Please recognize that it has taken a long time to find suitable lamps to calibrate the TTF filters. If you have recommendations for better lamps, please let me know. In a nutshell, we have established suitable lamps for the RTTF (see below) but the situation is still not satisfactory for the BTTF. Click on one of the filters above to go straight to the recommended calibration lines.
Here are sample arcs from an UCLES campaign with Steve Lee. Here is OSRAM's helpful table of lamp lines or, if you prefer, a zipped ps file. The problem is that most of these lamp lines are faint. In practice, lamps with ideal elements like Ca and Rb are hard to find. The entire chimney is soon to be reworked with a whole battery of low-power lamps enclosed within condensing optics, much like the WHT. This will render even the faintest lines much brighter.
At site, we have access to
FeAr CuAr ThAr Ne He D HgCd Hg Na K CaKr Xe
We have the following lamps on order (Nov '02)
Cd Zn HgCd Hg Rb Tl Kr
The D lamp is a fairly useful whitelight source in the blue although Kr and Xe are better, and W is good everywhere else. Here are some really good arc lists from the Figaro release. The deuterium and hydrogen lists come from calculations I did with the aid of Haken & Wolf. Don't forget that the deuterium isotope requires a reduced mass to get the orbits right.
Deuterium & HydrogenIf you still cannot find what you want, see the ESO, KPNO and WHT web pages, and the claret-coloured Bureau of Standards book in the AAT library. The WHT web site (http://ing.iac.es) provides good calibration spectra of many lamps. A very useful new facility is called Spectral Atlas Central (http://www.noao.edu/kpno/specatlas) at KPNO. Here, you can dial up spectra for
Iron Argon (300-1090nm)
Helium Neon Argon (330-1110nm)
Copper Argon (under construction)
Calibration lamps for specific TTF filters.
Arcs discussed here apply to the intermediate band 5" diameter blockers, and the narrowband 3" diameter blockers. For the UBVRIz filters, there are no suitable arcs which produce a few unambiguous lines over the broad window. This is an age old problem. In their place, either extrapolate from calibrations within the intermediate blockers, or use whitelight through the narrow 3" filters to simulate an emission line. Both methods work well enough although the former is more accurate than the latter.
There are some intense lines available from internal dome lights, street lights and the atmosphere in general.
The line blends below are estimated from Morlen (1973)
Physica Scripta, 8, 249 with flux weighting. The
first number indicates a very rough relative flux in arbitrary units; zero
means very faint line.
Intermediate band blockers(5" diameter; very expensive so do take care)
|375nm/19 (B0): Three
methods. I. low power Kr pen lamp.
II. high power Xe lamp for continuum
structure + Hg line at 3650. III.[OII]
nebular doublet at 3726, 3729.
I. Low power Kr pen lamp glued to front of diffuser disk.
(get exact wavelengths from book, p 7-60)
For more details on Method II, see below for B1.
spectrum for low power Kr lamp (Zc=-2); lots of lines.
This is the best calibration and requires dz=3 stepping, and 10 sec per
step (we are still waiting for brigher lamp). Recommend
that you calibrate with Kr once in daylight then track with bright Hg line
Example spectrum for high power Xe lamp (Zc=-2); not entirely clear why asymmetry goes the other way. Faint lines?
|390nm/17 (B1): Two
methods. I. low
power Xe pen lamp. II. high
power Xe lamp for continuum structure + CuHe for single line.
power Kr pen lamp for continuum structure + emission line.
I. Low power Xe lamp glued to front of diffuser disk. (Do not stick lamp on guide probe since this is 0.1m from focal plane rather than 4m; even though you get a lot more flux, you will not see lines in the spectrum.)
II. Use Steve Lee's blackbox high power Xe lamp for strong blue continuum which we install at prime focus access into top end; this is precarious as lamp is controlled by heavy transformer. At Zc=-2, we see well separated orders (FSR ~ 200A), and we can use the filter FWHM edges for calibration (3904-83-dT, 3904+83-dT) where dT = 0.2*(24-4) C = 4A offsets the Barr measurements for the cold dome temp. The He lamp gives a nice line at 3888.658 which proves that the corrected filter edges calibrate the response well. (We tried HgCd, CuAr and Steve Lee's small Kr lamp but all too weak even with mitll_on_off_large; we also tried using VPH but this cuts out just below B2.)
III. Low power Kr pen lamp reveals continuum structure as shown in the picture opposite. The line messes things up a little.
spectrum for low power Xe (Zc=-2); fit faint line
with gaussian, bright lines with lorentzian. This gives very accurate fit
which one can improve a bit by subtracting the asymmetric continuum. This
requires dz=3 stepping and 10 sec per step. This spectrum
shows the continuum underneath and this spectrum
shows how the arc looks once the continuum is removed.
Recommend that you calibrate with Xe once in daylight then track with bright Hg line during night.
Example spectrum for high power Xe (Zc=-2); note that the asymmetry is due to the rapid decline of the EEV2 response. You can divide this out with a straight line.
Example spectrum for CuHe (Zc=-2); the first feature is the He line, the second is noisy continuum structure as seen in the Xe spectrum (ignore the plot header).
