CMB-S4 frequency bands v1.99
For the low frequencies, it uses the definitions suggested by Adrian ( email from March 21st 2017) for the first 3 bands (10,15,20 GHz). For the upper bands it just uses the existing 8 bands of the S4 Data Challenge 1.0 (30-270), as defined on the S4 data challenge page: 
This gives 11 possible bands, distributed over what I suggest we might call five "windows" 0: 0-25 GHz 1: 25-50 GHz 2: 75-115 GHz 3: 125-175 GHz 4: 190-300 GHz
For the relative NETs, the spreadsheet lists in yellow the values that were the basis for the 1.0 forecasting effort which come from , in blue values which Adrian had emailed and in green values which Denis has updated his posting: 
Whatever we use for NETs, I want to emphasize that our approach has been to use these numbers only to assign ratios of sensitivities between bands, used to scale achieved noise performance from actual maps made in the closest available existing band. That has been how we have set the map noise levels for DC 1.0.