Readout noise of digital frequency multiplexed TES detectors for CUPID

Abstract

Superconducting transition-edge sensor (TES) detectors have been the standard in Cosmic Microwave Background experiments for almost two decades and are now being adapted for use in nuclear physics, such as neutrinoless double beta decay searches. In this paper we focus on a new high-bandwidth frequency multiplexed TES readout system developed for CUPID, a neutrinoless double beta decay experiment that will replace CUORE. In order to achieve the high energy resolution requirements for CUPID, the readout noise of the system must be kept to a minimum. Low TES operating resistance and long wiring between the readout SQUID and the warm electronics are needed for CUPID, prompting a careful consideration of the design parameters of this application of frequency multiplexing. In this work, we characterize the readout noise of the newly designed frequency multiplexed TES readout system for CUPID and construct a noise model to understand it. We find that current sharing between the SQUID coil impedance and other branches of the circuit, as well as the long output wiring, worsen the readout noise of the system. To meet noise requirements, a SQUID with a low input inductance, high transimpedance and/or low dynamic impedance is needed, and the wiring capacitance should be kept as small as possible. Alternatively, the option of adding a cryogenic low-noise amplifier at the output of the SQUID should be explored.

Figures (8)

• Figure 1: Simplified schematic of the CUPID fMUX TES readout. The bi-layer IrPt TES detectors are on the mixing chamber plate (blue), the resonator bank and readout SQUID on the still plate (yellow), while the ICEboard, its accessories and the analog/digital converters are at room temperature outside the cryostat (red).
• Figure 2: Circuit schematic of the cold electronics board with the SQUID and LC resonators, showing current sharing paths for the nuller (blue arrows). Ideally, all the nuller current would flow through the SQUID input coil (purple).
• Figure 3: Modeled current sharing factor for our system parameters. Current sharing increases with frequency due to rising SQUID coil impedance and spikes on-resonance due to low TES operating resistance.
• Figure 4: Modeled SQUID output filter magnitude response for a dynamic impedance of 820 $\Omega$ and wire harness capacitance of 120 pF.
• Figure 5: On-resonance readout noise model prediction for our system parameters, showing contributions from different noise sources and the total noise in black, including the TES Johnson noise.