CCAT: Readout of over 10,000 280 GHz KIDs in Mod-Cam using RFSoC Electronics

TL;DR

The paper addresses the challenge of scaling KID readout for CCAT’s Prime-Cam by implementing a five-RFSoC warm readout system capable of simultaneously interrogating >10,000 KIDs across a 280 GHz module. It introduces a complete hardware/software pipeline (RFSoC boards, primecam_readout, rfsoc-streamer, ccatkidlib) that performs VNA-style tone sweeps, targeted tone placement, and time-ordered data streaming, validated with VNA traces and FTS-based spectral measurements. The results demonstrate end-to-end readout of thousands of detectors in a testbed, establishing a practical path toward Prime-Cam’s eventual ~100,000 KID deployment and informing detector tuning and noise performance at scale. This work has significant implications for high-density, millimeter-wave detector readouts in large-aperture submillimeter surveys and cosmology experiments.

Abstract

Over the past decade, kinetic inductance detectors (KIDs) have emerged as a viable superconducting technology for astrophysics at millimeter and submillimeter wavelengths. KIDs spanning 210 - 850 GHz across seven instrument modules will be deployed in the Prime-Cam instrument of CCAT Observatory's Fred Young Submillimeter Telescope at an elevation of 5600 m on Cerro Chajnantor in Chile's Atacama Desert. The natural frequency-division multiplexed readout of KIDs allows hundreds of detectors to be coupled to a single radio frequency (RF) transmission line, but requires sophisticated warm readout electronics. The FPGA-based Xilinx ZCU111 radio frequency system on chip (RFSoC) offers a promising and flexible solution to the challenge of warm readout. CCAT uses custom packaged RFSoCs to read out KIDs in the Prime-Cam instrument. Each RFSoC can simultaneously read out four RF channels with up to 1,000 detectors spanning a 512 MHz bandwidth per channel using the current firmware. We use five RFSoCs to read out the >10,000 KIDs in the broadband 280 GHz instrument module inside a testbed receiver. Here, we describe and demonstrate the readout software and pipeline for the RFSoC system. We also present the preliminary averaged spectral responses of the 280 GHz instrument module using KIDs from the TiN array and the first Al array as a demonstration of the end-to-end performance of the readout and optical systems. These measurements demonstrate the foundation that will enable us to simultaneously read out over 10,000 KIDs with the RFSoC and represent a critical step toward reading out the ~100,000 KIDs in Prime-Cam in its future full capacity configuration.

Figures (7)

• Figure 1: The full 280 GHz focal plane with one TiN KID array with Al feedhorns (top right) and two Al KID arrays with silicon feedhorns. Each array contains $\sim$3400 polarization sensitive KIDs and is divided into six networks with $\sim$570 KIDs each.
• Figure 2: A radio frequency system on a chip (RFSoC) board (A) inside a custom 1U rack-mountable enclosure. The RFSoC FPGA fabric is cooled with a copper heat sink (B). The RFSoC analog-to-digital and digital-to-analog converters are broken out using a custom board and connected to digitally controlled variable attenuators (C) with up to 31.75 dB of attenuation. An Opsero Ethernet FPGA Mezzanine card (D) is attached to the board to stream time-ordered data. The remainder of the enclosure (E) contains the power supply, an unmanaged switch for breaking out Ethernet connections, and additional accessories for debugging.
• Figure 3: A high-level schematic diagram of the software pipeline used to take data with the RFSoCs from a centralized control computer. Only a single RFSoC is included in the diagram for simplicity but multiple can be connected at the network switch interface. Software packages are denoted by the orange circles with their hierarchy indicated by the solid arrows. Networking connections between the control computer and the RFSoCs are denoted in purple with the dashed arrows indicating communication via TCP (for the Redis server) or UDP (for time-ordered data streaming). Data products are denoted by the red hexagons with the dotted lines indicating which software packages create them.
• Figure 4: Top Left: A Fourier Transform Spectrometer (FTS) with a $\sim700$ K thermal blackbody source positioned in front of the Mod-Cam receiver while optically open to the room to measure the spectral response of the 280 GHz instrument module. An aperture stop is placed in front of the thermal source to limit the amount of light entering the FTS, and activated charcoal cloth is used to reduce stray reflections. Bottom Left: An unobstructed view of the FTS showing the input and output mirrors, polarized wire grids, and motorized central mirror. Right: Five rack-mounted radio frequency systems on a chip (RFSoCs) connected to the Mod-Cam RF readout feedthrough with 36 SMA connections. Also pictured is a custom biasing system from ASU used to bias the 18 low-noise amplifiers.
• Figure 5: Schematic diagram of the Mod-Cam and Prime-Cam cold readout chains adapted from DuellThesis. Attenuator values and distribution vary between individual networks and are still being finalized.