Multiplexed SQUID-Based Readout
- Multiplexed SQUID-based readout is a method for simultaneously reading out high-density cryogenic detector arrays using superconducting quantum interference devices with minimal noise.
- It employs frequency-domain (fMUX) and microwave (μMUX) architectures along with flux-ramp modulation to efficiently encode and demultiplex detector signals.
- This scalable technology supports thousands of detectors with low crosstalk and high sensitivity, crucial for applications in cosmic microwave background research and x-ray astronomy.
Multiplexed SQUID-based readout refers to a suite of mature and evolving technologies for the simultaneous low-noise, high-density readout of cryogenic detector arrays—such as transition-edge sensors (TES), metallic magnetic calorimeters (MMC), and magnetic microbolometers—using superconducting quantum interference devices (SQUIDs) as the front-end amplifiers and frequency or code-division multiplexing to efficiently utilize available bandwidth and wiring resources. The multiplexing schemes have enabled compact, scalable readout architectures supporting O(10²–10⁴) detectors per pair of cryogenic lines, and now underpin the focal-plane instrumentation of leading experiments in cosmic microwave background (CMB) research and high-rate x-ray and sub-mm astronomy.
1. Frequency-Multiplexed SQUID Readout Architectures
In classic low-frequency frequency-domain multiplexing (fMUX) (Dobbs et al., 2011, Haan et al., 2019, Lowitz et al., 2018), TES or MMC detectors are grouped into modules, each module using a unique set of series LC resonators and a common AC bias. Each detector in the module is assigned a distinct MHz-band carrier frequency, with amplitude modulation of the carrier encoding the sky or signal power. The bias comb is summed and applied on a single pair of wires, with the detector signals separated in frequency space by the LC resonators. The combined current is routed to a SQUID amplifier (commonly, a series array operated at 4 K or 100 mK), after which room-temperature demodulation and digitization recover the per-detector signals. State-of-the-art sub-Kelvin fMUX implementations co-locate the SQUIDs and LC networks adjacent to the detectors, minimizing parasitic inductance and series resistance, and eliminating the need for cold electronics at intermediate stages (Haan et al., 2019, Lowitz et al., 2018).
In GHz-band multiplexed readout (microwave SQUID multiplexing, or μMUX), a transition is made to readout at microwave (4–8 GHz) frequencies, exploiting lithographically defined high-Q coplanar waveguide or lumped-element resonators. Each detector (via its SQUID) terminates a unique resonator with resonance frequency fₙ, permitting N∼1000 resonances per feedline (Redondo et al., 28 Sep 2025, Liu et al., 2024, Henderson et al., 2018). The demodulation proceeds digitally by tracking the amplitude and phase of each microwave probe tone.
2. SQUID Amplifier Topologies and Modulation Techniques
The SQUID functions as a low-noise, low-impedance amplifier, with topology determined by the multiplexing scheme:
- Series Array SQUID: Used in sub-MHz fMUX; input coil inductance and feedback topologies (shunt, flux-locked loops) are optimized for low input impedance and high dynamic range (Dobbs et al., 2011). Closed-loop (FLL) operation extends dynamic range, e.g., from I_max=I_Φ₀/2 (open-loop) to ≈7–23 times greater, depending on loop gain.
- rf-SQUID/Resonator Combination: In μMUX, the rf-SQUID's Josephson inductance is flux-modulated by the detector signal and loads a microwave resonator, shifting its frequency. Non-hysteretic operation is ensured by constraining the SQUID screening parameter β_L<1 (Neidig et al., 10 Dec 2025, Redondo et al., 28 Sep 2025).
Flux-ramp modulation, a periodic sawtooth current applied to all SQUIDs, linearizes the otherwise 2π-periodic SQUID transfer function, enabling simultaneous readout of large arrays by turning the detector signal into a phase shift in the periodic response (Neidig et al., 9 Sep 2025, Malnou et al., 2023). MHz-band approaches with flux-ramp modulation enable per-channel MHz bandwidth in hybrid or MHz FDM multiplexers (Richter et al., 2021).
3. Multiplexing Factor, Bandwidth, and Channelization
The practical multiplexing limit N is set by the bandwidth of the SQUID (for fMUX), the loaded Q and separation of microwave resonators (for μMUX), and system constraints on noise and cross-talk:
- In MHz fMUX, N = BW_s / Δf, with Δf > resonator bandwidth; N≈7 displayed per SQUID at 1 MHz usable bandwidth in (Dobbs et al., 2011), up to N>100 in optimized sub-K fMUX (Haan et al., 2019, Lowitz et al., 2018).
- In μMUX, Δf_chan ≈ resonance bandwidth = f₀ / Q_L, so N ≈ (BW_total) / (f₀ / Q_L) (Liu et al., 2024, Neidig et al., 10 Dec 2025). Demonstrations report N=528 (Henderson et al., 2018), N=16–18 (Müller et al., 27 Sep 2025, Neidig et al., 9 Sep 2025), with targets and designs up to N∼4000 per line.
- Hybrid multiplexing (multiplexed SQUID + code division or per-resonator multiplexing) further increases N by a factor L, the code length or number of SQUIDs per resonator (Schuster et al., 2022, Yu et al., 2020).
Channel spacing is always chosen to keep cross-talk well below the noise floor, with measured crosstalk typically <0.4% (fMUX: <0.4% (Dobbs et al., 2011); μMUX: <0.3% (Dober et al., 2017)) for optical or electrical signals.
