposure frame).

A quick check on NIST database revealed a good match with all Ne I lines. See here:



I did it hastily, please check for yourself, below the DB link (you need to provide the element, Ne, and spectral range).

https://physics.nist.gov/PhysRefData/AS ... _form.html

Pay attention to blended lines, not suitable for the calibration purpose.

Let us know,

Paolo

[James Foster]

RE:"The lamp spectrum should be correct but your lamp doesn't contain argon!"
This is great to know! I wonder why this is the case (No Ar)? I just bought from Shel yak in Jun17. I did order many (6) extra
Ar-Ne lamps so I may try to replace this and shoot again. Here is a cell phone image of my now identified Ne (only) lamp; my Alpy calibration box has a more magenta light source.

Thanks again for your analysis!

James

[James Foster]

To: Peter,

RE:"3 lines is the bare minimum for order=2, you should include as much as possible!
3 lines also won't determine any error, RMS will always be "too nice" with (order+1)."
I reprocessed my CH Cyg Hb spectrum with the Ne lines provided by Paolo, Here is the 2nd order error for 6 lines I used in my .lst calibr. file (note Ne line 4488.92 used, not shown):

From Isis Dispersion window, 2nd order:
4837.276 0.034
4863.148 -0.078
4884.897 0.023
4928.193 0.047
4957.057 -0.027
RMS = 0.060

James

[Paolo Berardi]

Hi James,
yes, the lamp colour confirms the standard Neon lamp. I suggest to change with the Relco type, especially if you want to observe in the blue-green spectral regions where there are few and too faint neon lines.

Using the main lines (NIST wavelengths) I obtain this result with a third grade polynomial:

Wavelength fitting error
Line #1 x = 216.809 lambda = 4788.920 dlambda = 0.000
Line #2 x = 431.240 lambda = 4821.920 dlambda = -0.002
Line #3 x = 531.720 lambda = 4837.310 dlambda = 0.002
Line #4 x = 844.711 lambda = 4884.920 dlambda = 0.000
Line #5 x = 1324.609 lambda = 4957.030 dlambda = 0.000
RMS : 0.002918

If you want to consider a second grade polynomial, the same lines return:

Wavelength fitting error
Line #1 x = 216.809 lambda = 4788.920 dlambda = -0.002
Line #2 x = 431.240 lambda = 4821.920 dlambda = 0.001
Line #3 x = 531.720 lambda = 4837.310 dlambda = 0.005
Line #4 x = 844.711 lambda = 4884.920 dlambda = -0.004
Line #5 x = 1324.609 lambda = 4957.030 dlambda = 0.001
RMS : 0.004921

I think this is not bad.

The calibration file I used with your data:

3
0.152
4788.92
4821.92
4837.31
4884.92
4957.03

Ciao
Paolo

PS: your spectrum is undersampled, I had to remove the cosmic ray filter in ISIS, otherwise the star continuum comes out altered. You need a larger spectrum image on the CCD sensor or smaller pixels (no binning if you are using it)

[James Foster]

I installed the Ar-Ne lamp I had, but since there was no diffuser, I used a piece "milky" plastic, but this ended-up not working well
since the filament projected in the middle of my _neon calibration files. I took this out and replaced it with a piece of translucent tape
as seen attached.

James

[Paolo Berardi]

Well done James!

Paolo

[James Foster]

Actually I don't think I needed the diffuser. What I thought was the filament projected onto the calibration _neon files was actually ionized H!
I moved the micrometer on the LhiresIII around my Ha and Hb settings (1970 and 1445 micrometers) and the "filament projection" follows the surrounding Ar-Ne lines.
I barely see it on Hy and not at all in the CaK region.

James