Example spectrum for Kr (Zc=-2); this is may be important lamp for B1 but we need a bright source (still on order!). Here is a long integration (20 sec per step) of the same. You see continuum structure and a line to the blue.
|411nm/15 (B2): Two
I. HgCd (these are nice and bright so only need mitll_on_off. We had this mounted with Ne in the top chimney. These lines become blended at Z_c > -2. If you are worried, try to extrapolate to HeII filter or H-beta in B4 lamp.)
II. FeAr (or CuAr) source (these lines are so faint that we put the FeAr lamp on a stick with a reflector and slid into the beam. We use mitll_on_off_large to get a 30x30 postage stamp.)
467nm/32 (f098): untried, belongs to WHT
477nm/32 (f099): untried, belongs to WHT
untried, belongs to WHT
|500nm/32 (B4): Three
I. Ne II.
followed by D III.
Method I is now the favoured calibration but we must use the D
lamp to demonstrate that Method I is giving the right calibration. Method
II is the safe approach and
The D lamp is awkward to use. Turn in on for 15 secs to warm up, turn it off then on again immediately to fire it: remember to turn off afterwards since no warning light here. The D lamp gives rise to structured continuum so, under iraf, fit the lines separately using splot/ddgqaanqq ddgqaanqq ddgqaanqqq combinations as opposed to the normal splot/ddlllqaanqqq method. (See also splot/kl)
I. Ne source (these lines are a little faint so use mitll_on_off_large. See the example spectrum on right hand side. You get a very accurate calibration if you use gaussian fits on the first three lines. You really want only Ne in the spectrum. Anything else messes you up, e.g. D line, Hg line will swamp spectrum.)
III. DHe source (Use mitll_on_off_large, you can see the He line at 5047 but it's blended?)
III. Compact Planetary Nebulae(Dopita & Hua 1997, ApJS, 108, 515) Only use this method in desperation. See the TTF Photometric Standards folder. Bright emission lines at H-beta and [OIII]. You will need to do APOFF to find object in small window and 10 sec per step, so maybe too slow. But excellent for photometric cals.
Example spectrum for Ne (Zc=-3, Zf=0)
Example spectrum for Ne (Zc=-1, Zf=0) - this gives a good 0.2% resid calibration with the bright line at 5193. I recommend that BTTF be used at Zc < 0 for most work since the arcs are excellent.
Example spectrum for Ne (Zc=+1, Zf=0) - this is probably not a good fit since 5193 has now wrapped onto the faint Ne lines! Compare with Zc=-1 example.
spectrum for D (Zc=-1, Zf=0) - this
is properly sampled so you can fit wing and core, but centroid of D-beta
is still not that great!
spectrum for DHe (Zc=-1, Zf=0) - this fit gave a superb
0.04% resid. Fit each line separately. Note how we fit D-beta with blue
wing and core. The sequence went splot/ddggqaanqq,
source (the two blue lines are resolved
at Z_c = 0)
|Example spectrumfor Ne (Zc=-3, Zf=0)|
570nm/45 (f133): untried, belongs to WHT
596nm/47 (f135): untried, belongs to WHT
untried, belongs to WHT
|580nm/18 (B6): Ne
source (Beware! If someone puts Ne
source on with Hg, there are two Hg lines at 5769.598and blend
5790.164. These could be used if no Ne. Curiously, all lines together do
not give a good fit and this may be due to Ne/Hg clash.)
spectrum (Zc = -1, Zf = 0)
I. Ne source II.
I. Ne source (this is much the preferred lamp as there are three lines)
II. CuAr source (this is a fairly decent back up; consider using awkward D lamp as check of calibration)
spectrum for Ne (Zc=+5, Zf=0)
Example spectrum for CuAr (Zc=+5, Zf=0)
Example spectrum for Ne (Zc=-2, Zf=0)
|710nm/26 (R1): CuAr
|Example spectrum for CuAr (Zc=-2,Zf=0)|
|733nm/115 (BP 733): CuAr source
These four lines give a very accurate fit, and correctly predict that the faint line to the blue is 7147.042, to an accuracy of 0.3A. The lines in between are too blended to be useful.
This filter belongs to Brian Schmidt and is kept at the MSO 2.3m - it was used very successfully by the Cooling Flows team of Edge, Bremer, Crawford, Baker, et al. in Mar 2001
|Example spectrum for CuAr (Zc=-2,Zf=0)|
|760nm/28 (R3): CuAr
|Example spectrum for CuAr (Zc=-2, Zf=0)|
|810nm/33 (R5): CuAr
|Example spectrum for CuAr (Zc=+4,Zf=0)|
|860nm/40 (R6): CuAr
source (At Zc < 0, use
for the blend wavelength)
spectrum (Zc=-2. Zf=0)
Example spectrum(Zc=+1, Zf=0)
|910nm/40 (R8): CuAr
Narrowband blockers (3" filters; rarely used for wavelength calibration)
If your intention is to use the narrowband blockers with the TTF, simply calibrate lambda-z with the intermediate filters and then switch to the narrower band when ready to observe.
These filters have another occasional use. When calibrating B-TTF and R-TTF at low resolution, the easiest way is to extrapolate to the line using accurate calibrations within the 5" filters. At present, we need the 3" filters as well to help us span the full optical range. These filters are:-
|[OI] 6310/40 (Ne
There are three bright lines. The blue line gets brighter with large tilt to filter. For sausage cube, 1 sec per step is fine. High resolution is best.