CMB Polarisation Signal Demodulation with a Rotating Half-Wave Plate

TL;DR

This paper compares two data demodulation techniques for CMB polarisation experiments using a rotating half-wave plate, finding that lock-in performs better overall but both methods are comparable at large scales, with implications for systematic effect mitigation.

Contribution

The study introduces a comparison between lock-in and map-making demodulation methods for CRHWP-based CMB experiments, highlighting their relative performance and potential for systematic effect analysis.

Findings

Lock-in technique outperforms map-making across most multipoles.

At large angular scales, both methods recover the B-mode signal equally well.

Detector differencing has minimal impact on signal recovery in most scenarios.

Abstract

Several forthcoming Cosmic Microwave Background polarisation experiments will employ a Continuously Rotating Half-Wave Plate (CRHWP), the primary purpose of which is to mitigate instrumental systematic effects. The use of a CRHWP necessitates demodulating the time-ordered data during the early stages of data processing. The standard approach is to ``lock in'' on the polarisation signal using the known polarisation modulation frequency and use Fourier techniques to filter out the remaining unwanted components. However, an alternative, less well-studied option is to incorporate the demodulation directly into the map-making step. Using simulations, we compare the performance of these two approaches to determine which is most effective for $B$-mode signal recovery. Testing the two techniques in multiple experimental scenarios, we find that the lock-in technique performs best over the full multipole range explored. However, for the recovery of the largest angular scales (multipoles, $\ell < 100$) we find essentially no difference in the recovery of the signal between the lock-in and map-making approaches, suggesting that a parallel analysis based on the latter approach could represent a powerful consistency check for primordial $B$-mode experiments employing a CRHWP. We also investigate the impact of a detector-differencing step, implemented prior to demodulation, finding that, in most scenarios, it makes no difference whether differencing is used or not. However, analysing detectors individually allows the point at which information from multiple detectors is combined to be moved to later stages in the analysis pipeline. This presents alternative options for dealing with additional instrumental systematic effects that are not mitigated by the CRHWP.