Hi John, gcp 459 was not an rgl--I have changed this to an rgl in my posting. It is indicative of the other pixels, however. That shape in figure 7b seems to be fairly repeatable. You can click through all the analyzed spectra in the link below it. I've added another consistency check to the posting where the direction changes. We're definitely seeing something interesting here and I'm not as of yet sure as to what it is. I will attempt to get FPU maps of this effect up by tomorrow's meeting. I haven't as of yet worked out a perfect spec (although I'm looking into the takahashi relgain spec)--I've been checking that the repeatability is within 1-2 frequency bins or less than 1.2 GHz. This is definitely a spec I'm open to changing if you have an idea. Grant and I were talking about this. I was under the impression that this is late-stage analysis to be done after consistency has been confirmed but the code is in place and I can do it. -Jon

Hi All, I've added a significant amount of plots to my posting (http://bicep0.caltech.edu/~spuder/analysis_logbook/analysis/20121216_FTS/) as well as some links to make it easier to navigate. I couldn't get everything up before the satellite pass ended so I'm sitting on a lot more interesting plots which I'll try to send through skynet tonight or GOES tomorrow morning. I'll walk through some of the changes (my local version is more up-to-date so forgive me if I describe plots which aren't there yet). Under the "Spectra" section, I've added figure 4 which terminates at a higher frequency to show we aren't getting much out of band power. I've plotted an inset which zooms in specifically in the 150 to 850 frequency range. Although there is repeatable signal in this range (just shy of 300), it is well below the percent level. In addition, I've added links to all the spectra analyzed so far. Feel free to browse around and note anything weird. I've gone back and analyzed the data for complete focal plane coverage. Figure 6 shows a detector map and gcp coordinate plot of the calculated bandwidths and band centers. Here, it's worthwhile to note that the way I calculate bandwidths locks to the nearest frequency bin so there are discreet values which the bandwidth can take. There's nothing particularly wrong with how I'm doing this but I'll probably make this code more elegant on a second pass. Essentially, the uncertainty in my bandwidth calculations is 0.6 GHz. I've added a significant amount of plots to the "Consistency checks" section. Firstly, figure 7 investigates the mismatch for spectra broken into left moving and right moving scans. I analyzed the dk315 data set to produce these values which gives total focal plane coverage. Figure 7 has 3 plots, the first being a focal plane detector map of the mismatch. The mismatch is calculated by taking (cf_set1 - cf_set2)/ cf_tile_average where here cf stands for center frequency. Note that the color scale is +/-1%! Next to this, I plot the same statistic but in gcp coordinates. The third plot is a scatter plot, plotting cf_set1 vs cf_set2. You can see that this seems to trace a line going through the origin fairly well. In my local copy (which I'll send up as soon as possible), I also make a histogram of the mismatch. This shows that 100% of the rgls have less than 0.5% band center mismatch. Figure 8 is the same as 7 but for the bandwidth. Keep in mind that the bandwidth takes discreet values which are multiples of the frequency bin (0.6 GHz) so although the scatter plot looks like it's not very populated, most of the points are actually on top of each other. The histogram clears this up. I've modified the dk rotation consistency check to add the dk315 data set. Figure 9 right shows a very repeatable sinusoidal residual in the spectra. It's quite possible that this is due to an illumination effect skewing the spectra. To test this, we've taken data at dk135 to have a 180 from dk315. This data set will be analyzed tonight and posted asap. Figure 11 shows scatter plots for the permutations of the three dk angles. The "Positive vs Negative from WLF scans" has also been updated to include the full focal plane. Recall that positive and negative are defined as the distances relative to the zero path. This test shows some very repeatable systematic where the negative path tends to give results higher than the positive path. This mismatch is still of order 1% so it should not be a big deal but it is interesting. There is also a much larger spread in the values. The final consistency check in my posting (more are being analyzed) is the scan speed check. We ran the 4 pointings at dk315 at 2mm/s and at 1mm/s. You can see from the normalized spectra in figure 14 right that these two data sets sit right on top of each other, though there is some small residual left over. Figures 16 and 17 are similar to 7 and 8, where I plot a detector map, gcp plot, scatter plot, and histogram (coming soon). Again, you can see that the mismatch is well below the % level, though there are a handful of outliers. Grant and I have looked at spectral mismatch and although we don't have the plots up yet, I can say that for the consistency check we've run, we're seeing what looks like sub % repeatability. I have to apply some cuts before I can sure up that number. We believe we've finished taking all the data requested today. We took data with a mirror in place of the input grid (time reverse), data at dk135 to have 180 from dk315, and took spectra after moving the short arm roof mirror. I'll email out updates to my posting as they come. -Jon