Thanks, John. There are a few non-nearest-neighbor channel combinations that have large matrix entries, but it will take more study to know whether they are really crosstalk. (They could also result from non-cosmic-ray glitches leaking in to the sample). From a quick look at the average across detectors, the next-nearest-neighbors are at about 0.05% and the other same-column channels are similar. I've been scratching my head about this since yesterday afternoon. The CMB analysis had much bigger error bars, so the "null result" was not a very strong one, but the A-B difference was clearly significant. I don't see any such difference in the cosmic ray analysis. - Walt

Hi Walt, The dominant cause of particle events in Planck detectors were hits to the silicon frame, not hits to the sensor itself. We have much larger frames, more cross section, higher temperature, more heat capacity, better heat sinking, and also a frame common to all the detectors in an arrays. Is there evidence for a frame hit population? It doesn't seem like it based on your results, definitely interesting(!), but one needs to be careful. Can you do any tests of this? E.g. do dark SQUIDs have the same crosstalk from a neighboring cosmic ray hit as a regular detector? Jamie -----Original Message----- From: bicep2-list-bounces(a)lists.fas.harvard.edu [mailto:bicep2-list-bounces@lists.fas.harvard.edu] On Behalf Of Walt Ogburn Sent: Friday, August 30, 2013 7:32 PM To: John Kovac Cc: bicep2-list Subject: Re: [Bicep2-list] Measuring crosstalk with cosmic ray hits Hi John, Thinking about this more, I have an idea. I'll try to think of a way to test this... and maybe a graphic to explain it more clearly. I suspect the difference is that the cosmic ray analysis truly isolates crosstalk, while the CMB map regression can be sensitive to other effects. Planck maps smoothed with our mean B_l file. For any detectors that aren't a good match to the mean B_l profile, there will be some residual. If there's no crosstalk, and only some beam shape residual, my regression will pick out the overlap between [beam residual] and [simulated xtalk beam]. Now the main beam and presumably any beam shape residuals are small at 1.5 degrees separation, so I initially thought this shouldn't matter for the CMB crosstalk regression. However, it's actually the overlap of the residual and the 1.5 degree displaced beam that matters. I think this can be large enough to shift the regression coefficients, and therefore the estimated crosstalk. There are lots of ways to test this, but one would be to repeat the sig+noise sim (blue lines in Fig. 1 of 20130826_cmb_xtalk) using a signal sim with no crosstalk, but with a wider beam profile. - Walt _______________________________________________ Bicep2-list mailing list Bicep2-list(a)lists.fas.harvard.edu https://lists.fas.harvard.edu/mailman/listinfo/bicep2-list