Resonator-Multiplexed Readout
Assigning each qubit's readout resonator a distinct frequency lets one feedline and one amplifier chain read many qubits at once.
One Line, Many Qubits
Reading each qubit through its own wire and its own amplifier chain does not scale: the wiring and the cryogenic amplifier count would explode. Frequency-division multiplexing solves this. Each qubit is given a readout resonator at a distinct frequency, and all the resonators couple to a single shared transmission line, the feedline. A single broadband amplifier chain then serves the whole group.
How It Works
To read the group, the control electronics generate a comb of probe tones, one at each resonator frequency, sum them into a single waveform, and send it down the feedline. The returned signal contains all the responses superimposed. Digital demodulation at each resonator frequency separates the individual qubit signals. Because a single arbitrary-waveform generator and digitizer can synthesize and analyze many tones, one instrument reads many qubits.
- Resonators are spaced in frequency to avoid crosstalk between readout channels.
- A broadband amplifier, ideally a traveling-wave parametric amplifier, covers the whole comb.
- Digital signal processing demodulates each tone independently.
Constraints
The number of qubits per line is bounded by the available bandwidth divided by the frequency spacing needed to keep channels from interfering, and by the total power the amplifier can handle without saturating, since power is shared among tones. Resonators must be fabricated on target frequency to fit their allotted slots, which places demands on fabrication precision. Simultaneous readout can also introduce weak measurement crosstalk that must be characterized and corrected.
Role in Scaling
Multiplexed readout is one of the pillars of scalable superconducting and spin-qubit systems, directly addressing the wiring bottleneck discussed elsewhere in this library. Combined with wideband parametric amplifiers it lets tens of qubits share a single output line while preserving single-shot fidelity, dramatically reducing the input and output count that a large processor demands.
The technique is a clear example of using classical signal-processing ideas to relieve a quantum-hardware scaling constraint.