Feb 25: I've just started looking over the test 950 data. This is 1.2 cm low noise data. It looks like we can get the wing for some/many modules from this data. George may be right about doing two sets of runs: one for wings, another for central portion of SRF. My worry is not so much the wing but the central portion...I'm not confident of being able to deter "true" SRF shape (or even the widths) to the desired 1% accuracy. None of the various test runs I've looked at has good consistent central SRF shape. I'm a bit surprised Paul vD said his off-axis computation is different than mine. I think my calc is pretty much exact. An interesting obseration: the fringes are shifted in phase depending on the off-axis angle. If the true AIRS SRF is close to symmetric, then this provides a means to determine the offaxis angle...the fringes of measured and calc will be in phase. This gave 0.4 degrees for M12. It doesn't seem possible to get an exact match between calc and measured using Geroge's G+L SRF model and off-axis. Either the data is bad (my guess) or AIR has an asymmetry. ------------------------------------------------------------------------------ Feb 23: I've re-processed the test 386 M12 fits using the corrected off-axis apodization function. The results are generally quite similar to what I got before. t386_m12_srfavg_0_1494.eps t386_m12_srfavg_4_1517.eps These two plots are the new replacements for the old figures 1 and 2 shown in pub/hannon/airs/t386/386.html. Not much change in the final obs-calc. If you have copies of the old plots with you, you can show them at the AIRS meeting instead of bother with the new plots above. One notable change was how I averaged the data. The old "average SRF" data was just a raw sum of the "good" channels, with absolutely no adjustments made for aligned the channel centers. The new average SRF comes an average where I aligned the mid-point of the SRF at half height (ie mid-point of the 1 FWHM chord). I compared this to the raw and peak aligned average SRFs...they all differ a bit. The fit error jumps up to ~2% for the "best" fit if I use the raw average. I did the fits for test 387 M12; the results are nearly identical to test 386 (as expected). The test 543 data looks rather dubious for SRF fitting...the SRF shape is jumping around too much to inspire much confidence in any fit result. (The shape is more stable than we see at 4 um, but it's the same type of thing). The test 386 data looks pretty stable in shape, at least when eyeballing it. t543_m12_srfavg_0_1446.eps t543_m12_srfavg_4_1460.eps These two plots are the fits of th test 543 data for M12. The noise is too high to see the wing, or even to be sure if there is any asymmetry in the SRF shoulder (there does seem to be a little). I used 0.4 degrees just because that is what I used before; it is not a fit for the "best" angle. Here's how George's Gauss-Lorentz model changed for the various fits of the M12 data pretending the average SRF was a single channel at 670 cm-1 (angle in degrees) test| angle | g_fraction | g_exponent | l_exponent | resolution ----|-------|------------|------------|------------|----------- 386 | 0.0 | 0.938 | 2.62 | 2.21 | 1494 386 | 0.4 | 0.936 | 3.05 | 2.24 | 1517 387 | 0.0 | 0.932 | 2.63 | 2.31 | 1494 387 | 0.4 | 0.930 | 3.07 | 2.34 | 1517 543 | 0.0 | 0.929 | 2.63 | 2.03 | 1446 543 | 0.4 | 0.927 | 2.76 | 2.05 | 1460 I've included the 387 results just to give an idea the stability of the 386/387 data. The average SRFs for 386 & 387 agree to an RMS of around 4E-4 (on a 0 to 1 SRF scale)....pretty good. The "g_fraction" is fairly consistent. The "g_exponent" varies quite a lot with offaxis angle for the test 366/387 data. The "l_exponent" more or less controls the wing. I would not trust the test 543 wing data at all. The resolution changes may be related to LMIRIS tweaking knobs between measurements, or may be an artifact of the different FOVs, or possibly the higher resolution of the 543 data (a result of the smaller FOV and a larger opd...3.50 cm & 1.9 degrees for 543 vs 2.88 cm and 2.9 degrees for 386). Overall I don't think we can say much about the "true" SRF shape or wing, other than it appears to be a bit higher res and smaller wing than expected. The wing from the test 386/387 is probably about right...I don't trust the central/wing shoulder, but the wing is probably about right. Paul VanDelst sent me email monday; he says he gets an off-axis apodization that is "similar, but different". I don't think I can get any better agreement between average SRF and convolved apod&AIRS using my stuff. The test 386 & 387 data had a 2.9 or 2.86 degree FOV...2.86 degrees corresponds to a 0.30" aperature at 6.0" distance. This gives more than twice the light as a 0.20" aperture (1.91 degrees) used for test 543. Paul said the following... --- BTW, the 2.9degree FTS FOV is exactly what I calculated to provide a 10cm beam projected onto the AIRS entrance apertures which should illuminate them all. The 1.94deg. would only give a 6.7cm beam which, by my calcs, means not all of the apertures would be illuminated (which was obviously not the case for the 386 data). ---