Superconducting Sensor Control
Designed for multiplexed microwave control of Kinetic Inductance Detector arrays in deployment environments.
Designed to operate large focal planes of Kinetic Inductance Detectors.
The rfmux firmware includes support for active feedback with configurable modes optimized for mapping/imaging, particle-physics, or single-photon detection.
White paper now available
The CRS: a scalable full-stack control system for Microwave Kinetic Inductance Detectors
Multiple Bandwidth Configurations
Each CRS supports a total of 4,096 control and readout channels. These can be allocated across different I/O modules and instantaneous bandwidth. The standard configuration allocates 625 MHz of instantaneous bandwidth and 1,024 channels for each of 4 I/O modules.
A wide bandwidth configuration is also available, which consolidates the full 2.5 GHz of instantaneous bandwidth and 4,096 channels into a single I/O module.
Standard Bandwidth Mode
4 active inputs (selectable)
4 active outputs (selectable)
1,024 channels per I/O
625 MHz of selectable instantaneous bandwidth per I/O
Wide Bandwidth Mode
1 active input (selectable)
1 active output (selectable)
4,096 channels per I/O
2.5 GHz of selectable instantaneous bandwidth per I/O
Multiple Data Streaming Options
For instruments that perform mapping-style imaging, where the target data is fully contained within narrow side-bands of the control tones, the full (I,Q) data for 4,096 channels can be continuously streamed with 2.4 KHz of complex bandwidth via standard 1G Ethernet.
Micro-calorimetry, particle-physics, or photon-number-resolving applications characterize discrete pulse shapes to measure energy deposition, and for these applications up to 512 channels can be continuously streamed with 2.44 MHz of complex bandwidth using the 100G QSFP28 interface.
It is also possible to stream the full 2.5 GHz instantaneous bandwidth via the QSFP28 interface.
Standard 1G Ethernet interface that streams (I,Q) data from all 4,096 channels with 2.4 KHz of complex bandwidth.
100G QSFP28 interface continuously streams (I,Q) data from 512 channels with 2.44 MHz of complex bandwidth, or the full 2.5 GHz of instantaneous I/O bandwidth.
Feedback Enabled
Microwave Control Firmware
The rfmux signal path is based on a coupled polyphase filter bank (PFB) synthesizer and demodulator. This allows high channel density and real-time feedback.
Feedback dynamically adjusts carrier tone amplitude, phase, and frequency as a function of the real-time (I,Q) data. It can be configured for a diverse set of digital control feedback techniques.
t0.technology collaborates with academic researchers to identify ways the CRS can support novel control and readout techniques, and to make the outcomes available to researchers.
One example of this is Active Resonator Feedback (ARC) for Kinetic Inductance Detectors. ARC is being pioneered by McGill researchers, and is described in (Rouble. et al 2024) as a technique that may allow operation beyond traditional bifurcation powers; extend detector dynamic range; and linearize detector response without using tone tracking.
t0.technology is working with researchers to support real-time firmware implementations of ARC on CRS boards with the rfmux firmware, and to characterize the behavior of the detectors under that feedback.
Multi-Probe View:
Research Applications
Superconducting resonators are often characterized with Vector Network Analyzer (VNA) systems, which use a single measurement tone swept up or down in frequency. However, the dynamics are better understood with a coincident view across the full resonance.
The rfmux firmware can allocate more than a thousand synchronously sampled probes, which use small amplitude perturbations to provide a complete view of the superconducting system as it evolves in time.
Shown here is a multi-probe measurement of the impedance of a Kinetic Inductance Detector while a large amplitude tone is swept across it. The multi-probe view shows behavior that is not apparent with a VNA.
The multi-probe measurement technique for characterizing the dynamics of KIDs was first described by a McGill research group (Rouble et al, 2024) while investigating detectors developed for South Pole Telescope.
These measurements are now natively supported on CRS hardware as part of the baseline readout for the proposed South Pole Telescope 3G+ instrument. This figure was made using a CRS board and the open-source rfmux API, and executing from hidfmux, a general KIDs control software developed by academic researchers and used on the South Pole Telescope (Rouble et al 2023).
t0.technology provides tools for scientists to understand complex systems, and pairs them with the open interface software. This combination allows researchers to creatively develop new operational algorithms or tailor measurements for their unique experimental parameters.
t0.technology provides the tools for researchers to understand the complex systems they work with.
Shown here is a multi-probe view of Kinetic Inductance Detector bifurcation, generated using the the CRS platform and third-party hidfmux control software developed by academic researchers.
Control Software
The t0.technology provides an open-source Python API to rfmux with an asynchronous framework for efficiently operating large numbers of CRS boards. It also includes a data acquisition system for continuously streamed data products. The rfmux API is designed to be forked and extended to generate customized algorithms your project uses for measurement and data visualization.
Each CRS board also hosts a JupyterLab instance customized for rfmux, allowing simple single-board operation through a web-application.
The CRS hardware and rfmux API is also compatible with the general KIDs control software, hidfmux, developed by researchers and used for the South Pole Telescope. hidfmux includes sophisticated KID characterization and optimization routines that now seamlessly integrate the CRS API.
The detector digest routine is part of the hidfmux control software, and was written by researchers to characterize Kinetic Inductance Detectors. The figure above used CRS hardware and rfmux API, executed within hidfmux.