4. Signal Processing and System Implementation
Modern multiplexed SQUID readout is tightly integrated with software-defined radio (SDR) or FPGA-based digital signal processing:
- Carrier and Flux Ramp Generation: High-precision DDS, NCO, or lookup tables generate the frequency comb for carrier injection and nullers (N-channel operation), and the flux ramp for μMUX or hybrid schemes (Dobbs et al., 2011, Redondo et al., 28 Sep 2025, Liu et al., 2024).
- Demodulation Path: Synchronous detection is performed via digital mixers, low-pass filtering, and anti-aliasing (fMUX: square-wave mixer, Bessel filter; μMUX: per-channel tracking engines, polyphase filterbanks, closed-loop digital PLLs) (Liu et al., 2024, Redondo et al., 28 Sep 2025).
- Flux-ramp demodulation: The phase of the modulated tone is extracted by unwrapping over each ramp period, outputting the recovered detector signal at the ramp rate (Neidig et al., 9 Sep 2025, Richter et al., 2021).
- Scaling and Integration: High channel counts are sustained with sample rates up to 1 kHz (fMUX), 100 kHz – MHz (μMUX), with event packaging and digitized data output handled in firmware (Muscheid et al., 5 Dec 2025).
Room-temperature readout SDR systems are built on RFSoC or FPGA platforms, allowing direct RF sampling (removing the need for analog IQ-mixers and associated calibration) (Redondo et al., 28 Sep 2025, Liu et al., 2024).
5. Noise, Crosstalk, and System Performance
The noise budget for multiplexed SQUID-based readout includes photon and phonon noise from the detector, Johnson (thermal) noise, readout chain noise (SQUID, amplifiers, electronics), and cross-talk between channels:
| Noise Source | Typical μMUX Value | fMUX Value |
|---|---|---|
| SQUID current noise | 3.5 pA/√Hz | 20 pA/√Hz (sub-K) |
| Readout (FRM-demod) flux | 1.4 μΦ₀/√Hz (Neidig et al., 9 Sep 2025) | 98 pA/√Hz (Dober et al., 2017) |
| On-sky NET (CMB) | 400 μK √s (Dobbs et al., 2011) | - |
| Crosstalk | <0.3% (Dober et al., 2017) | <0.4% (Dobbs et al., 2011) |
SQUID and electronic readout noise are maintained well below the detector or photon noise, ensuring that multiplexing does not degrade overall system sensitivity. Integration of quantum-limited amplification stages, such as kinetic-inductance traveling-wave parametric amplifiers, reduces the system noise floor below 1 μΦ₀/√Hz (Malnou et al., 2023).
6. Advanced Multiplexing: Hybrid and Code-Division Schemes
Recent developments include hybrid microwave SQUID multiplexers (Schuster et al., 2022) and impedance-modulated code-division microwave SQUID multiplexers (Z-CDM) (Yu et al., 2020). The hybrid schemes relax the requirement for one resonator per detector by multiplexing multiple SQUIDs on a single resonator, assigning different flux ramp frequencies and thus encoding detector signals as sidebands. Z-CDM uses digital Walsh code modulation, multiplying the channel density by the code length and substantially relaxing resonator-frequency requirements for dense arrays (up to 10⁴–10⁵ detectors per readout line). Trade-offs include modest penalties in sampling density and amplifier noise due to shared resonators, but practical demonstrations confirm multiplexing factors >10,000 are feasible for narrow-bandwidth detectors.
7. Outlook: Scalability, Limitations, and Future Directions
Multiplexed SQUID readout is now field-proven to O(10³)–O(10⁴) detectors per system, with the Simons Observatory deploying μMUX at ∼1000x multiplexing (Vavagiakis et al., 2020). System limits are currently set by available RF bandwidth (commonly 4–8 GHz), resonator fabrication tolerances, amplifier power handling and linearity, and digital signal processing resources. Fundamental scaling is further constrained by resonator Q, crosstalk, and flux ramp demodulation bandwidth.
Progress areas include:
- Direct RF Sampling and Digital Up/Downconversion: RFSoC-based implementations eliminate analog IQ chains, lower power, and offer >1,000 channel readout per module (Liu et al., 2024, Redondo et al., 28 Sep 2025).
- High-Density Microwave Multiplexing: Integration of improved parametric amplifiers, dynamic closed-loop tone-tracking (SMuRF (Henderson et al., 2018)), and advanced multiplexing protocols (CDM, hybrid schemes) provide pathways toward 10⁴–10⁵ detector arrays.
- Sub-Kelvin Integration: Moving the SQUIDs to the sub-K stage (with the detectors) minimizes wiring parasitics, improves noise performance, reduces crosstalk, and enables compact focal-plane integration at negligible additional heat load (Haan et al., 2019, Lowitz et al., 2018).
- Comprehensive Modelling: Advanced numerical models incorporating readout power effects, Josephson junction inhomogeneities, and noise analysis guide next-generation μMUX optimization and fabrication (Neidig et al., 10 Dec 2025, Schuster et al., 2022).
In summary, multiplexed SQUID-based readout—spanning both low-frequency (fMUX) and microwave (μMUX) domains, with continual innovation in digital signal processing, hardware integration, and system design—remains a central enabling technology for large-format cryogenic sensor arrays in astrophysics, quantum sensing, and related fields (Dobbs et al., 2011, Dober et al., 2017, Redondo et al., 28 Sep 2025, Liu et al., 2024, Neidig et al., 10 Dec 2025).