Friday, May 12, 2017

SaltLick Update and Data Summary

Folks,

FYI I've upgraded the SaltLick server in the lab.  It now contains three disk drives as a single 4TB redundant (RAID5) volume.  There is room to add one more drive, which would max out the storage at 6TB.  We're presently sitting at 58% of capacity, so there's no imminent need to add that disk.  I did go ahead and buy a complete set of replacement disks so that we're OK for future failures that might occur.

Now that there is room, I've consolidated the data from Everything EXES on dropbox with all the other lab data that isn't on the dropbox, so SaltLick is now (I believe) a comprehensive raw data set.  I need to collect some of the reduced data still, and once that's done SaltLick will be a complete dataset.

Included below is a table summarizing the data that presently housed on SaltLick and on DB/EE/Data, in case you're wondering where to find something someday.  This table is also stored at the root of SaltLick/raw, and I can update it as more data finds its way home.  Please let me know if you see any errors that I've made.




<!--This file was converted to xhtml by LibreOffice - see http://cgit.freedesktop.org/libreoffice/core/tree/filter/source/xslt for the code.- no title specified
cooldownSL_rawSL_reducedDBEE_datadate 
11.7GB----Apr-Jul 09 
2-------- 
32.2MB----Oct-Nov 09temp data only
411GB----May-Jun 10 
5-------- 
6-------- 
71.3GB----Jul-Aug 10 
86.3GB----Aug-Sep 10 
923.4GB0.5MB--Oct-Nov 10 
1015.4GB----Nov-Dec10 
1164GB----Jan 11IRTF
12142.8GB----Mar11–Dec12 
1333.5GB----Feb-Aug 12 
1443.7GB--15.8GBJan-May 13 
1561.8GB--50MBJun-Aug 13 
1690.9GB----Nov-Dec 13 
17224GB--137.5GBApr-Oct 14Cyclops
1820.2GB--3.3GBOct-Dec 14 
197.3GB--1.6GBJan-Mar 15 
20-------- 
21190.8GB8.8GB199GBFeb-Mar 15Wolverine
22114.8GB50.8GB79GBMar-Sep 15Storm
23-------- 
246.1GB--6.1GBFeb-Apr 16 
25157.1GB7.3GB157.1GBMar16Colossus
266.9GB--6.9GBApr16 
2731GB--31GBOct16 
2812.3GB--12.3GBDec16 
29186.8GB28.9GB186.8GBJan17Deadpool
30269.1GB--269.1GTBFeb-Apr 17Kitty
      
      

-->

Thursday, May 11, 2017

CD31 resolution tests

We took C2H2 Q branch spectra in himed and med modes with all slits and the pinhole.  We also took rinsed cell spectra to go with them.  Script and save files are in analysis/cd31/specfocus.

I used the pinholes to trace the path of dispersion and find the dispersion in wn/pix.  I used a linear fit for wn/pix, although a quadratic would probably be <=3% better.

I used clickfit to fit along pixels and corrected for the fact that the slits didn't run along pixels later.  This was a small effect.

For the high-med, I fit Q(9)-Q(15) and Q(15)-Q(19).  For the med, I fit Q(10) through Q(23) along with R(0) and skipping Q(20) because of a hot band.  As the slits got wider, the lower lines really blended together.  In fact, no fits were good for slit 6.

In all cases, I fit many rows of each line and then took a median of all the fits that looked good for a particular line in order to calculate the resolving power in the bottom figures.

himed Q branch with slit 4 (rotated 90 degrees in IDL to work with clickfit)
med Q branch with software rotate and only Q(10)-Q(23) (right to left) extracted

himed resolving power as a function of x pixel.  Each slit is different color.
Medium mode resolving power as function of X pixel.  Each slit is in different color.  Predictions from ETC are horizontal lines with appropriate color.  Note that Slit 6 was so broad that the fits didn't work.

After finding that C2H2 med mode was ~85% of the prediction, which seems reasonable since we bias ourselves toward the high resolution, we looked at H2O at John Carr's 1838 cm-1 setting.  The actually wn0 is a little different to provide good lines to fit.  

Followed the same procedure with pinhole and all slits.  The lines were present in the cell even when continuously pumping.  Used an aluminum plate right in front of the window and adjusted the level in software to provide a good reference.  Chose to divide rather than subtract.  

Got resolving powers that were far below ETC prediction.  We're going to look at more settings going toward longer wavelengths.  Will check the ETC, too.

Resolving power for 1838 (nearJohn Carr's H2O setting) as a function of x pixel.  Note that  the line  near x=850 was the brightest.  The fits make me think maybe it was resolved by the narrowest slits.