Nestor Espinoza presented some calculations on bright limits/efficiencies for what would superstripes/substripe look like if implemented on NIRISS/SOSS (along with code on how those calculations were made). If you want slides, let him know:
- Michael Regan : efficiency seems large. Nestor Espinoza comments that the efficiency he plotted was the total efficiency on reading the entire subarray size. True that this is not the efficiency per stripe/wavelength region.
- Elena Manjavacas : To make a strong science case, would be good to include how many more planets would such a change allow to observe on different modes. This would also make the case for pushing for extra calibrations. Do we know which ones this would imply? Michael Regan notes certainly darks, biases; unclear (unlikely?) if flats.
- Everett Schlawin : also mentions on the efficiency question. Intuitively, after each stripe one needs to reset all the detector, so that on its own is a lot of time. Nestor Espinoza mentions that this is indeed accounted in the calculation (via fast reset reads). Michael Regan notes you don't do this per pixel, but per row.
- Loic Albert : notes that there are really two efficiencies — one full read efficiency (the one shown), and the other one that actually defines the SNR as a function of magnitude because "throwing away" photons. There might be a sweetspot somewhere across those "efficiencies".
- Michael Regan : might be good to think about doing different setups for different people: bright star folks might be happy with small stripes SNR-wise, but faint object folks would probably not.
- Aarynn Carter : are we fixed on a given set of parameters per instrument? Nestor Espinoza mentions this came up on a fortuitous conversation with Nikolay Nikolov and Mario Gennaro on the halls. No, there could be M-G-K-focused parameters, for instance. Michael Regan notes that each of them would be like a new subarray. Nestor Espinoza notes that, e.g., in APT, one could think this appears as "SStripe M" or "SStripe G" and then one would select the one that makes more sense to a given spectral type, say. Michael Regan notes it would be wise to ask for what you want (e.g., this) on the first try.
- Nikolay Nikolov will check if the orientations are correct for NIRSpec/PRISM "flatten the curve" idea. We want pixels to start being read on the red/faint end for this to work.