Systematic errors in spectral measurements with the Tenerife Microwave Spectrometer
We present an analytical instrument model of the TMS radiometer, a pseudo-correlation system designed for absolute sky-temperature measurements through a continuous comparison between sky and reference-load signals. The goal of this work is to quantify and understand the impact of instrumental non-idealities that are intrinsic to absolute radiometric measurements. We used a combination of the Jones-matrix formalism, to describe non-ideal signal mixing effects in the different instrument components, and the Friis formalism, to account for noise-temperature contributions which introduce specific offset terms. The model includes all components from the cryostat entrance window onwards and allows us to propagate realistic losses, return losses, and noise figures through the system. We find that non-ideal mixing effects, such as intensity-to-polarization and sky-to-load leakage, are expected to appear at the percent level, but they could be in principle calibrated out. The total TMS intensity response exhibits a frequency-dependent offset with a band-averaged level of 6.9 K, dominated by the cryostat window, and with smaller contributions from the infrared filter and orthomode transducers. Under the required TMS thermal stability of 1 mK over one hour, the resulting variation in the output signal can reach up to 91.3 mK, which sets the fundamental limit on the absolute sky-temperature accuracy. In contrast, relative spectral measurements across the 10-20 GHz band are stable at the few-microkelvin level, consistent with the instrument target sensitivity of ~ 10 Jy/sr. This work provides a detailed and quantitative assessment of the systematic effects affecting absolute radiometric measurements and establishes a robust framework for the calibration and performance optimization of the TMS instrument.